为什么 AI evals 是产品构建者最热门的新技能 | Hamel Husain & Shreya Shankar
Why AI evals are the hottest new skill for product builders | Hamel Husain & Shreya Shankar
Guest Intro & the Rise of Evals
Lenny Rachitsky: To build great AI products, you need to be really good at building evals. It’s the highest ROI activity you can engage in.
Hamel Husain: This process is a lot of fun. Everyone that does this immediately gets addicted to it. When you’re building an AI application, you just learn a lot.
What Exactly Are Evals?
Lenny Rachitsky: What’s cool about this is you don’t need to do this many, many times. For most products, you do this process once and then you build on it.
Shreya Shankar: The goal is not to do evals perfectly, it’s to actionably improve your product.
Lenny Rachitsky: I did not realize how much controversy and drama there is around evals. There’s a lot of people with very strong opinions.
Shreya Shankar: People have been burned by evals in the past. People have done evals badly, so then they didn’t trust it anymore, and then they’re like, “Oh, I’m anti evals.”
Lenny Rachitsky: What are a couple of the most common misconceptions people have with evals?
Evals Are More Than Tests
Hamel Husain: The top one is, “We live in the age of AI. Can’t the AI just eval it?” But it doesn’t work.
Lenny Rachitsky: A term that you used in your posts that I love is this idea of a benevolent dictator.
Real-World Case: Nurture Boss
Hamel Husain: When you’re doing this open coding, a lot of teams get bogged down in having a committee do this. For a lot of situations, that’s wholly unnecessary. You don’t want to make this process so expensive that you can’t do it. You can appoint one person whose taste that you trust. It should be the person with domain expertise. Oftentimes, it is the product manager.
Lenny Rachitsky: Today, my guests are Hamel Husain and Shreya Shankar. One of the most trending topics on this podcast over the past year has been the rise of evals. Both the chief product officers of Anthropic and OpenAI shared that evals are becoming the most important new skill for product builders. And since then, this has been a recurring theme across many of the top AI builders I’ve had on. Two years ago, I had never heard the term evals. Now it’s coming up constantly. When was the last time that a new skill emerged that product builders had to get good at to be successful?
Hamel and Shreya have played a major role in shifting evals from being an obscure, mysterious subject to one of the most necessary skills for AI product builders. They teach the definitive online course on evals, which happens to be the number one course on Maven. They’ve now taught over 2,000 PMs and engineers across 500 companies, including large swaths of the OpenAI and Anthropic teams along with every other major AI lab.
In this conversation, we do a lot of show versus tell. We walk through the process of developing an effective eval, explain what the heck evals are and what they look like, address many of the major misconceptions with evals, give you the first few steps you can take to start building evals for your product, and also share just a ton of best practices that Hamel and Shreya have developed over the past few years. This episode is the deepest yet most understandable primer you’ll find on the world of evals. And honestly, it got me excited to write evals, even though I have nothing to write evals for. I think you’ll feel the same way as you watch this.
If this conversation gets you excited, definitely check out Hamel and Shreya’s course on Maven. We’ll link to it in the show notes. If you use the code LENNYSLIST when you purchase the course, you’ll get 35% off the price of the course. With that, I bring you Hamel Husain and Shreya Shankar.
And Fin is trusted by over 5,000 customer service leaders and top AI companies like Anthropic and Synthesia. And because Fin is powered by the Fin AI engine, which is a continuously improving system that allows you to analyze, train, test, and deploy with ease, Fin can continuously improve your results too. So if you’re ready to transform your customer service and scale your support, give Fin a try for only 99 cents per resolution. Plus, Fin comes with a 90-day money-back guarantee. Find out how Fin can work for your team at fin.ai/lenny. That’s fin.ai/lenny.
Inspecting Real Conversation Traces
Hamel Husain: Thank you for having us.
Shreya Shankar: Yeah, super excited.
Methods for Error Analysis
Lenny Rachitsky: I’m even more excited. Okay, so a couple years ago, I had never heard the term evals. Now it’s one of the most trending topics on my podcast, essentially, that to build great AI products, you need to be really good at building evals. Also, it turns out some of the fastest-growing companies in the world are basically building and selling and creating evals for AI labs. I just had the CEO of Mercor on the podcast. So there’s something really big happening here. I want to use this conversation to basically help people understand this space deeply, but let’s start with the basics. Just what the heck are evals? For folks that have no idea what we’re talking about, give us just a quick understanding of what an eval is, and let’s start with Hamel.
Note-Taking Rule: Log First Error Only
Hamel Husain: Sure. Evals is a way to systematically measure and improve an AI application, and it really doesn’t have to be scary or unapproachable at all. It really is, at its core, data analytics on your LLM application and a systematic way of looking at that data, and where necessary, creating metrics around things so you can measure what’s happening, and then so you can iterate and do experiments and improve.
Examples of LLM Hallucinations
Lenny Rachitsky: So that’s a really good broad way of thinking about it. If you go one level deeper just to give people a very, even more concrete way of imagining and visualizing what we’re talking about, even if you have a example to show would be even better, what’s an even deeper way of understanding what an eval is?
Can LLMs Automate Error Analysis?
Hamel Husain: Let’s say you have a real estate assistant application and it’s not working the way you want. It’s not writing emails to customers the way you want, or it’s not calling the right tools, or any number of errors. And before evals, you would be left with guessing. You would maybe fix a prompt and hope that you’re not breaking anything else with that prompt, and you might rely on vibe checks, which is totally fine.
And vibe checks are good and you should do vibe checks initially, but it can become very unmanageable very fast because as your application grows, it’s really hard to rely on vibe checks. You just feel lost. And so evals help you create metrics that you can use to measure how your application is doing and kind of give you a way to improve your application with confidence. That you have a feedback signal in which to iterate against.
The Benevolent Dictator Approach
Lenny Rachitsky: So just to make very real, so imagining this real estate agent, maybe they’re helping you book a listing or go see an open house. The idea here is you have this agent talking to people, it’s answering questions, pointing them to things. As a builder of that agent, how do you know if it’s giving them good advice, good answers? Is it telling them things that are completely wrong?
So the idea of evals, essentially, is to build a set of tests that tell you, how often is this agent doing something wrong that you don’t want it to do? And there’s a bunch of ways you could define wrong. It could be just making up stuff. It could be just answering in a really strange way. The way I think about evals, and tell me if this is wrong, just simply is like unit tests for code. You’re smiling. You’re like, “No, you idiot.”
Shreya Shankar: No, that’s not what I was thinking.
How Many Traces to Sample?
Lenny Rachitsky: Okay. Okay, okay, tell me. Tell me, how does that feel as a metaphor?
Shreya Shankar: Okay. I like what you said first, which is we had a very broad definition. Evals is a big spectrum of ways to measure application quality. Now, unit tests are one way of doing this. Maybe there are some non-negotiable functionalities that you want your AI assistant to have, and unit tests are going to be able to check that. Now, maybe you also, because these AI assistants are doing such open-ended tasks, you kind of also want to measure how good are they at very vague or ambiguous things like responding to new types of user requests or figuring out if there’s new distributions of data like new users are coming and using your real estate agent that you didn’t even know would use your product. And then all of a sudden, you think, “Oh, there’s a different way you want to kind of accommodate this new group of people.”
So evals could also be a way of looking at your data regularly to find these new cohorts of people. Evals could also be like metrics that you just want to track over time, like you want to track people saying, “Yes. Thumbs up. I liked your message.” You want very, very basic things that are not necessarily AI-related but can go back into this flywheel of improving your product. So I would say, overall, unit tests are a very small part of that very big puzzle.
Lenny Rachitsky: Awesome. You guys actually brought an example of an eval just to show us exactly what the hell we’re talking about. We’re talking in these big ideas. So how about let’s pull one up and show people, “Here’s what an eval is.”
From Open to Axial Coding
Hamel Husain: Yeah, let me just set the stage for it a little bit. So to echo what Shreya said, it’s really important that we don’t think of evals as just tests. There’s a common trap that a lot of people fall into because they jump straight to the test like, “Let me write some tests,” and usually that’s not what you want to do. You should start with some kind of data analysis to ground what you should even test, and that’s a little bit different than software engineering where you have a lot more expectations of how the system is going to work. With LLMs, it’s a lot more surface area. It’s very stochastic, so you kind of have a different flavor here.
And so the example I’m going to show you today, it’s actually a real estate example. It’s a different kind of real estate example. It’s from a company called Nurture Boss. I can share my screen to show you their website just to help you understand this use case a little bit, so let me share my screen. So this is a company that I worked with. It’s called Nurture Boss, and it is a AI assistant for property managers who are managing apartments, and it helps with various tasks such as inbound leads, customer service, booking appointments, so on and so forth. It’s like all the different sort of operations you might be doing as a property manager, it helps you with that. And so you can see kind of what they do. It’s a very good example because it has a lot of the complexities of a modern AI application.
So there’s lots of different channels that you can interact through the AI with like chat, text, voice, but also, there’s tool calls, lots of tool calls for booking appointments, getting information about availability, so on and so forth. There’s also RAG retrieval, getting information about customers and properties and things like that. So it’s pretty fully fleshed in terms of an AI application. And so they have been really generous with me in allowing me to use their data as a teaching example. And so we have anonymized it, but what I’m going to walk through today is, okay, let’s do the first part of how we would start to build evals for Nurture Boss. Why would we even want to do that?
So let’s go through the very beginning stage, what we call error analysis, which is, let’s look at the data of their application and first start with what’s going wrong. So I’m going to jump to that next, and I’m going to open an observability tool. And you can use whatever you want here. I just happen to have this data loaded in a tool called Braintrust, but you can load it in anything. We don’t have a favorite tool or anything in the blog post that we wrote with you. We had the same example but in Phoenix Arize, and I think Aman, on your blog post, used Phoenix Arize as well. And there’s also LangSmith. So these are kind of like different tools that you can use.
So what you see here on the screen, this is logs from the application, and let me just show you how it looks. So what you see here is, and let me make it full screen, this is one particular interaction that a customer had with the Nurture Boss application, and what it is is a detailed log of everything that happened. So it’s called a trace, and it’s just the engineering term for logs of a sequence of events. The concept of a trace has been around for a really long time, but it’s especially really important when it comes to AI applications.
And so we have all the different components and pieces and information that the AI needs to do its job, and we are logged all of it and we’re looking at a view of that. And so you see here a system prompt. The system prompt says, “You are an AI assistant working as a leasing team member at Retreat at Acme Apartments.” Remember, I said this is anonymized, so that’s why the name is Acme Apartments. “Your primary role is to respond to text messages from both current residents and prospective residents. Your goal is to provide accurate, helpful information,” yada, yada, yada. And then there’s a lot of detail around guidelines of how we want this thing to behave.
Lenny Rachitsky: Is this their actual system prompt, by the way, for this company?
Basic Counting: The Most Powerful Technique
Hamel Husain: It is. Yes, it is.
Customizing Axial Coding Prompts Flexibly
Lenny Rachitsky: Amazing. That’s so cool.
Hamel Husain: It’s a real system prompt.
Lenny Rachitsky: That’s amazing because it’s rare you see a actual company product’s system prompt. That’s like their crown jewels a lot of times, so this is actually very cool on its own.
Hamel Husain: Yeah. Yeah, it’s really cool. And you see all of these different sort of features that are different use cases, so things about tour scheduling, handling applications, guidance on how to talk to different personas, so on and so forth. And you can see the user just kind of jumps in here and asks, “Okay, do you have a one-bedroom with study available? I saw it on virtual tours.” And then you can see that the LLM calls some tools. It calls this get individual’s information tool, and it pulls back that person’s information. And then it gets the community’s availability. So it’s querying a database with the availability for that apartment complex.
And then finally, the AI responds, “Hey, we have several one-bedroom apartments available, but none specifically listed with a study. Here are a few options.”
And then it says, “Can you let me know when one with a study is available?”
And then it says, “I currently don’t have specific information on the availability of a one-bedroom apartment.”
User says, “Thank you.”
And the AI says, “You’re welcome. If you have any more questions, feel free to reach out.” Now, this is an example of a trace, and we’re looking at one specific data point. And so one thing that’s really important to do when you’re doing data analysis of your LLM application is to look at data. Now, you might wonder, “There’s a lot of these logs. It’s kind of messy. There’s a lot of things going on here. How in the hell are you supposed to look at this data? Do you want to just drown in this data? How do you even analyze this data?”
So it turns out there is a way to do it that is completely manageable, and it’s not something that we invented. It’s been around in machine learning and data science for a really long time, and it’s called error analysis. And what you do is, the first step in conquering data like this is just to write notes. Okay? So you got to put your product hat on, which is why we’re talking to you, because product people have to be in the room and they have to be involved in sort of doing this. Usually a developer is not suited to do this, especially if it’s not a coding application.
Lenny Rachitsky: And just to mirror back, why I think you’re saying that is because this is the user experience of your product. People talking to this agent is the entire product essentially, and so it makes sense for the product person to be super involved in this.
Hamel Husain: Yeah. So let’s reflect on this conversation. Okay, a user asked about availability. The AI said, “Oh, we don’t really have that. Have a nice day.” Now, for a product that is helping you with lead management, is that good? Do you feel like this is the way we want it to go?
Lenny Rachitsky: Not ideal.
Hamel Husain: Yes, not ideal, and I’m glad you said that. A lot of people would say, “Oh, it’s great. The AI did the right thing. It looked, it said, ‘We didn’t have available,’ and it’s not available.” But with your product hat on, you know that’s not correct. And so what you would do is you would just write a quick note here. You would say, “Okay.” You might pop in here, and you can write a note. So every observability application has ability to write notes, and you wouldn’t try to figure out if something is wrong. In this case, it’s kind of not doing the right thing, but you just write a quick note, “Should have handed off to a human.”
Lenny Rachitsky: And as we watch this happening, it’s like you mention this and you’ll explain more. You’re doing this, this feels very manual and unscalable, but as you said, this is just one step of the process and there’s a system to this. That was just the first one.
Hamel Husain: Yeah, and you don’t have to do it for all of your data. You sample your data and just take a look, and it’s surprising how much you learn when you do this. Everyone that does this immediately gets addicted to it and they say, “This is the greatest thing that you can do when you’re building an AI application.” You just learn a lot and you’re like, “Hmm, this is not how I want it to work. Okay.” And so that’s just an example.
So you write this note, and then we can go on to the next trace. So this is the next trace. I just pushed a hot key on my keyboard. Let me go back to looking at it.
Lenny Rachitsky: And these tools make it easy to go through a bunch and add these notes quickly.
Hamel Husain: Yes. And so this is another one. Similar system prompt. We don’t need to go through all of it again. We’ll just jump right into the user question. “Okay, I’ve been texting you all day.” Isn’t that funny? And the user says, “Please.” Okay, yeah, this one is just like an error in the application where this is a text message application, sorry, the channel through which the customer is communicating is through text message, and you’re just getting really garbled. And you can see here that it kind of doesn’t make sense. The words are being cut off like, “In the meantime,” and then the system doesn’t know how to respond, because you know how people text message, they write short phrases. They split their sentence across four or five different turns. So in this case-
Lenny Rachitsky: Yeah, so what do you do with something like that?
Hamel Husain: Yeah, so this is a different kind of error.
Lenny Rachitsky: Mm.
Hamel Husain: This is more of, “Hey, we’re not handling this interaction correctly. This is more of a technical problem,” rather than, “Hey, the AI is not doing exactly what we want.” So we would write that down too.
Lenny Rachitsky: Which is still really cool.
Hamel Husain: Yeah.
Lenny Rachitsky: It’s amazing you’re catching that, too, here. Otherwise, you’d have no idea this was happening.
Hamel Husain: Yeah, you might not know this is happening, right? And so you would just say, “Okay.” You would write a note like, “Oh, conversation flow is janky because of text message.”
Lenny Rachitsky: And I like that, I like that you’re using the word janky. It shows you just how informal this can be at this stage.
Hamel Husain: Yeah, it’s supposed to be chill. Just don’t overthink it. And there’s a way to do this. So the question always comes up, how do you do this? Do you try to find all the different problems in this trace? What do you write a note about? And the answer is, just write down the first thing that you see that’s wrong, the most upstream error. Don’t worry about all the errors, just capture the first thing that you see that’s wrong, and stop, and move on. And you can get really good at this. The first two or three can be very painful, but you can do a bunch of them really fast.
So here’s another one, and let’s skip the system prompt again. And the user asks, “Hey, I’m looking for a two- to three-bedroom with either one or two baths. Do you provide virtual tours?”
And a bunch of tools are called and it says, “Hi Sarah. Currently, we have three-bedroom, two-and-a-half-bathroom apartment available for $2,175. Unfortunately, we don’t have any two-bedroom options at the moment. We do offer virtual tours. You can schedule a tour,” blah, blah. It just so happens that there is no virtual tour, right?
Lenny Rachitsky: Mm-hmm. Nice.
Hamel Husain: So it is hallucinating something that doesn’t exist. Then you kind of have to bring your context as an engineer, or even product content, and say, “Hey, this is kind of weird. We shouldn’t be telling a person about virtual tour when it’s not offered.”
So you would say, “Okay, offered virtual tour,” and you just write the note. So you can see there’s a diversity of different kinds of errors that we’re seeing, and we’re actually learning a lot about your application in a very short amount of time.
Foundations of the Methodology
Shreya Shankar: One common question that we get from people at this stage is, “Okay, I understand what’s going on. Can I ask an LLM to do this process for me?”
Live Demo: Tools and Workflows
Lenny Rachitsky: Mm, great question.
Shreya Shankar: And I loved Hamel’s most recent example because what we usually find when we try to ask an LLM to do this error analysis is it just says the trace looks good because it doesn’t have the context needed to understand whether something might be bad product smell or not. For example, the hallucination about scheduling the tour, right? I can guarantee you, I would bet money on this, if I put that into chat GPT and asked, “Is there an error?” it would say, “No, did a great job.”
But Hamel had the context of knowing, “Oh, we don’t actually have this virtual tour functionality,” right? So I think, in these cases, it’s so important to make sure you are manually doing this yourself. And we can talk a little bit more about when to use LLMs in the process later, but number one pitfall right here is people are like, “Let me automate this with an LLM.”
Lenny Rachitsky: Do you think we’ll get to a place where an agent can do this, where it has that context?
Shreya Shankar: Oh, no. No, no, no. Sorry. There are parts of error analysis that an LLM is suited for, which we could talk about later in this podcast. But right now, in this stage of free form, note-taking is not the place for an LLM.
Lenny Rachitsky: Got it. And this is something you call open coding, this step?
Shreya Shankar: Yes, absolutely.
Lenny Rachitsky: Cool. Another term that you used in your posts that I love and that fits into this step is this idea of a benevolent dictator. Maybe just talk about what that is, and maybe, Shreya, cover that.
Shreya Shankar: Yeah, so Hamel actually came up with this term.
Lenny Rachitsky: Okay, maybe Hamel cover that, actually.
From Pivot Tables to AI Evals
Hamel Husain: No problem. And we’ll actually show the LLM automation in this example, because we’re going to take this example, we’re going to go all the way through.
Lenny Rachitsky: Amazing.
Selecting Questions and Building Evals
Hamel Husain: And so benevolent dictator is just a catchy term for the fact that when you’re doing this open coding, a lot of teams get bogged down in having a committee do this. And for a lot of situations, that’s wholly unnecessary. People get really uncomfortable with, “Okay, we want everybody on board. We want everybody involved,” so on and so forth. You need to cut through the noise. And a lot of organizations, if you look really deeply, especially small, medium-sized companies, you can appoint one person whose tastes that you trust. And you can do this with a small number of people and often one person, and it’s really important to make this tractable. You don’t want to make this process so expensive that you can’t do it. You’re going to lose out.
So that’s the idea behind benevolent dictator, is, “Hey, you need to simplify this across as many dimensions as you can.” Another thing that we’ll talk about later is when it goes to building an LLM as a judge, you need a binary score. You don’t want to think about, “Is this like a 1, 2, 3, 4, 5?” Like, assign a score to it. You can’t. That’s going to slow it down.
Lenny Rachitsky: Just to make sure this benevolent dictator point is really clear, basically, this is the person that-
Make sure this benevolent dictator point is really clear. Basically, this is the person that does this note-taking, and ideally they’re the expert on the stuff. So if it’s law stuff, maybe there’s a legal person that owns this, it could be a product manager. Give us advice on who this person should be?
Practical Examples of LLM Judges
Hamel Husain: Yeah. It should be the person with domain expertise. So in this case, it would be the person who understands the business of leasing, apartment leasing, and has context to understand if this makes sense. It’s always a domain expert, like you said. Okay. For legal, it would be a law person. For mental health, it would be the mental health expert, whether that’s a psychiatrist or someone else.
Lenny Rachitsky: Cool.
Building Judge Prompts & Human Alignment
Hamel Husain: Though oftentimes, it is the product manager.
Lenny Rachitsky: Cool. So the advice here is pick that person. It may not feel so super fair that they’re the one in charge and they’re the dictator, but they’re benevolent. It’s going to be okay.
Checking the Judge Consistency Matrix
Hamel Husain: Yeah. It’s going to be okay. It’s not perfection. You’re just trying to make progress and get signal quickly so you have an idea of what to work on because it can become infinitely expensive if you’re not careful.
Evals Are the New PRD
Lenny Rachitsky: Yeah. Okay, cool. Let’s go back to your examples.
Hamel Husain: Yeah, no problem. So this is another example where we have someone saying, “Okay. Do you have any specials?” And the assistant or the AI responds, “Hey, we have a 5% military discount.” User responds, and it switches the subject, “Can you tell me how many floors there are? Do you have any one-bedrooms available or one-bedrooms on the first floor?” And the AI responds, “Yeah, okay. We have several one-bedroom apartments available.” And then the user wants to confirm, “Any of those on the first floor and how much are the one-bedrooms?” And then also, it’s a current resident, so they’re also asking, “I need a maintenance request.”
You could see the messiness of the real world in here, and the assistant just calls a tool that says transfer call, but it doesn’t say anything. It just abruptly does transfer call, so it’s pretty jank, I would say. It’s just not-
Who Validates the Validated & Criteria Drift
Lenny Rachitsky: Another jank.
Advanced Data Analysis Techniques
Hamel Husain: Another kind of jank, a different kind of jank. So when you write the open note, you don’t want to say jank, because what we want to do is we want to understand, and when we look at the notes later on, we want to understand what happened.
So you just want to say, “Did not confirm call transfer with user.” And it doesn’t have to be perfect. You just have to have a general idea of what’s going on.
Lenny Rachitsky: Cool.
Hamel Husain: So, okay. So let’s say we do, and Shreya and I, we recommend doing at least 100 of these. The question is always, “How many of this do you do?” And so there’s not a magic number. We say 100 just because we know that as soon as you start doing this, once you do 20 of these, you will automatically find it so useful that you will continue doing it.
So we just say 100 to mentally unblock you, so it’s not intimidating. It’s like, “Don’t worry, you’re only going to do 100.” And there is a term for that, so the right answer is, “Keep looking at traces until you feel like you’re not learning anything new.” Maybe Shreya should talk about-
Further Applications for LLM Judges
Shreya Shankar: Yeah. So there’s actually a term-
Hamel Husain: … that.
Shreya Shankar: … in data analysis and qualitative analysis called theoretical saturation. So what this means is when you do all of these processes of looking at your data, when do you stop? It’s when you are theoretically saturating or you’re not uncovering any new types of notes, new types of concepts, or nothing that will materially change the next part of your process.
And this kind of takes a little bit of intuition to develop, so typically, people don’t really know when they’ve reached theoretical saturation yet. That’s totally fine. When you do two or three examples or rounds of this, you will develop the intuition. A lot of people realize, “Oh, okay. I only need to do 40, I only need to do 60. Actually, I only need to do 15.” I don’t know. Depends on the application and depends on how savvy you are with error analysis for sure.
Controversies Surrounding AI Evals
Lenny Rachitsky: And your point about you’re going to want to do a bunch. I imagine it’s because you’re just like, “Oh, I’m discovering all these problems. I got to see what else is going on here.”
Shreya Shankar: Exactly.
The Evals vs. A/B Testing Debate
Lenny Rachitsky: Is that right?
Shreya Shankar: And promise, at some point, you’re not going to discover new types of problems.
How Evals Relate to A/B Testing
Lenny Rachitsky: Yeah. Awesome. So let’s say you did 100 of these, what’s the next step?
Hamel Husain: Yeah. Okay. So you did 100 of these. Now you have all these notes. So this is where you can start using AI to help you. So the part where you looked at this data is important, like we discussed. You don’t want to automate this part too much.
OpenAI Acquires Statsig
Lenny Rachitsky: Humans will still have jobs. This is the takeaway here. That’s great.
Hamel Husain: Yes.
Common Misconceptions About Evals
Lenny Rachitsky: Just reviewing traces. At least there’s one job left for now. Great.
Hamel Husain: So, yeah. Exactly. And so, okay. You have all these notes. Now, to turn this into something useful, you can do basic counting. So basic counting is the most powerful analytical technique in data science because it’s so simple and it’s kind of undervalued in many cases, and so it’s very approachable for people.
And so the first thing you want to do is take these notes, and you can categorize them with an LLM, and so there’s a lot of different ways to do that. Right before this podcast, I took three different coding agents or AI tools in how to categorize these notes. So one is, “Okay, I uploaded into a cloud project, I uploaded a CSV of these notes, and I just exported them directly from this interface.” There’s a lot of different ways to do this, but I’m showing you the simple, stupid way, the most basic way of doing things.
And so I dumped the CSV in here and I said, “Please analyze the following CSV file.” And I told it there’s a metadata field that has a note in it, but what I said is I used the word open codes, and I said, “Hey, I have different open codes,” and that’s a term of art. LLMs know what open codes are and they know what axial codes are because it is a concept that’s been around for a really long time, so those words help me shortcut what I’m trying to do.
Practical Advice for Evals
Lenny Rachitsky: That’s awesome. And the end of the prompt is telling it to create axial codes?
Building Your Own Data Viewer
Hamel Husain: Yes. Creating axial codes, so what it does is-
Time Investment and Ongoing Maintenance
Shreya Shankar: So maybe it’s worth talking about what are axial codes or what’s the point here? You have a mess of open codes, and you don’t have 100 distinct problems. Actually, many of them are repeats, but because you phrased them differently, and that you shouldn’t have tried to create your taxonomy of failures as you’re open coding. You just want to get down what’s wrong and then organize, “Okay, what’s the most common failure mode?”
So the purpose, axial code basically is just a failure mode. It’s the label or category. And what our goal is, is to get to this clusters of failure modes and figure out what is the most prevalent, so then you can go and run and attack that problem.
Data-Driven Product Insights
Lenny Rachitsky: That is really helpful. Basically, just synthesizing all these-
Shreya Shankar: Absolutely.
Recommended Evals Courses
Lenny Rachitsky: … into categories and themes. Super cool. And we’ll include this prompt in our show notes for folks so they don’t have to sit there and screenshot it and try to type it up themselves.
Hamel Husain: Yeah. Great idea. And so Claude went ahead and analyzed the CSV file and decided how to parse it, blah, blah, blah. We don’t need to worry about all that stuff, but it came up with a bunch of axial codes. Basically, axial codes are categories, like Shreya said. So one is, okay, capability limitations, misrepresentation, process and protocol violations, human handoff issues, communication, quality. It created these categories.
Now, do I like all the categories? Not really. I like some of them. It’s a good first stab at it. I would probably rename it a little bit because some of them are a bit too generic. Like what is capability limitation? That’s a little bit too broad. It’s not actionable. I want to get a little bit more actionable with it so that if I do decide it’s a problem, I know what to do with it, but we’ll discuss that in a little bit. So you can do this with anything, and this is the dumbest way to do it, but dumb sometimes is a good way to get started, so-
Rapid Fire Q&A
Lenny Rachitsky: And this is what LLMS are really good at, taking a bunch of information and synthesizing it.
Shreya Shankar: Absolutely. Synthesizing for us to make sense of, right? Note that it’s not automatically proposing fixes or anything, that’s our job, but now, we can wade through this mess of open codes a lot easier.
Another thing that’s interesting here in this prompt to generate the axial codes is you can be very detailed if you want, right? You can say, “I want each axial code to actually be some actionable failure mode,” and maybe the LLM will understand that and propose it, or, “I want you to group these open codes by what stage of the user story that it’s in.” So this is where you can be creative or do what’s best for you as a product manager or engineer working on this, and that will help you do the improvement later.
Lenny Rachitsky: So there’s no definitive prompt of, “Here’s the one way to do it”?
Shreya Shankar: Absolutely.
Lenny Rachitsky: You’re saying you can iterate, see what works for you?
Shreya Shankar: Absolutely.
Lenny Rachitsky: It’s interesting the tools don’t do this, or do they try and they just don’t do a great job?
Shreya Shankar: No, I don’t think they do it. We’ve been screaming from the rooftops, “Please, please-”
Lenny Rachitsky: Oh, wow.
Shreya Shankar: ”… do this.” I do think it’s a little bit hard, right? Part of this whole experience with the eval scores Hamel and I are teaching are a lot of people don’t actually know this, so maybe it’s that people don’t know this and they don’t know how to build tools for it. And hopefully, we can demystify some of this magic.
Lenny Rachitsky: And just to double-click on this point, this is not a thing everyone does or knows. This is something you two developed based on your experience doing data analysis and data science at other companies?
Shreya Shankar: Well, I want to caveat that we didn’t invent error analysis. We don’t actually want to invent things. That’s bad signal. If somebody is coming to you with a way to do something that’s entirely new and not grounded in hundreds of years of theory and literature, then you should, I don’t know, be a little bit wary of that.
But what we tried to do was distill, “Okay, what are the new tools and techniques that you need to make sense of the LLM error-out analysis?” And then we created a curriculum or structured way of doing this. So this is all very tailored to LLMs, but the terms open coding, axial coding, are grounded in social science.
Lenny Rachitsky: Amazing. Okay. What’s funny about you guys doing this is I just want to go do this somewhere. I don’t have any AI product to do this on, but it’s just like, “Oh, this would be so fun.” Just sit there and find all the problems I’m running into and categorize them and then try to fix them.
Shreya Shankar: I love that.
Lenny Rachitsky: Hamel pulled up a video. What do you got going on here?
Hamel Husain: Yeah. So I pulled up a video just to drive home Shreya’s point. We are not inventing anything, so what you see on the screen here is Andrew Ng, one of the famous machine learning researchers in the world who have taught a lot of people, frankly, machine learning. And you can see this is an eight-year-old video, and he’s talking about error analysis.
And so this is a technique that’s been used to analyze stochastic systems for ages, and it’s something that it was just using the same machine learning ideas and principles, just bringing them into here, because again, these are stochastic systems.
Lenny Rachitsky: Awesome. Well, one thing, we’re working on getting Andrew on the podcast, we’re chatting, so that will-
Shreya Shankar: Nice.
Lenny Rachitsky: … be really fun. Two, I love that my podcast episode just came out today is in your feed there, and it’s standing out really well in that feed, so I’m really happy about that [inaudible 00:39:13].
Hamel Husain: Very nice. Yeah. The recommendation algorithm is quite good.
Lenny Rachitsky: Yes. Here we go. Hope you click on that. Don’t screw my algorithm. Okay, cool. So we’ve done some synthesis. I know we’re not going to go through the entire step. This is you have a whole course that takes many days to learn this whole process. What else do you want to share about how to go about this process?
Hamel Husain: Okay. So you can do this through anything, and the same thing works just fine in ChatGPT, the same exact prompt. You can see it made axial codes. I really like using Julius AI. It’s one of my favorite tools.
Julius is kind of this third-party tool that uses notebooks. I personally like Jupiter notebooks a lot, and so it’s more of a data science thing, but a lot of product managers that are kind of learning notebooks nowadays, and it’s kind of cool. It’s like a fun playground where you can write code and look at data. But we don’t have to go deeply into that. Just wanted to mention, you can use a lot. AI is really good at this.
So let’s go to the fun part. Here we go. So now we have these axial codes. So the first thing I like to do, I have these open codes, and I have the axial codes, let’s say, that we assigned from the cloud project or the ChatGPT. And so what I do is I collect them first and I take a look, like, “Does these axial codes make sense?” And I look at the correspondence between the different axial codes and the open codes, and I go through an exercise and I say, “Hmm. Do I like these codes? Can I make them better? Can I refine them? Can I make them more specific?” Instead of being generic, I make them very specific and actionable.
So you see the ones that I came up with here are tour scheduling, rescheduling issues, human handoff or transfer issue, formatting error with an output, conversational flow. We saw the conversational flow issue with the text messages. Making follow-up promises not kept.
And so basically, what I can do, what you can do now is you have these axial codes, and so I just collect them into a list, so this is an Excel formula. Just collect these codes into a list, and now we have a comma-separated list of these codes. And then what you can simply do is you could take your notes that you have, those open codes, and you can tell an AI, and this is using Gemini and AI just for simplicity, this is, again, we’re trying to keep it simple, categorize the following note into one of the following categories as always.
Lenny Rachitsky: For folks watching, I like all these different prompts and formulas you’re sharing. This is the Google Sheets AI prompt.
Shreya Shankar: Huge fan.
Hamel Husain: And so basically, what you could do is you can categorize your traces into one of the buckets, and that’s what we have here. We have categorized all those problems that we encountered into one of these things.
Shreya Shankar: And this is automatic, which is very exciting. I mean, the AI is doing it. So this also drives home the point that your open codes have to be detailed, right? You can’t just say janky because if the AI is reading janky, it’s not going to be able to categorize it. Even a human wouldn’t, right? It would have to go and remember why you said janky, so it’s important to be somewhat detailed in your open code.
Lenny Rachitsky: Okay. So avoid the word janky. It’s a good rule of thumb.
Shreya Shankar: Yeah. Or have it with 10 other words.
Lenny Rachitsky: Oh, okay. What is-
Hamel Husain: Yeah. I was being funny.
Lenny Rachitsky: Yeah, okay. What are some of those other words that people often use that you think are not good?
Shreya Shankar: I don’t think it’s specific words. I think it’s just people are not detailed enough in the open code, so it’s hard to do the categorization.
Lenny Rachitsky: Great. And by the way, the reason you have to map them back is because, say, Claude or ChatGPT gave you suggestions and you change them and iterated on them, so you can’t just go back and say, “Cool, whatever,” in each bucket?
Hamel Husain: Yeah, yeah.
Lenny Rachitsky: Great.
Hamel Husain: That’s a really good question, actually. It’s good to iterate and think about it a little bit like, “Do I like these open codes? Do these actually make sense to me?” Just like anything that AI does, it’s really good to kind of put yourself in the middle just a little bit.
Lenny Rachitsky: It’s in the loop. Still space for us. Great.
Shreya Shankar: One of the things that I like to do with this step if I’m trying to use AI to do this labeling, is also have a new category called none of the above. So an AI can actually say, “None of the above,” in the axial code, and that informs me, “Okay, my axial codes are not complete. Let’s go look at those open codes, let’s figure out what some new categories are or figure out how to reword my other axial codes.”
Lenny Rachitsky: Awesome. And what’s cool about this is you don’t need to do this many, many times.
Shreya Shankar: No.
Lenny Rachitsky: For most products, you do this process once, and then you build on it, I imagine, and you just tweak it over time?
Shreya Shankar: Absolutely. And it gets so fast. People do this once a week, and you can do all of this in 30 minutes, and suddenly your product is so much better than if you were never aware of any of these problems.
Lenny Rachitsky: Yeah. It’s absurd to feel like you wouldn’t know this is happening. Watching this happening, I’m like, “How could you not do this to your product?”
Shreya Shankar: A lot of people have no idea.
Lenny Rachitsky: Most people. Yeah. We’ll talk about that. There’s a whole debate around this stuff that we want to talk about. Okay, cool. So you have the sheet. What comes next?
Hamel Husain: Okay. So here’s sort of the big unveil. This is the magic moment right now. So we have all these codes that we applied, the ones that we like on our traces. Now, you can do the ta-da, you can count them.
So here’s a pivot table, and we just can do pivot table on those, and we can count how many times those different things occurred. So what do we find? Find on these traces that we categorized? We found 17 conversational flow issues. And I really like pivot tables because you can do cool things. You can double-click on these. You can say, “Oh, okay. Let me take a look at those,” but that’s going into an aside about pivot tables, how cool they are.
But now, we have just a nice, rough cut of what are our problems? And now, we have gone from chaos to some kind of thinking around, “Oh, you know what? These are my biggest problems. I need to fix conversational issues, maybe these human handoff issues.” It’s not necessarily the count is the most important thing. It might be something that’s just really bad and you want to fix that, but okay. Now, you have some way of looking at your problem, and now you can think about whether you need evals for some of these.
So there might be some of these things that might be just dumb engineering errors that you don’t need to write an eval for because it’s very obvious on how to fix them. Maybe the formatting error with output, maybe you just forgot to tell the LLM how you want it to be formatted, and you didn’t even say that in the prompt. So just go ahead and fix the prompt maybe, and we can decide, “Okay, do you want to write an eval for that?” You might still want to write an eval for that because you might be able to test that with just code. You could just test the string, does it have the right formatting potentially? Without running an LLM.
So there’s a cost-benefit trade-off to evals. You don’t want to get carried away with it, but you want to usually ground yourself in your actual errors. You don’t want to skip this step. And so the reason I’m kind of spending so much time on this is this is where people get lost. They go straight into evals like, “Let me just write some tests,” and that is where things go off the rails.
Okay. So let’s say we want to tackle one of these things. So for example, let’s say we want to tackle this human handoff issue, and we’re like, “Hmm, I’m not really sure how to fix this. That’s a kind of subjective sort of judgment call on should we be handing off to a human? And I don’t know immediately how to fix it. It’s not super obvious per se. Yeah. I can change my prompt, but I’m not sure. I’m not 100% sure.”
Well, that might be sort of an interesting thing for an LLM as a judge, for example. So there’s different kinds of evals. One is code-based, which you should try to do if you can because they’re cheaper. LLM as a judge is something, it’s like a meta eval. You have to eval that eval to make sure the LLM that’s judging is doing the right thing, which we’ll talk about in a second.
So, okay. LLM as a judge, that’s one thing. Okay. How do you build an LLM as a judge?
Lenny Rachitsky: Before we get into that actually, just to make sure people know exactly what you’re describing there, these two types of evals. One is you said it’s code-based and one is LLM as judge. Maybe Shreya, just help us understand what code-based eval even is? It’s essentially a unit test? Is that a simple way to think about it?
Shreya Shankar: Yeah. Maybe eval is not the right term here, but think automated evaluator. So when we find these failure modes, one of the things we want is, “Okay. Can we now go check the prevalence of that failure mode in an automated way without me manually labeling and doing all the coding and the grouping, and I want to run it on thousands and thousands of traces, I want to run it every week.” That is, okay. You should probably build an automated evaluator to check for that failure mode.
Now, when we’re saying code-based versus LLM-based, we’re saying, “Okay. So maybe I could write a Python function or a piece of code to check whether that failure mode is present in a trace or not.” And that’s possible to do for certain things like checking the output is JSON, or checking that it’s markdown, or checking that it’s short. These are all things you can capture in code or you could approximately capture in code.
When we’re talking about LLM judge here, we’re saying that this is a complex failure mode and we don’t know how to evaluate in an automated way. So maybe we will try to use an LLM to evaluate this very, very narrow, specific failure mode of handoffs.
Lenny Rachitsky: So just to try to mirror back what you’re describing, you want to test what your, say, agent or AI product is doing. You ask it a question, it gets back with something.
One way to test if it’s giving you the right answer is if it’s consistently doing the same thing, that you could write a code to tell you this is true or false. For example, will it ever say there’s a virtual tour? So you could ask it.
Shreya Shankar: Yes.
Lenny Rachitsky: “Do you provide virtual tours?” It says yes or no, and then you could write code to tell you if it’s correct based on that specific answer.
But if you’re asking about something more complicated and it’s not binary, in one world, you need a human to tell you this is correct. The solution to avoid humans having to review all this every time automatically is LLMs replacing human judgment, and you’d call it an LLM as judge. The LLM as being the judge if this is correct or not.
Shreya Shankar: Absolutely. You nailed it.
Lenny Rachitsky: Great.
Shreya Shankar: So people always think, “Oh, this is at least as hard as my problem of creating the original agent.” And it’s not, because you’re asking the judge to do one thing, evaluate one failure mode, so the scope of the problem is very small and the output of this LLM judge is pass or fail. So it is a very, very tightly scoped thing that LLM judges are very capable of doing very reliably.
Lenny Rachitsky: And the goal here is just to have a suite of tests that run before you ship to production that tell you things are going the way you want them to? The way your agent is interacting is correct?
Shreya Shankar: The beautiful thing about LLM judges, you can use them in unit tests or CI, sure, but you could also use it online for monitoring, right? I can sample 1000 traces every day, run my LLM judge, real production traces, and see what the failure rate is there. This is not a unit test, but still now we get an extremely specific measure of application quality.
Lenny Rachitsky: Cool. That’s a really great point because a lot of people just see evals for being this not-real-life thing. It’s a thing that you test before it’s actually in the real world. And what’s actually happening in the real world, you’re saying you should actually do exactly that?
Shreya Shankar: Yeah.
Lenny Rachitsky: Test your real thing running in production? And it’s a daily, hourly sort of thing you could be running?
Shreya Shankar: Totally.
Lenny Rachitsky: Awesome. Okay. Hamel’s got an example of an actual LLM as a judge eval here, so let’s take a look.
Hamel Husain: I love how Shreya really teed it up for me, so thank you so much. So what we have is a LLM as a judge prompt for this one specific failure. Like Shreya said, you would want to do one specific failure and you want to make it binary because we want to simplify things. We don’t want, “Hey, score this on a rating of one to five. How good is it?” That’s just in most cases, that’s a weasel way of not making a decision. Like, “No, you need to make a decision. Is this good enough or not? Yes or no?”
It can be painful to think about what that is, but you should absolutely do it. Otherwise, this thing becomes very untractable, and then when you report these metrics, no one knows what 3.2 versus 3.7 means, so.
Shreya Shankar: Yeah. We see this all the time also, and even with expert-curated content on the internet where it’s like, “Oh, here’s your LLM judge evaluator prompt. Here’s a one-to-seven scale.”
And I always text Hamel like, “Oh, no. Now, we have to fight the misinformation again because we know somebody is going to try it out and then come back to us and say, ‘Oh, I have 4.2 average,’” and we’re going to be like, “Okay.”
Lenny Rachitsky: It’s wild how much drama there is in the evals space. We’re going to get to that. Oh, man.
Meticulously designed to be an intuitive and simple experience, and Mercury brings all the ways that you use money into a single product, including credit cards, invoicing, bill pay, reimbursements for your teammates and capital. Whether you’re a funded tech startup looking for ways to pay contractors and earn yield on your idle cash, or an agency that needs to invoice customers and keep them current, or an e-commerce brand that needs to stay on top of cash flow and access capital, Mercury can be tailored to help your business perform at its highest level. See what over 200,000 entrepreneurs love about Mercury. Visit mercury.com to apply online in 10 minutes. Mercury is a fintech, not a bank. Banking services provided through Mercury’s FDIC insured partner banks. For more details, check out the show notes.
Hamel Husain: Okay, so this is your judge prompt. There’s no one way to do it. It’s okay to use an LLM to help you create it, but again, put yourself in the loop. Don’t just blindly accept what the LLM does, and in all of these cases, that’s what we did. With the axial codes, we iterated on this. You can use an LLM to help you create this prompt, but make sure you read it, make sure you edit it, whatever. This is not necessarily the perfect prompt. This is just the stupid, keeping it very simple just to show you the idea. It’s like, “Okay, for this handoff failure,” I said, “Okay, I want you to output true or false,” it’s a binary judge. That’s what we recommend. Then I just go through and say, “Okay, when should you be doing a handoff?” And I just list them out.
Okay, explicit human requests ignored or looped, some policy-mandated transfer, sensitive resident issues, tool data, unavailability, same day walk-in or tour requests. You need to talk to a human for that, so on and so forth. The idea is, now that I know that this is a failure from my data, I’m interested in iterating on it, because I know this is actually happening all the time. Like Shreya said, it would be nice to have a way not only to evaluate this on the data I have, but also on production data, just to get a sense of, what scales is this happening? Let me find more traces, let me have a way to iterate on this. We can take this prompt and I’m going to use the spreadsheet again. The first step is, okay, when I’m doing this judge… I wrote the prompt.
Now, a lot of people stop there and they say, “Okay, I have my judge prompt. We’re done. Good, let’s just ship it,” and the prompt says… If the judge says it’s wrong, it’s wrong. They just accept it as the gospel, be like, “Okay, the LLM says it’s wrong, it must be wrong. Don’t do that, because that’s the fastest way that you can have evals that don’t match what’s going on, and when people lose trust in your evals, they lose trust in you. It’s really important that you don’t do that, so before you release your LLM as a judge, you want to make sure it’s aligned to the human. How do you do that? You have those axial codes and you want to measure your judge against the axial code, and say like, “Hey, does it agree with me? My own judge, does it agree with me?” Just measure it.
What we have here is, okay, I say, “Assess this LLM trace.” Again, I’m using just spreadsheets here, “Assess this LM trace according to these rules,” and the rules are just the prompt that I just showed you. I ask it, “Okay, is there a handoff error, true or false?” Then this column, let me just zoom in a bit. Column H, I have, “Okay, did this error occur?” Column G is whether I thought the error occurred or not. You can see-
Lenny Rachitsky: You’re going through manually, you do that.
Hamel Husain: Yeah, yeah, which we already did. We already went through it manually. It’s not like we have to do it again, because we have that cheat code from the axial coding, we already did it. You might have to go through it again if you need more data, and there’s a lot of details to this on how to do this correctly. You want to split your data and do all these things, so that you’re not cheating, but I just want to show you the concept. Basically, what you can do is measure the agreement. Now, one thing you should know, as a product manager, is a lot of people go straight to this agreement. They say, “Okay, my judge agrees with the human some percentage of the time.”
Now that sounds appealing, but it’s a very dangerous metric to use, because a lot of times, errors, they only happen on the long tail and they don’t happen as frequently, so if you only have the error 10% of the time, then you can easily have 90% agreement by just having a judge say it passes all the time. Does that make sense? 90% agreement look good on paper, but it might be misleading.
Lenny Rachitsky: It’s rare, it’s a rare error. Yeah.
Hamel Husain: As a product manager or someone, even if you’re not doing this calculation yourself, if someone ever reports to you agreement, you should immediately ask, “Okay, tell me more.” You need to look into it. They give you more intuition, here is like a matrix of this specific judge in the Google sheet, and this is, again, a pivot table, just keeping it dumb and simple. “Okay, on the rows I have, what did the human think? What did I think? Did it have an error, true or false? Then did my judge have an error, true or false?”
Shreya Shankar: The intuition here is exactly what Hamel said, where you need to look at each type of error. When the human said false, but the judge said true, or vice versa, so those non-green diagonals here, and if they’re too large, then go iterate on your prompt, make it more clear to the LLM judge, so that you can reduce that misalignment. You want to get to a point where most… You’re going to have some misalignment, that’s okay. We talk about in our course, also how to code correct that misalignment, but in this stage, if you’re a product manager and the person who’s building the LLM judge eval has not done this, they’re saying like, “It agrees 75% of the time, we’re good.” They don’t have this matrix and they haven’t iterated to make sure that these two types of errors have gone down to zero, then it’s a bad smell. Go and ask them to go fix that.
Lenny Rachitsky: Awesome. That’s a really good tip, what to look for when someone’s doing this wrong.
Shreya Shankar: Yeah.
Lenny Rachitsky: Actually, can you take us back to the LLM as judge prompt? I just want to highlight something really interesting here. I’ve had some guests on the podcast recently who’ve been saying, “Evals are the new PRDs,” and if you look at this, this is exactly what this is. Product managers, product teams, here’s what the product should be, here’s all the requirements, here’s the how it should work. They built a thing and then they test it. Manually, often. What’s cool about this is this is exactly that same thing, and it’s running constantly. It’s telling you, “Here’s how this agent should respond,” and it’s very specific ways. “If it’s this, this, this, do that. If it’s this, this, that, do that.” It’s exactly what I’ve been hearing again and again, you could see right here. This is the purest sense of what a product requirements document should be, is this eval judge that’s telling you exactly what it should be, and it’s automatic and running constantly.
Shreya Shankar: Yeah, absolutely. It’s derived from our own data, so of course, it’s a product manager’s expectations. What I find that a lot of people miss is they just put in what their expectations are before looking at their data, but as we look at our data, we uncover more expectations that we couldn’t have dreamed up in the first place, and that ends up going into this prompt.
Lenny Rachitsky: That is interesting. Your advice is not skip straight to evals and LLM as judge prompts before you build the product, still write traditional one-pagers PRDs to tell your team what we’re doing, why we’re doing it, what success looks like. But then at the end, you could probably pull from that and even improve that original PRD if you’re evolving the product using this process.
Shreya Shankar: I would go even further to say you’re going to improve… It’s going to change. You’re never going to know what the failure modes are going to be upfront, and you’re always going to uncover new vibes that you think that your product should have. You don’t really know what you want until you see it with these LLMs, so you got to be flexible, have to look at your data, have to… PRDs are a great abstraction for thinking about this. It’s not the end all, be all. It’s going to change.
Lenny Rachitsky: I love that, and Hamel’s pulling up some cool research report. What’s this about?
Hamel Husain: This is one of the coolest research reports you can possibly read if you want to know about evals. It was authored by someone named Shreya Shankar.
Shreya Shankar: Oh, my God.
Hamel Husain: And her collaborators. It’s called “Who Validates the Validated?”
Lenny Rachitsky: That’s the best name for a researcher.
Shreya Shankar: Thank you, thank you.
Hamel Husain: I should let Shreya talk about this. I think one of the most important things to pay attention in this paper are the criteria drift, and what she found.
Shreya Shankar: We did this super fun study when we were doing user studies with people who were trying to write LLM judges or just validate their own LLM outputs. I think this was before evals was extremely popular, I feel like, on the internet. We did this project late 2023 was when we started it. But then the thing that really was burning in my mind as a researcher is like, “Why is this problem so hard? We’ve been having machine learning and AI for so long, it’s not new, but suddenly, this time around, everything is really difficult.” We just did this user study with a bunch of developers and we realized, “Okay, what’s new here is that you can’t figure out your rubrics upfront. People’s opinions of good and bad change as they review more outputs, they think of failure modes only after seeing 10 outputs they would never have dreamed of in the first place,” and these are experts. These are people who have built many LLM pipelines and now agents before, and you can’t ever dream up everything in the first place. I think that’s so key in today’s world of AI development.
Lenny Rachitsky: That is a really good point. That’s very much reinforcing what we were just talking about and that’s why I’ll pull this up, is just… Okay-
Shreya Shankar: The research behind it.
Lenny Rachitsky: Yeah, okay, great. You still got to do product the same way, but now you have this really powerful tool that helps you make sure what you’ve built is correct. It’s not going to replace the PRD process. Cool. How many, say, I don’t know, LLM as judge prompts, do you end up with usually say… I don’t know. I know, obviously, depends complexity to the product, but what’s a number in your experience?
Shreya Shankar: For me, between four and seven.
Lenny Rachitsky: That’s it.
Shreya Shankar: It’s not that many, because a lot of the failure modes, as Hamel said earlier, can be fixed by just fixing your prompt. You just didn’t think to put it in your prompts, so now you put it in your… You shouldn’t do an eval like this for everything, just the pesky ones that you’ve described your ideal behavior in your agent prompt, but it’s still failing.
Lenny Rachitsky: Got it. Say you found a problem, you fixed it. In traditional software development, you’d write a unit test to make sure it doesn’t happen again. Is your insight here is, “Don’t even bother writing an eval around that if it’s just gone”?
Shreya Shankar: I think you can if you want to, but the whole game here is about prioritizing. You have finite resources and finite time, you can’t write an eval for everything, so prioritize the ones that are the more pesky areas.
Lenny Rachitsky: Probably the ones that are most risky to your business if they say something like Mecha Hitler, Grok.
Shreya Shankar: Yikes.
Lenny Rachitsky: Cool. Okay, so that’s very relieving, because this prompt was a lot of work to really think through all these details.
Shreya Shankar: But it’s a lot of one-time cost. Right now, forever, you can run this on your application.
Hamel Husain: Okay, data analysis is super powerful, is going to drive lots of improvements very quickly to your application. We showed the most basic kind of data analysis, which is counting, which is accessible to everyone. You can get more sophisticated with the data analysis. There’s lots of different ways to sample, look at data. We made it look easy in a sense, but there’s a lot of skills here to do to it well. Building an intuition and a nose for how to sort through this data. For example, let’s say I find conversational issues, this conversational flow issues. Maybe if I was trying to chase down this problem further, I would think about ways to find other conversational flow issues that I didn’t code. I would maybe dig through the data in several ways, and there’s different ways to go about this. It’s very similar, if not almost exactly similar as traditional analytics techniques that you would do on any product.
Lenny Rachitsky: Give us just a quick sense of what comes next and then let’s talk about the debate around evals and a couple more things.
Shreya Shankar: What comes next after you’ve built your LLM judge? Well, we find that people just try to use that everywhere they can, so they’ll put the LLM judge in unit tests and they will build, “Here are some example traces where we saw that failure, because we labeled it. Now we’re going to make those part of unit tests and make sure that, every time we push a change to our code, these tests are going to pass.” They also use it for online monitoring. People are making dashboards on this, and I think that’s incredible. I think the products that are doing this, they have a very sharp sense of how well their application is performing, and people don’t talk about it, because this is their moat. People are not going to go and share all of these things, because it makes sense. If you are an email-writing assistant, and you’re doing this and you’re doing it well, you don’t want somebody else to go and build an email-writing assistant and then get you out of business.
I really want to stress the point that it’s try to use these artifacts that you’re building wherever possible online, repeatedly use them to drive improvements to your product. Oftentimes, Hamel and I will tell people how to do this up to this very point, and it clicks for people and then they never come back again. Either they have, I don’t know, quit their jobs, they’re not doing AI development anymore, or they know what to do from here on out. I think it’s the latter, but I think it’s very powerful.
Lenny Rachitsky: Just watching you do this really opened my eyes to what this is and how systematic the process is. I always imagine you just sit on a computer, “Okay, what are the things I need to make sure work correctly?” What you’re showing us here is it’s a very simple step-by-step based on real things that are happening in your product, how to catch them, identify them, prioritize them, and then catch them if they happen again and fix them.
Shreya Shankar: Yeah, it’s not magic. Anyone can do this, you’re going to have to practice the skill, like any new skill, you have to practice, but you can do it. I think what’s very empowering now is that product managers are doing this and can do this, and can really build very, very profitable products with this skill set.
Lenny Rachitsky: Okay, great segue to a debate that we got pulled into that was happening on X the other day. I did not realize how much controversy and drama there is around evals. There’s a lot of people with very strong opinions. How about Shreya? Give us just a sense of the two sides of the debate around the importance and value of evals, and then give us your perspective.
Shreya Shankar: Yeah. All right, I’ll be a little bit placating and I say I think everyone is on the same side. I think the misconception is that people have very rigid definitions of what evals is. For example, they might think that evals is just unit tests or they might think that evals is just the data analysis part and no online monitoring or no monitoring of product-specific metrics, like actually number of chats engaged in or whatnot. I think everyone has a different mindset of evals going in, and the other thing I will say is that people have been burned by evals in the past. I think people have done evals badly. One concrete example of this is they’ve tried to do an LLM judge, but it has not aligned with their expectations. They only uncovered this later on and then they didn’t trust it anymore, and then they’re like, “I’m anti evals.”
I 100% empathize with that, because you should be anti Likert scale LLM judge. I absolutely agree with you, we are anti that as well. A lot of the misconception stems from two things, like people having a narrow definition of evals and then people not doing it well and then getting burned and then wanting to avoid other people making that mistake. Then, unfortunately, X or Twitter is a medium where people are misinterpreting what everybody is saying all the time, and you just get all these strong opinions of, “Don’t do evals, it’s bad. We tried it, it doesn’t work. We’re Claude Code,” or whatever other famous product, “And we don’t do evals.” There’s just so much nuance behind all of it, because a lot of these applications are standing on the shoulders of evals. Coding agents is a great example of that, Claude Code. They’re standing on the shoulders of Claude base model… Not base, but the fine-tuned Claude models have been evaluated on many coding benchmarks. Can’t argue against that.
Lenny Rachitsky: Just to make clear exactly what you’re talking about there, one of the heads, I think maybe the head engineer of Claude Code, went on a podcast and he’s like, “We don’t do evals, we just vibe. We just look at vibes,” and vibes meaning they just use it and feel if it’s right or wrong.
Shreya Shankar: I think that works. There’s two things to that, right? One is they’re standing on the shoulders of the evals that their colleagues are doing for coding.
Lenny Rachitsky: Of the Claude foundational model.
Shreya Shankar: Absolutely, right? We know that they report those numbers, because we see the benchmarks, we know who’s doing well on those. The other thing is they are actually probably very systematic about the error analysis to some extent. I bet you that they’re monitoring who is using Claude, how many people are using Claude, how many traps are being created, how long these chats are. They’re also probably monitoring in their internal team, they’re dogfooding. Anytime something is off, they maybe have a cue or they send it to the person developing Claude Code, and this person is implicitly doing some form of hair error analysis that Hamel talked about. All of this is evals, right? There’s no world in which they’re just being like, “I made Claude Code, I’m never looking at anything,” and unfortunately, when you don’t think about that or talk about that, I think that the community…
Most of the community is beginners or people who don’t know about evals and want to learn about it, and it sends the wrong message there. Now, I don’t know what Claude Code is doing, obviously, but I would be willing to bet money that they’re doing something in the form of evals.
Hamel Husain: We’ll also say that coding agents are fundamentally very different than other AI products, because the developer is the domain expert, so you can short circuit a lot of things, and also, the developer is using it all day long, so there’s a type of dogfooding and type of domain expertise that is… You can collapse the activities, you don’t need as much data, you don’t need as much feedback or exploration, because you know, so your eval process should look different.
Lenny Rachitsky: Because you’re seeing the code, you see the code it’s generating. You can tell, “This is great, this is terrible.”
Hamel Husain: Yeah, yeah. I think a lot of people had generalized coding agents, because coding agents are the first AI product released into the wild, and I think it’s a mistake to try to generalize that at large.
Shreya Shankar: The other thing is, yeah, engineers have a dogfooding personality. There are plenty of applications where people are trying to build AI in certain domains and they don’t have dogfooding for doctors, for example, or not out there trying to get all the most incorrect advice from AI and be tolerant and receptive to that. It’s very important to keep, I think these nuanced things in mind.
Lenny Rachitsky: What I’m hearing from you, Shreya, interestingly, is that if humans on the team are doing very close data analysis, error analysis, dogfooding like crazy, and essentially, they’re the human evals and you’re describing that as that’s within the umbrella of evals. You could do it that way if you have time and motivation to do that, or you could set these things up to be automatic.
Shreya Shankar: Absolutely, it’s also about the skills. People who work at Anthropic are very, very highly skilled. They’ve been trained in data analysis or software engineering or AI, and whatnot. You can get there, anyone can get there, of course, by learning the concepts, but most people don’t have that skill right now.
Hamel Husain: Dogfooding is a dangerous one, only because a lot of people will say they’re dogfooding. They’re like, “Yeah, we dogfooded,” but are they, really? A lot of people aren’t really dogfooding it at that visceral level that you would need to close that feedback loop. That’s the only caveat I would add.
Lenny Rachitsky: There’s also this, feels like, straw man argument of evals versus A-B tests. Talk about your thoughts there, because that feels like a big part of this debate. People are having like, “Do you need evals if you have A-B tests that are testing production level metrics?”
Shreya Shankar: A-B tests are, again, another form of evals ,I imagine, right? When you’re doing an A-B test, you have two different experimental conditions and then you have a metric that quantifies the success of something, and you’re comparing the metric. Again, an eval in our mind is systematic measurement of quality, some metric. You can’t really do an A-B test without the eval to compare, so maybe we just have a different weird take on it.
Lenny Rachitsky: Yeah, okay. What I’m hearing is you consider A-B tests as part of the suite of evals that you do. I think when people think A-B tests, it’s like we’re changing something in the product, we’re going to see if this improves some metric we care about. Is that enough? Why do we need to test every little feature? If it’s impacting a metric we care about as a business, we have a bunch of A-B tests that are just constantly running.
Shreya Shankar: This is now a great point. I think a lot of people prematurely do A-B tests, because they’ve never done any error analysis in the first place. They just have hypothetically come up with their product requirements and they believe that, “We should test these things,” but it turns out, when you get into the data, as Hamel showed, that the errors that you’re seeing are not what you thought what the errors might be. They were these weird handoff issues or, I don’t know, the text message thing was strange. I would say that, if you’re going to do A-B tests and they’re powered by actual error analysis as we’ve shown today, then that’s great, go do it. But if you’re just going to do them, which we find that people try to do, just want to do them based on what you hypothetically think is what is important, then I would encourage people to go and rethink that and ground your hypotheses.
Lenny Rachitsky: Do you have thoughts on what Statsig is going to do at OpenAI? Is there anything there that’s interesting? That was a big deal, a huge acquisition. A- B test company people are like, “A-B test, the future.” Thoughts?
Hamel Husain: Just to add to the previous question a little bit, why is there this debate, A-B testing versus evals? I think, fundamentally, evals is… People are trying to wrap their head around how to improve their applications and fundamentally need to do… Data science is useful in products. Looking at data, doing data analytics. There’s many different suite of tools, and you don’t need to invent anything new. Sure, you don’t need necessarily the whole breadth of data science, and it looks slightly different, just slightly, with LLMs. Your tactics might be different, so really what it is is using analytic tools to understand your product. Now, people say the word “Evals,” trying to carve out this new thing, and saying evals and then A-B testing, but if you zoom out, it’s the same data science as before, and I think that’s what’s causing the confusion is, “Hey, we need data science thinking,” and AI product is helpful to have that thinking in AI products like it is in any product is my take on that.
Lenny Rachitsky: That’s a really good take, I think just the word “Evals” triggers people now.
Shreya Shankar: Yeah.
Lenny Rachitsky: If you just call it, “We’re just doing error analysis, doing data science to understand where our product breaks and just setting up tests to make sure we know-”
Shreya Shankar: That’s boring, sounds boring. No, no, no. We need a mysterious term, like “Evals,” to really get the momentum going. Your question about Statsig, I think it’s very exciting. To be honest, I don’t know much about it, because I just imagine that they’re this company that… There’s a tool that many people use, and maybe it just so happened that OpenAI acquired them. I’m sure they’ve been using them in the past, I’m sure OpenAI’s competitors are using Statsig as well, so maybe there is something strategic in that acquisition. I have no idea, I don’t know anything there, but I think those are really the bigger questions for me than, “Is this fundamentally changing A-B testing or making evals more of a priority?” I think they’ve always been a priority, I think OpenAI has always been doing some form of them, and OpenAI has gone so far, historically speaking, as to go and look at all the Twitter sentiment and try to do some retrospective on that, and then tie that back to their products. Certainly, they’re doing-
Then, tie that back to their products. Certainly, they’re doing some amount of evals before they ship their new foundation models, but they’re going so much beyond and being like, “Okay, let’s find all the tweets that are complaining about it, all the Reddit threads that are complaining about it, and go try to figure out what’s going on.” It goes to show that evals are very, very important. No one has really figured it out yet. People are using all the available sources signal that they can to improve their products.
Hamel Husain: What I’ll say is I’m really hopeful that it might shift or create a focus within OpenAI, hopefully. Up until now, a lot of the big labs understandably focused on general benchmarks like MMLU score, human eval, things like that, which are very important for foundation models. Those not very related to product specific evals, like the ones we talked about today, but handoff and stuff like that, they tend not to correlate.
Shreya Shankar: Yeah, they don’t correlate with math problem-solving, sorry to say.
Hamel Husain: Exactly. If you look at the eval products, let’s say the ones up until recently that some of the big labs have, they don’t have error analysis. They have a suite of generic tools, cosine similarity, hallucination score, whatever, and that doesn’t work. It’s a good first stab at it. It’s okay. At least you’re doing something, getting people, maybe it’s like getting people look at data. But eventually, what we hope to see is, okay, a bit more data science thinking in this eval process. That’s hopefully the tools we’ll get to.
Shreya Shankar: Yeah, Pamela and I should not be the only two people on the planet that are promoting a structured way of thinking about application specific evals. It’s mind-boggling to me. Why are we the only two people doing this the whole world? What’s wrong? I hope that we’re not the only people and that more people catch on.
Lenny Rachitsky: The fact that your course on Maven is the number one highest grossing course in Maven, clearly there’s demand and interest, and there’s more people I think on your side. Interestingly, just as an example you’ve been sharing on Twitter that I think is informative, everyone’s been saying how cloud code doesn’t care about evals. They’re all about vibes, and everyone’s like, and they’re the best coding agent out there, so clearly, this is right. More recently, there’s all this talk about Codex, OpenAI Codex being better and everyone’s switching and they’re so pro evals.
Shreya Shankar: I know.
Lenny Rachitsky: Yeah.
Shreya Shankar: It gets me every time. The Internet’s so inconsistent. My favorite thing was yesterday, I believe, a couple of lab mates and I were out getting dessert or something, and somebody said like, “Oh, do you like Codex or Claude better or whatever?” The other person said, “Oh, I like Claude.” Then, someone else said, “But the new version of Codex is better.” Then, the first person said, “Oh, but the last I checked was two days ago, so maybe my thoughts, maybe I’m not up-to-date.” I was like, “Oh, my God.”
Lenny Rachitsky: So true, so true. This is the world we live in. Oh, my God. Okay. I want to ask about just top misconceptions people have with evals and top tips and tricks for being successful. Maybe just share one or two each of each. Let me just start with misconceptions, and maybe I’ll go to the Hamel first. Just what are a couple of the most common misconceptions people have with eval still?
Hamel Husain: The top one is, “Hey, I can just buy a tool, plug it in, and it’ll do the eval for you. Why do I have to worry about this? We live in the age of AI. Can’t the AI just eval it?” That’s the most common misconception, and people want that so much that people do sell it, but it doesn’t work. That’s the first one.
Lenny Rachitsky: Shoot, many humans are still great. I think that’s great news.
Hamel Husain: The second one that I see a lot is, “Hey, just not looking at the data.” In my consulting, people come to me with problems all the time, and the first thing I’ll say is, “Let’s go look at your traces.” You can see their eyes pop open and be like, “What do you mean?” I’m like, “Yeah, let’s look at it right now.” They’re surprised that I am going to go look at individual traces, and it always 100% of the time learn a lot and figure out what the problem is. I think people just don’t know how powerful looking at the data is like we showed on this podcast.
Shreya Shankar: I would agree with that.
Lenny Rachitsky: Those are the top two? Okay.
Shreya Shankar: Yes.
Lenny Rachitsky: Is there anything else or those are the ones solve those problems.
Shreya Shankar: Oh, those are definitely… Then, I guess the third one I would add is, there’s no one correct way to do evals. There are many incorrect ways of doing evals, but there are also many correct ways of doing it. You got to think about where you are at with your product, how much resources you have, and figure out the plan that works best for you. It’ll always involve some form of error analysis as we showed today, but how you operationalize those metrics is going to change based on where you’re at.
Lenny Rachitsky: Amazing. Okay. What are a couple of just tips and tricks you want to leave people with as they start on their eval journey or just try to get better at something they’re already doing?
Shreya Shankar: Tip number one is just don’t be alarmed or don’t be scared of looking at your data. The process, we try to make it as structured as possible. There are inevitably questions that are going to come up. That’s totally fine. You might feel like you’re not doing it perfectly. That’s also fine. The goal is not to do evals perfectly, it’s to actionably improve your product. We guarantee you, no matter what you do, if you’re doing parts of these process, you’re going to find ways of actionable improvement, and then you’re going to iterate on your own process from there.
The other tip that I would say is, we are very pro-AI. Use LLMs to help you organize any thoughts that you have throughout this entire process. This could be everything ranging from initial product requirements. Figure out how to organize them for yourself. Figure out how to improve on that product requirements doc based on the open codes that you’ve created. Don’t be afraid to use AI in ways that present information better for you.
Lenny Rachitsky: Sweet, so don’t be scared. Use LLMs as much as you can throughout the process.
Shreya Shankar: But not to replace yourself.
Lenny Rachitsky: Right. Okay, great. There’s still jobs. It’s great. Hamel.
Hamel Husain: Yeah. Let me actually share my screen, because I want to show something. To piggyback of what Shreya said is, if you heard any phrase in this podcast, you’ve probably heard look at your data more than anything else. It’s so important that we teach that you should create your own tools to make it as easy as possible. I showed you some tools when we’re going through the live example of how to annotate data. Most of the people I work with, they realize how important this is and they vibe code their own tools, or we shouldn’t say vibe code. They make their own tools, and it’s cheaper than ever before because you have AI that can help you.
AI is really good at creating simple web applications that can show you data, that can write to a database. It’s very simple. For the Nurture Boss use case, we wanted to remove all the friction of looking at data. What you see here is just some screenshots of what the application that they created looks like. It’s just, “Okay, they have the different channels, voice, email, text. They have the different threads, they hid the system prompt by default.” Little quality of life improvements. Then, they actually have this axial coding part here where you can see in red the count of different errors. They automated that part in a nice way and they created this within a few hours. It’s really hard to have a one size fits all thing for looking at your data. You don’t have to go here immediately, but something to think about is make it as easy as possible because, again, it’s the most powerful activity that you can engage in. It’s the highest ROI activity you can engage in. With AI, yeah, just remove all the friction.
Lenny Rachitsky: That’s amazing. Again, I think that ROI piece is so important. We haven’t even touched on this enough. The goal here is to make your product better, which will make your business more successful. This isn’t just a little exercise to catch bugs and things like that. This is the way to make AI products better because the experience is how users interact with your AI.
Hamel Husain: Absolutely. If any, we teach our students, “Hey, when you’re doing these evals, if you see something that’s wrong, just go fix it.” The whole point is not to have evals, a beautiful eval suite, where you can point at it, edit it and say, Oh, look at my evals.” No, just fix your application, make it better. If it’s obvious, do it. Totally agree with you.
Lenny Rachitsky: Amazing. A question I didn’t ask, but this is I think something people are thinking about. How long do you spend on this? How long does it usually take to do? The first time
Shreya Shankar: I can answer for myself for applications that I work with. Usually, I’ll spend three to four days really working with whoever to do initial rounds of error analysis. A lot of labeling, feel like we’re in a good place to create the spreadsheet that Hamel had and everyone’s on-board and convinced, and even a few LLM judge evaluators. But this is one-time cost. Once I figured out how to integrate that in unit tests, or I have a script that automatically runs it on samples and I’ll create a Cron Job to just do this every week. I would say it’s like, I don’t know, I find myself probably spending more time looking at data because I’m just data hungry like that. I’m so curious.
I’m like, I’ve gained so much from this process and it’s put me above and beyond in any of my collaborations with folks, so I want to keep doing it, but I don’t have to. I would say maybe 30 minutes a week after that.
Lenny Rachitsky: It’s a week essentially, a week essentially upfront, and then 30 minutes to keep improving on adding to your suite?
Shreya Shankar: Yeah, it’s really not that much time. I think people just get overwhelmed by how much time they spend up front and then thinking that they have to keep doing this all the time.
Lenny Rachitsky: Amazing. Is there anything else that you wanted to share or leave listeners with? Anything else you wanted to double down as a point before we get to a very exciting lightning round?
Hamel Husain: I would say this process is a lot of fun, actually. It’s like, okay, you’re looking at data. Oh, it sounds like you’re annotating things. Okay. Actually, I was just looking at a client’s data yesterday, the same exact process. It’s a application that sends emails, recruiting emails to try to get candidates to apply for a job. We decided to start looking at traces. We jumped right into it. “Hey, let’s look at your traces.” We looked at a trace, the first thing I saw was this email that is worded, “Given your background, blah, blah, blah, blah, blah.” I asked the person right away, and this is where putting your product hat on and just being critical, and this is where the fun part is.
I said, “You know what? I hate this email. Do you like the email, given your background?” When I receive a message given your background, comma, I just delete that. I’m like, “What is this, given your background with machine learning and blah blah?” I’m like, “This is a generic thing.” I asked the person like, “Hey, can we do better than this? This sounds like generic recruiting.” They’re like, “Oh, yeah, maybe.” Because they were proud of it, they’re like, “The AI is doing the right thing, it’s sending this email with the right information, with the right link, with the right name, everything.” That’s where the fun part is, is put your product hat on and get into, is this really good?
Lenny Rachitsky: Something I want to make sure we cover before we get to a very exciting lightning round is, this is just scratching the surface of all the things you need to know to do this well. I think this is the best primer I’ve ever seen on how to do this well.
Shreya Shankar: Nice.
Lenny Rachitsky: But I think we did it. But you guys teach a course that goes much, much deeper for people that really want to get good at this and take this really seriously. Share what else you teach in the course that we didn’t cover, and what else you get as a student being part of the course you teach at Maven.
Shreya Shankar: Yeah, I can talk about the syllabus a little bit, and then Hamel can talk about all the perks. We go through a lifecycle of error analysis, then automated evaluators, then how to improve your application, how do you create that flywheel for yourself? We also have a few special topics that we find pretty much no one has ever heard of or taught before, which is exciting. One is, how do you build your own interfaces for error analysis? We go through actual interfaces that we’ve built and we also live code them on the spot for new data. We show how we use Claude code cursor, whatever we’re feeling in the moment that day to build these interfaces.
We also talk about broadly cost-optimization as well. A couple of people that I’ve worked with, they get to a point where their evals are very good, their product is very good, but it’s all very expensive because they’re using state-of-the-art models. How can we replace certain uses of the most expensive GPT-5, with 5-nano, 4-mini whatnot and save a lot of money, but still maintain the same quality? We also give some tips for that. Hamel, you’re on. We also have many perks.
Lenny Rachitsky: Yeah. Talk about the perks.
Hamel Husain: Okay, the perks. My favorite perk is there’s 160 page book that’s meticulously written, that we’ve created, that walks through the entire process in detail of how to do evals that supplement the course. You don’t have to sit there and take all these notes. We’ve done all the hard work for you and we have documented it in detail and organize things. That is really useful. Another really interesting thing, and something that I got the idea from you, Lenny, is, okay, this is an AI course. Education shouldn’t be this thing where you are only watching lectures and doing homework assignments. Students should have access to an AI that also helps them. What we have done is we’ve, just like there’s the LennyBot that you have.
Lenny Rachitsky: Dot com.
Hamel Husain: Yeah, lennybot.com, we have made the same thing with the same software that you’re using, and we have put everything we’ve ever said about evals into that. Every single lesson, every office hours, every Discord chat, any blogs, papers, anything that we’ve ever said publicly and within our course, we’ve put it in there. We’ve tested it with a bunch of students and they’ve said it’s helpful. We’re giving all students 10 months free unlimited access to that alongside the course.
Lenny Rachitsky: Amazing. Then, you’ll charge for that later down the road?
Hamel Husain: I have no idea. I just take one month at a time. I don’t know where we’re going with that.
Lenny Rachitsky: Eight months and then we’ll have to figure it out. I was thinking this whole interview should have just been our bots talking to each other.
Shreya Shankar: That’s amazing. I would watch that, only for 10 minutes then I don’t know what they’re talking about.
Lenny Rachitsky: Yeah, maybe 30 seconds. Do you guys train it on the voice mode, by the way? That’s my favorite feature of Delphi’s product. If not, you should do that.
Hamel Husain: Oh, I think, I can’t remember, I should look at it.
Lenny Rachitsky: You definitely should. Now that we have this podcast episode, you could use this content to train it. It’s 11Labs powered. It’s so good. Okay, so how do they get to… I guess that’s okay. They get to that once they become, enter your course.
Shreya Shankar: Yeah, sign up for the course and then you’ll get a bunch of emails. Everything will be clear, hopefully.
Lenny Rachitsky: Amazing. Okay.
Shreya Shankar: We also have a Discord of all the students who have ever taken the class. That Discord is so active. I can’t go on vacation without getting notified on the plane.
Lenny Rachitsky: Bittersweet, bittersweet. Incredible. Okay. With that, we’ve reached our very exciting lightning round. I’ve got five questions for you. Are you ready?
Shreya Shankar: Yes. Let’s go.
Lenny Rachitsky: Let’s do it. Okay. I’m going to bounce between you two. Share something if you want. You can pass if you want. First question, Shreya, what are two or three books that you find yourself recommending most to other people?
Shreya Shankar: I like to recommend a fiction book because life is about more than evals. Recently, I read Pachinko by Min Jin Lee. A really great book. Then, I also am currently reading Apple in China, which the name of the author is slipping my mind, but this is more of an exposition, written by a journalist on how Apple did a lot of manufacturing processes in Asia over the last couple, several decades. Very eye-opening.
Lenny Rachitsky: Amazing. Hamel.
Hamel Husain: Yeah, I have them right here. I’m a nerd. Okay, so I’m not as cool as Shreya is. I actually have textbooks, which are my favorite. This one is a very classic one, Machine Learning by Mitchell. Now, it’s theoretical, but the thing I like about it is it really drives home the fact that Occam’s razor is prevalent not only in science, but also in machine learning and AI. A lot of times the simplest, and also engineering, so a lot of times the simpler approach generalizes better. That’s the thing I internalize deeply from that book. I also really like this one. Another textbook. I told you I’m a nerd. This is also a very old one, and this is Norvig algorithms. I really like it because it’s just human ingenuity and it’s lots of clever useful things in computing.
Shreya Shankar: They’re down the street, him and Berkeley.
Lenny Rachitsky: The people that did that research?
Shreya Shankar: Yeah, textbook authors.
Lenny Rachitsky: Super cool. Oh, man, nerds, I love it. Okay, next question. Favorite recent movie or TV show? I’ll jump to Hamel first.
Hamel Husain: Okay, so I’m a dad of two parents. I have two parents. Sorry, two kids. Yeah, I’m a dad of two kids, and I don’t really get the time to watch any TV or movies, so I watch whatever my kids are watching. I’ve watched Frozen three times in the last week.
Lenny Rachitsky: Only three? Oh, okay. In the last week. Okay.
Hamel Husain: That’s my life.
Lenny Rachitsky: Great, Hamel. Frozen. I love it. Okay, Shreya.
Shreya Shankar: Yeah, I don’t have kids, so I can give all these amazing answers. Actually, so my husband and I have been watching The Wire recently. We never actually saw it growing up, so we started watching it and it’s great.
Lenny Rachitsky: I feel like everyone goes through that. Eventually in their life they decide, I will watch The Wire.
Shreya Shankar: I know, so we are in that right now.
Lenny Rachitsky: It’s like a year of your life. It’s great. It’s such a great show. Oh, man. But it’s so many episodes and everyone’s an hour long.
Shreya Shankar: I know. I know.
Lenny Rachitsky: It’s such a commitment.
Shreya Shankar: We get through two or three a week, so we’re very slow.
Lenny Rachitsky: Worth it. Okay, next question. Do you have a favorite product you’ve recently discovered that you really love? We’ll start with Shreya.
Shreya Shankar: Yeah. I really like using Cursor, honestly. Now, Claude Code. I’ll say why. I’m a researcher more so than anything else. I write papers, I write code, I build systems, everything, and I find that a tool… I’m so bullish on AI assisted coding because I have to wear a lot of hats all the time. Now, I can be more ambitious with the things that I build and write papers about, so I’m super excited about those. Cursor was my entry point into this, but I’m starting to find myself always trying to keep up with all these AI assisted coding tools.
Lenny Rachitsky: Hamel?
Hamel Husain: Yeah, I really like Claude Code and I like it because I feel like the UX is outstanding. There’s a lot of love that went into that. It’s just really impressive as a terminal application that is that nice.
Lenny Rachitsky: Ironic that you two both love Claude Code when it’s just built on vibes.
Shreya Shankar: I think it’s false. It’s not just built on vibes.
Lenny Rachitsky: There we go. Okay, two more questions. Hamel, do you have a favorite life motto that you find yourself using in coming back to in work or in life?
Hamel Husain: Keep learning in. Think like a beginner.
Lenny Rachitsky: Beautiful. Shreya?
Shreya Shankar: I like that. For me, it’s to always try to think about the other side’s argument. I find myself sometimes just encountering arguments on the internet, like this race to eval debates and really think, “Okay, put myself in their shoes. There’s probably a generous take, generous interpretation.” I think we’re all much stronger together than if we start picking fights. My vision for evals is not that Hamel and I become billionaires. It is that everyone can build AI products, and we’re all on the same page
Lenny Rachitsky: Slash everyone becomes billionaires.
Shreya Shankar: Yes.
Lenny Rachitsky: Amazing. Final question. When I have two guests on, I always like to ask this question and I’ll start with Hamel. What’s something about Shreya that you like most? What do you like most about Shreya? I’m going to ask her the same question in reverse.
Hamel Husain: Yeah. Shreya is one of the wisest people that I know, especially for being so young relative to me. I feel like she’s much wiser than I am, honestly, seriously. She’s very grounded and has a very even perspective on things. I’m just really impressed by that all the time.
Lenny Rachitsky: Shreya?
Shreya Shankar: Yeah. My favorite thing about Hamel is his energy. I don’t know anybody who consistently maintains momentum and energy like Hamel does. I often think that I would start carrying much less about evals, if not for Hamel. Everyone needs a Hamel in their life, for sure.
Lenny Rachitsky: Well, we all have a Hamel in our life now. This was incredible. This was everything I’d hoped it’d be. I feel like this is the most interesting in-depth consumable primer on evals that I’ve ever seen. I’m really thankful you two made time for this. Two final questions. Where can folks find you? Where can they find the course and how can listeners be useful to you? I’ll start with Shreya.
Shreya Shankar: Yeah, you can reach me via email. It’s on my website. If you Google my name, that is the easiest way to get to my website. You can find the course if you Google AI Evals for engineers and product managers, or just AI Evals course, you’ll find it. We’ll send some links hopefully after this, so it’s easy. How to be helpful? Two things always for me. One is ask me questions when you have them. I’ll try to get to the respond as soon as I can. The other one is tell us your successes. One of the things that keeps us going is somebody tells us what they implemented or what they did, a real case study. Hamel and I gets so excited from these and it really keeps us going, so please share.
Hamel Husain: Yeah, it’s pretty easy to find me. My website is Hamel.dev. I’ll give you the link. You can find me on social media, LinkedIn, Twitter. The thing that’s most helpful is to echo what Shreya said, we would be delighted if we are not the only people teaching evals. We would love other people teach evals. Any kind of blog posts, writing, especially that as you go through this and learn this that you want to share, we would be delighted to help re-share that or amplify that.
Lenny Rachitsky: Amazing. Very generous. Thank you two, so much for being here. I really appreciate it, and you guys have a lot going on, so thank you.
Shreya Shankar: Thanks, Lenny, for having us and for all the compliments.
Lenny Rachitsky: My pleasure. Bye everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at Lennyspodcast.com. See you in the next episode.
Reformatted by reformat_english_direct.py
为什么 AI evals 是产品构建者最热门的新技能 | Hamel Husain & Shreya Shankar
文字记录
Lenny Rachitsky: 要打造优秀的 AI 产品,你必须擅长构建 AI评估(evals)。这是你能从事的投资回报率最高的活动。
Hamel Husain: 这个过程非常有趣。每个做过这件事的人都会立刻上瘾。在构建 AI 应用时,你会学到很多东西。
Lenny Rachitsky: 这件事很酷的地方在于,你不需要反复做很多次。对大多数产品来说,你只需要完成一次这个过程,然后在此基础上构建就行了。
Shreya Shankar: 目标不是把 evals 做得完美,而是切实改进你的产品。
Lenny Rachitsky: 我没想到 evals 周围有这么多争议和风波。很多人持有非常强烈的观点。
Shreya Shankar: 人们过去在 evals 上吃过亏。他们做过效果不好的 evals,于是不再信任它,然后就变成了”哦,我反对 evals”。
Lenny Rachitsky: 关于 evals,人们最常见的几个误区是什么?
Hamel Husain: 最主要的一个是,“我们生活在 AI 时代,难道不能让 AI 自己来做评估吗?“但这行不通。
Lenny Rachitsky: 你在文章中用过一个我很喜欢的词——“仁慈独裁者”(benevolent dictator)。
Hamel Husain: 在做开放式编码时,很多团队会陷入委员会集体决策的泥潭。在很多情况下,这完全没有必要。你不想把这个流程搞得太昂贵以至于无法执行。你可以指定一个你信任其品味的人,应该是具有领域专业知识的人。通常,这个人选是产品经理。
嘉宾介绍与 evals 的崛起
Lenny Rachitsky: 今天我的嘉宾是 Hamel Husain 和 Shreya Shankar。过去一年里,这个播客上最热门的话题之一就是 evals 的兴起。Anthropic 和 OpenAI 的首席产品官都表示,evals 正在成为产品构建者最重要的新技能。从那以后,这成为了我邀请的众多顶级 AI 构建者口中反复出现的主题。两年前,我从未听说过 evals 这个词。现在它不断出现。上一次出现产品构建者必须掌握才能成功的新技能,是什么时候的事了?
Hamel 和 Shreya 在将 evals 从一个冷门、神秘的课题转变为 AI 产品构建者最必备的技能之一方面,发挥了重要作用。他们在 Maven 上教授最权威的 evals 在线课程,恰好也是 Maven 上排名第一的课程。他们已经教过来自 500 家公司的超过 2000 名产品经理和工程师,其中包括 OpenAI 和 Anthropic 的大量团队成员,以及其他所有主要 AI 实验室的人。
在这次对话中,我们会大量展示实操而非空谈。我们将逐步演示开发有效 eval 的流程,解释 evals 到底是什么、长什么样,讨论关于 evals 的主要误区,给出你可以立即开始为你的产品构建 evals 的最初几个步骤,并分享 Hamel 和 Shreya 过去几年积累的大量最佳实践。这期节目是关于 evals 世界最深入同时也最易懂的入门指南。说实话,它甚至让我对写 evals 产生了兴趣,尽管我根本没有什么需要写 evals 的东西。我想你看完之后也会有同样的感受。
如果这次对话让你兴奋起来,一定要去看看 Hamel 和 Shreya 在 Maven 上的课程。我们会在节目备注中附上链接。购买课程时使用优惠码 LENNYSLIST,可以享受 35% 的折扣。下面,有请 Hamel Husain 和 Shreya Shankar。
[广告已跳过]
Lenny Rachitsky: Hamel、Shreya,非常感谢你们来到这里,欢迎来到播客。
Hamel Husain: 谢谢你的邀请。
Shreya Shankar: 太兴奋了。
到底什么是 evals
Lenny Rachitsky: 我比你们更兴奋。好的,几年前我从未听说过 evals 这个词。现在它基本上是我播客上最热门的话题之一——要打造优秀的 AI 产品,你必须擅长构建 evals。而且事实证明,世界上一些增长最快的公司基本上就是在为 AI 实验室构建、销售和创建 evals。我之前刚请了 Mercor 的 CEO 来上播客。所以这里确实正在发生一件大事。我想通过这次对话帮助人们深入理解这个领域,但让我们从基础开始。到底什么是 evals?对于完全不知道我们在说什么的人,给我们一个快速的理解,什么是 eval?先请 Hamel 来说。
Hamel Husain: 好的。Evals 是一种系统化衡量和改进 AI 应用的方法,它完全不需要令人畏惧或遥不可及。它的核心其实就是对你的 LLM 应用做数据分析,以一种系统化的方式审视这些数据,必要时围绕各种事物创建指标,从而衡量正在发生的情况,然后据此迭代、做实验并改进。
Lenny Rachitsky: 这是一个非常好的宏观理解方式。如果再深入一层,给人们一个更具体的想象和可视化方式——如果你能展示一个例子就更好了——关于 eval 是什么,还有没有更深入的理解方式?
Hamel Husain: 假设你有一个房地产助手应用,它没有按你想要的方式工作——它没有按你的要求给客户写邮件,或者没有调用正确的工具,或者出现了各种错误。在 evals 出现之前,你只能靠猜。你可能会修改一个提示词,然后祈祷不会因此破坏其他东西,你可能依赖”感觉检查”(vibe checks),这完全没问题。
evals 不仅仅是测试
Hamel Husain: 感觉检查是好的,初期你应该做感觉检查,但它很快就会变得非常难以管理,因为随着应用的增长,仅凭感觉检查是很困难的。你会感到无所适从。而 evals 帮助你创建可以用以衡量应用表现的指标,在某种程度上给你一种自信改进应用的方式——你拥有一个可以据此迭代的反馈信号。
Lenny Rachitsky: 为了让大家更直观地理解,我们继续想象这个房地产助手的例子——也许它帮客户预约看房或者安排开放日参观。它的场景是这个助手在和客户对话,回答问题,引导他们做各种事情。作为这个助手的构建者,你怎么知道它给出的建议好不好?回答是否正确?它有没有在说完全错误的东西?
所以 evals 的核心理念,本质上就是构建一组测试,告诉你这个助手有多经常做出你不希望它做的事情。而”错误”的定义可以有很多种方式。可能是凭空编造信息,可能是以一种非常奇怪的方式回答。我对 evals 的理解——告诉我这么说对不对——简单来说就像是代码的单元测试。你在笑。你是不是在想,“不,你这个白痴。”
Shreya Shankar: 不,我不是那个意思。
Lenny Rachitsky: 好好好,那你说说,这个比喻你觉得怎么样?
Shreya Shankar: 好的。我同意你一开始说的,我们给出了一个非常宽泛的定义。Evals 是衡量应用质量的一个很大的光谱。单元测试确实是其中一种方式。也许你的 AI 助手有一些不可妥协的功能要求,单元测试可以检查这些。但与此同时,因为这些 AI 助手执行的任务非常开放,你还需要衡量它们在模糊或含混不清的事情上表现如何——比如应对新型用户请求,或者判断是否出现了新的数据分布,比如突然有新的用户群体开始使用你的房地产助手,而这些人你甚至没想到会使用你的产品。然后你会想,“哦,我需要用一种不同的方式来适应这个新用户群体。”
所以 evals 也可以是一种定期查看数据以发现这些新用户群的方式。Evals 也可以是你想要随时间追踪的指标,比如追踪用户说”是的,点赞,我喜欢这个回答”。这些非常基础的东西不一定与 AI 相关,但可以回到改进产品的飞轮中。所以总体来说,单元测试只是这个巨大拼图里很小的一部分。
Lenny Rachitsky: 太好了。你们实际上还带来了一个 eval 的具体例子,就是想给我们展示一下我们到底在说什么。我们一直在讲这些大概念。那不如我们拉出一个例子,让大家看看,“这就是一个 eval。“
真实案例:Nurture Boss
Hamel Husain: 好的,让我先稍微铺垫一下背景。呼应 Shreya 刚才说的,非常重要的是我们不要把 evals 仅仅等同于测试。很多人会掉进一个常见的陷阱——他们直接跳到写测试,“让我来写一些测试”,但通常这不是你应该做的。你应该从某种数据分析开始,以此为依据来确定你到底应该测试什么。这和传统软件工程有点不同,在传统软件工程中你对系统行为有更多的预期。而面对 LLM,表面积大得多,随机性很强,所以这里的情况有所不同。
今天我要给你们看的例子,正好是一个房地产的例子,不过是不太一样的那种。它来自一家叫 Nurture Boss 的公司。我可以分享一下屏幕,展示一下他们的网站,帮助大家理解这个用例,让我来共享屏幕。这是一家我合作过的公司,叫 Nurture Boss,它是一个面向物业管理人员的 AI 助手,帮助管理公寓,处理各种事务,比如潜在客户线索、客户服务、预约安排等等。物业管理人员日常可能做的各种运营工作,它都能帮忙。你可以看到他们做的事情。这是一个非常好的例子,因为它包含了现代 AI 应用的很多复杂性。
它有很多不同的交互渠道,比如聊天、短信、语音,同时还有工具调用——大量的工具调用,用于预约安排、获取可用信息等等。还有 RAG 检索,获取关于客户和物业的信息之类的。所以作为一个 AI 应用来说,它是相当完整的。他们非常慷慨地允许我使用他们的数据作为教学案例。我们已经做了匿名化处理,但今天我要带大家走一遍的是,好的,我们如何开始为 Nurture Boss 构建 evals。我们为什么需要做这件事?
让我们看看最开始的阶段,我们称之为错误分析(error analysis),也就是去看看他们应用的数据,首先搞清楚出了什么问题。我来跳到下一步,打开一个可观测性工具(observability tool)。这里你可以用任何你喜欢的工具。我只是碰巧把数据加载到了一个叫 Braintrust 的工具里,但你可以加载到任何工具中。在我们和你一起写的那篇博文里,我们没有偏好某个特定工具。我们有同样的例子但用的是 Phoenix Arize,我想 Aman 在他自己的博文里也用了 Phoenix Arize。还有 LangSmith。这些就是你可以使用的各种不同工具。
现在屏幕上显示的是应用的日志,让我给你们看看它是什么样的。让我把它全屏——这是某个客户与 Nurture Boss 应用的一次具体交互,它是一份详细记录了所有发生事情的日志。这叫做 trace,就是一系列事件日志的工程术语。Trace 这个概念已经存在了很长时间,但在 AI 应用领域它尤为重要。
我们在这里能看到 AI 完成其工作所需的所有不同组件、部分和信息,所有这些都被记录下来了,我们现在看到的就是这样一个视图。你看这里有一个系统提示词(system prompt)。系统提示词写的是,“你是一名 AI 助手,在 Retreat at Acme Apartments 担任租赁团队成员。” 记住我说过这是匿名化的,所以名字是 Acme Apartments。“你的主要职责是回复来自现有住户和潜在住户的短信。你的目标是提供准确、有用的信息,” 等等等等。然后还有很多关于我们希望它如何行为的详细指引。
Lenny Rachitsky: 顺便问一下,这是这家公司实际使用的系统提示词吗?
Hamel Husain: 是的,确实是。
Lenny Rachitsky: 太棒了,太酷了。
Hamel Husain: 这是真实的系统提示词。
Lenny Rachitsky: 太厉害了,因为你很少能看到一个真实的公司产品的系统提示词。那通常被视为他们的核心机密,所以这本身就非常酷。
查看真实的对话 trace
Hamel Husain: 对,真的很酷。你可以看到各种不同的功能用例,比如房源预约、处理申请、针对不同角色的沟通指引等等。你可以看到用户直接进来问:“你们有带书房的一居室吗?我在虚拟看房里看到的。“然后你可以看到 LLM 调用了一些工具——它调用了获取个人信息工具,把那个人的信息拉了出来。接着它又查询了社区的可租房源——也就是去查那个公寓楼的数据库,看有哪些房源可用。
然后 AI 回复说:“我们有几套一居室公寓可租,但没有专门标注带书房的。以下是一些选项。“接着用户说:“有一套带书房的了能通知我吗?“AI 回复说:“我目前没有关于一居室公寓带书房的具体房源信息。“用户说:“谢谢。“AI 说:“不客气,如果还有其他问题,随时联系我们。”
这是一个 trace 的例子,我们正在看的是其中一条具体的数据点。在做 LLM 应用的数据分析时,一件非常重要的事情就是——看数据。你可能会想:“这些日志量太大了,乱糟糟的,各种信息混杂在一起。到底该怎么看这些数据?难道要被数据淹没吗?怎么分析?“
错误分析(error analysis)方法
其实有一种完全可控的方法,而且这不是我们发明的,它在机器学习和数据科学领域已经存在很长时间了,叫做错误分析(error analysis)。具体怎么做呢?掌控这类数据的第一步,就是写笔记。你需要戴上产品经理的帽子——这也是为什么我们要和你聊这个话题,因为产品人员必须参与其中,亲自做这件事。通常开发人员并不适合做这件事,尤其是当这个应用不是编程类应用的时候。
Lenny Rachitsky: 我想复述一下,确认我理解了你的意思——你之所以这么说,是因为这就是你的产品的用户体验。用户和这个 agent 对话,本质上就是整个产品,所以产品经理深度参与这件事是完全合理的。
Hamel Husain: 对。那我们来审视一下这段对话。用户询问了房源可用性,AI 说:“哦,我们没有那种房型,祝您愉快。“对于一个做线索管理的产品来说,这算好吗?你觉得这是我们期望的结果吗?
Lenny Rachitsky: 不太理想。
Hamel Husain: 对,不理想,我很高兴你这么说。很多人会说:“哦,挺好的,AI 做得没错。它查了,说了没有可用的,确实没有嘛。“但如果你戴上产品经理的帽子,你就知道这样是不对的。所以你要做的就是在这里快速写一条笔记。你可能会进入这个界面,写一条备注——每个可观测性工具(observability tool)都支持写备注的功能。你不需要去判断到底是哪里出了问题——在这个案例中,它的行为确实不太对——你只需要快速写一句:“应该转交给人工处理。”
Lenny Rachitsky: 看着这个过程,就像你提到的这一点,后面还会进一步展开说明——你正在做的这件事,感觉非常手动、很难规模化,但正如你所说,这只是流程中的一步,背后有一套体系。这只是第一步。
Hamel Husain: 对,而且你不需要对所有数据都这么做。你抽样一部分数据看一看,做下来你会惊讶于自己能学到多少东西。每个这样做的人都会立刻上瘾,他们说:“这是构建 AI 应用时你能做的最有价值的事情。“你真的能学到很多,然后你会想:“嗯,这不是我想要它运行的方式,好吧。“这只是其中一个例子。
所以你写下这条笔记,然后我们可以看下一条 trace。这就是下一条 trace,我在键盘上按了个快捷键切过来的。
Lenny Rachitsky: 这些工具让你可以很方便地批量浏览并快速添加备注。
Hamel Husain: 对。这是另一条。类似的系统提示词(system prompt),我们不需要再过一遍了。直接跳到用户的问题。用户说:“我给你发了一整天短信了。“挺有意思的吧?用户说:“请……”好吧,这个案例其实是一个应用程序的错误——这是一个短信应用,准确地说是客户通过短信渠道进行沟通的,你会收到非常混乱的内容。你可以看到这里的对话基本上读不通,消息被截断了,比如”与此同时,“然后就断了,系统也不知道该怎么回复。因为大家发短信的方式就是那样,写很短的短语,一句话拆成四五条消息来发。
Lenny Rachitsky: 那遇到这种情况你怎么办?
Hamel Husain: 这是一种不同类型的错误。这更像是”我们没有正确处理这种交互”,是一个技术层面的问题,而不是”AI 没有按我们的要求去做”的问题。所以这个我们也要记下来。
Lenny Rachitsky: 能发现这种问题也真的很酷。
Hamel Husain: 对。
Lenny Rachitsky: 能在这里捕捉到这种问题太厉害了。否则你根本不知道这种情况在发生。
写笔记的原则:只记第一个错误
Hamel Husain: 对,你可能完全不知道发生了这种事。所以你就写一条备注:“由于短信渠道导致对话流程很 janky(混乱)。”
Lenny Rachitsky: 我喜欢你用了 janky 这个词。这说明这个阶段可以非常随意。
Hamel Husain: 对,就是要轻松随意,别想太多。这是有方法可循的。大家经常问的一个问题是:具体怎么做?你是要试图找出这条 trace 里所有的问题吗?到底该写什么?答案是——只写你看到的第一个问题,也就是最上游的那个错误。不要管其他所有错误,只抓住你看到的第一个问题,然后停下来,继续看下一条。你可以做得很快——前两三条可能会很痛苦,但熟练之后就能飞速处理。
幻觉问题的例子
再看这一条,系统提示词(system prompt)我们也跳过。用户问:“你好,我在找两到三居室,一到两个卫生间的,你们提供虚拟看房吗?”
然后调用了一堆工具,AI 回复说:“Hi Sarah,目前我们有一套三居室、两个半卫生间的公寓,租金 $2,175。不幸的是,目前没有两居室的选择。我们确实提供虚拟看房,您可以预约看房,“等等等等。但问题是,根本没有虚拟看房这个功能。
Lenny Rachitsky: 嗯,好。
Hamel Husain: 所以这是在编造一个不存在的东西——产生了幻觉。这时候你就需要带入你作为工程师或产品人员的上下文判断,说:“这个很奇怪。我们不应该告诉用户有虚拟看房,因为我们根本不提供这个服务。“
能用 LLM 自动做错误分析吗?
Hamel Husain: 所以你会说,“好的,推荐了虚拟看房”,然后直接写下这条笔记。你可以看到,我们发现的问题类型是多种多样的,而且我们实际上在很短的时间内就对你的应用了解了很多。
Shreya Shankar: 我们在这个阶段经常被问到的一个问题是,“好吧,我理解这是怎么回事了。能不能让 LLM 来替我做这个过程?”
Lenny Rachitsky: 嗯,好问题。
Shreya Shankar: 我很喜欢 Hamel 刚才举的最新例子,因为我们通常发现,当我们尝试让 LLM 来做这种错误分析时,它只会说这个 trace 看起来不错,因为它没有理解某个东西是否有产品层面的问题所需的上下文。比如刚才那个编造预约看房的幻觉,对吧?我敢保证,我愿意为此打赌——如果我把它放进 ChatGPT,问”有没有错误?“它会说,“没有,做得很好。”
但 Hamel 有上下文,他知道,“哦,我们其实没有虚拟看房这个功能。“所以我认为,在这种情况下,确保你自己亲手做这件事是非常重要的。我们稍后可以再谈谈在过程中何时使用 LLM,但这里的头号陷阱就是人们说,“让我用 LLM 把这个过程自动化。”
Lenny Rachitsky: 你觉得我们会不会走到这样一个阶段,一个 agent 能做这件事,具备那样的上下文?
Shreya Shankar: 哦,不会。不不不。抱歉。错误分析中确实有一些部分适合用 LLM 来做,我们稍后可以在播客中讨论。但现在这个自由形式、记笔记的阶段,不是 LLM 该介入的地方。
Lenny Rachitsky: 明白了。你把这个步骤叫做开放式编码(open coding),对吗?
Shreya Shankar: 没错,完全正确。
仁慈独裁者(Benevolent Dictator)
Lenny Rachitsky: 好的。你们在文章中用到的另一个我很喜欢的术语,也和这个步骤相关,就是”仁慈独裁者”这个概念。也许可以谈谈这是什么,Shreya 你来讲讲?
Shreya Shankar: 好的,这个词其实是 Hamel 发明的。
Lenny Rachitsky: 那还是 Hamel 来讲吧。
Hamel Husain: 没问题。而且我们实际上会在这个例子中展示 LLM 自动化,因为我们会用这个例子一路走到底。
Lenny Rachitsky: 太好了。
Hamel Husain: 所以”仁慈独裁者”只是一个比较醒目的说法,表达的是这样一个事实:在做这种开放式编码时,很多团队会被委员会式的工作方式拖慢。在很多情况下,这完全没有必要。人们会非常不安,觉得”我们需要所有人都参与进来,我们需要所有人的认可”之类的。你需要穿过这些噪音。在很多组织中,如果你深入观察,尤其是中小型公司,你可以指定一个你信任其判断力的人。你可以让很少的人来做这件事,通常一个人就够了。让这个过程可执行是非常重要的。你不能把这个过程搞得太昂贵以至于根本做不了,那样你就亏了。
这就是”仁慈独裁者”背后的理念——“嘿,你需要在尽可能多的维度上简化这件事。“我们稍后还会谈到另一个相关的事情:当你要构建一个 LLM 作为评判者时,你需要一个二元的评分。你不想去想”这是 1 分、2 分、3 分、4 分还是 5 分?“给它打个分。你做不到的,那会拖慢进度。
Lenny Rachitsky: 为了确保”仁慈独裁者”这个概念说得很清楚——基本上,这个人就是负责做笔记的,而且理想情况下他们应该是这个领域的专家。如果是法律相关的,可能是一个法务人员来负责;也可以是产品经理。给我们一些关于这个人应该是什么角色的建议?
Hamel Husain: 对,应该是具有领域专业知识的人。在这个案例中,就是了解公寓租赁业务、有上下文来判断这样做是否合理的人。就像你说的,始终是一个领域专家。如果是法律领域,就是法律专业人士。如果是心理健康领域,就是心理健康专家,可能是精神科医生或其他相关人员。
Lenny Rachitsky: 好的。
Hamel Husain: 不过很多时候,这个人是产品经理。
Lenny Rachitsky: 好的。所以建议就是:选定这个人。可能感觉不太公平——他一个人说了算,他是独裁者,但他是仁慈的。没关系的。
Hamel Husain: 对,没关系的。不需要追求完美。你只是想取得进展,快速获得信号,这样你就知道接下来该做什么,因为如果不小心的话,这件事的成本可以无限膨胀。
抽样多少条 trace 才够?
Lenny Rachitsky: 嗯,好的。我们回到你的例子。
Hamel Husain: 好的,没问题。这是另一个例子,有人说,“好吧,你们有什么优惠吗?“然后助手或 AI 回答说,“嘿,我们有 5% 的军人折扣。“用户接着回复,话题转换了,“能告诉我有多少层吗?你们有没有一居室?或者一楼有没有一居室?“AI 回答说,“好的,我们有几套一居室公寓。“用户想确认,“那些有在一楼的吗?一居室多少钱?“同时这是一个现有住户,所以还在问,“我需要提交一个维修请求。”
你可以在这里看到现实世界的混乱,而助手直接调用了一个工具说转接电话,但它什么都没说。就是突然执行了转接电话,所以我觉得这相当 janky。就是——
Lenny Rachitsky: 又一个 jank。
Hamel Husain: 另一种 jank,不同类型的 jank。所以当你写开放式笔记时,你不想只写”jank”,因为我们想要的是理解——当我们之后回顾这些笔记时,我们想要理解发生了什么。
所以你只需要写,“未与用户确认通话转接。“不需要完美,只要对发生了什么有个大概的了解就行。
Lenny Rachitsky: 好的。
Hamel Husain: 好的。假设我们做了——Shreya 和我建议至少做 100 条。大家总是问,“到底要做多少条?“这没有神奇的数字。我们说 100 条,只是因为我们知道,一旦你开始做这件事,做了 20 条之后,你会自动发现它太有用了,你会继续做下去。
所以我们说 100 条只是为了在心理上帮你解绑,让它不那么令人望而生畏。就像,“别担心,你只需要做 100 条。“这其实有一个术语,所以正确答案是,“持续查看 trace,直到你觉得自己学不到新东西了。“也许 Shreya 可以讲讲——
Shreya Shankar: 对,其实在数据分析和质性分析中有一个术语,叫做理论饱和(theoretical saturation)。它的意思是,当你做所有这些查看数据的过程时,什么时候停下来?就是当你达到了理论上的饱和——你不再发现新类型的笔记、新类型的概念,或者任何会在你流程的下一个阶段产生实质性改变的东西。
从开放式编码到轴向编码
Hamel Husain: 这种直觉需要一点时间来培养,所以通常人们并不真的知道自己什么时候达到了理论饱和。这完全没问题。当你做了两三个例子或两三轮之后,你就会逐渐培养起这种直觉。很多人会意识到,“哦,好吧,我只需要做 40 条,只需要做 60 条。其实我只需要做 15 条。“因情况而异,也取决于你对错误分析(error analysis)的熟练程度,这是肯定的。
Lenny Rachitsky: 你说的”你会想要做很多条”,我猜是因为你会觉得,“哦,我发现了这么多问题,我得看看还有什么别的。”
Shreya Shankar: 没错。
Lenny Rachitsky: 是这样吗?
Shreya Shankar: 而且我保证,到了某个点之后,你就不会再发现新类型的问题了。
Lenny Rachitsky: 好的。假设你做了 100 条,下一步是什么?
Hamel Husain: 好,你做了 100 条,现在你有了所有这些笔记。这时候你可以开始用 AI 来帮你了。正如我们讨论过的,亲自查看数据这个环节很重要,你不想把这个部分过度自动化。
Lenny Rachitsky: 人类还是会有工作的。这是关键结论。太好了。
Hamel Husain: 是的。
Lenny Rachitsky: 就是审查 trace。至少目前还剩这一份工作。很好。
基础计数——最强大的分析技术
Hamel Husain: 对。好,你现在有了所有这些笔记。要把它变成有用的东西,你可以做基础计数。基础计数是数据科学中最强大的分析技术,因为它如此简单,而且在很多情况下被低估了,所以对人们来说非常容易上手。
所以你要做的第一件事就是拿着这些笔记,用 LLM 对它们进行分类,方法有很多种。就在录这期播客之前,我试了三种不同的编程代理或 AI 工具来给这些笔记分类。一种是,我把一个包含这些笔记的 CSV 上传到了一个 Cloud 项目里,直接从这个界面导出。方法有很多,但我展示的是最简单、最笨的方式,最基础的做法。
我把 CSV 丢进去,然后说”请分析以下 CSV 文件”。我告诉它有一个元数据字段包含笔记内容,但我用的是”open codes”(开放式编码)这个词,我说”我有不同的 open codes”,这是一个专业术语。LLM 知道什么是 open codes,也知道什么是 axial codes,因为这是一个存在了很久的概念,所以这些词能帮我快速表达我想要做的事。
Lenny Rachitsky: 太好了。prompt 的末尾是让它创建 axial codes?
Hamel Husain: 对。创建 axial codes,它的作用是——
Shreya Shankar: 也许值得讲讲什么是 axial codes,目的是什么?你面对的是一堆混乱的 open codes,你并没有 100 个不同的问题。实际上很多是重复的,只是因为你的措辞不同。而且你不应该在开放式编码的时候就去创建自己的失败分类体系。你只需要把问题记下来,然后再去整理,“好,最常见的失败模式是什么?”
所以 axial code 的目的基本上就是一个失败模式。它是一个标签或类别。我们的目标是得到这些失败模式的聚类,找出哪个最普遍,这样你就可以着手去解决那个问题。
Lenny Rachitsky: 这真的很有帮助。基本上就是把所有这些——
Shreya Shankar: 完全正确。
Lenny Rachitsky: 综合成类别和主题。非常酷。我们会把这个 prompt 放在节目说明里,这样大家就不用坐在那里截图然后自己手敲了。
Hamel Husain: 好主意。然后 Claude 分析了 CSV 文件,决定如何解析,等等这些细节我们不用操心,但它给出了一堆 axial codes。正如 Shreya 所说,axial codes 基本上就是类别。比如,能力限制、错误陈述、流程和协议违规、人工交接问题、沟通质量。它创建了这些类别。
我喜不喜欢所有这些类别呢?并不完全喜欢。我喜欢其中一些。这是一个不错的初步尝试。我可能会给它改改名,因为有些太笼统了。比如”能力限制”是什么意思?有点太宽泛了,不具有可操作性。我想让它更具可操作性,这样如果我确定它是个问题,我就知道该怎么处理它。不过这个我们稍后再讨论。你可以用任何工具来做这件事,这是最笨的方法,但有时候笨方法正是一个好的起步方式。
Lenny Rachitsky: 而这正是 LLM 真正擅长的事情——把大量信息拿来并综合提炼。
Shreya Shankar: 完全同意。帮我们综合提炼,让我们能够理解,对吧?注意,它并不是自动提出修复方案或做什么,那是我们的工作,但现在我们可以更轻松地梳理这堆混乱的 open codes 了。
灵活定制轴向编码的 prompt
还有一点很有趣,在这个生成 axial codes 的 prompt 中,你可以非常详细地指定。你可以说”我希望每个 axial code 实际上是一个可操作的失败模式”,也许 LLM 会理解并据此提出建议。或者你说”我希望你按照用户故事的阶段来对这些 open codes 分组”。这就是你可以发挥创意的地方,或者作为产品经理或工程师做最适合你的事情,这将帮助你后续进行改进。
Lenny Rachitsky: 所以并没有一个终极的 prompt,“这就是唯一正确的方法”?
Shreya Shankar: 完全没有。
Lenny Rachitsky: 你是说可以迭代,看什么适合自己?
Shreya Shankar: 完全正确。
Lenny Rachitsky: 有意思的是现有工具不做这件事,还是说它们尝试了但做得不好?
Shreya Shankar: 不,我觉得它们没做这件事。我们一直在到处呼吁,“拜托,拜托——”
Lenny Rachitsky: 哦,真的吗。
Shreya Shankar: “请做这件事吧。“不过我确实觉得这有点难。Hamel 和我教的 eval 课程中的经验之一是,很多人其实不知道这些,所以也许是人们不知道,也就不知道如何为此构建工具。希望我们能揭开这层神秘面纱。
Lenny Rachitsky: 再深入聊一下这一点,这不是所有人都知道或都在做的事情。这是你们两位基于在其他公司做数据分析和数据科学的经验发展出来的?
方法论的根基
Shreya Shankar: 我想说明的是,我们没有发明错误分析(error analysis)。我们其实不想发明新东西。那是不好的信号。如果有人带着一种全新的、不是建立在数百年理论和文献基础上的方法来找你,那你应该——我不知道,稍微警惕一点。
但我们努力做的是提炼出,“好,要理解 LLM 的错误分析,你需要哪些新的工具和技术?“然后我们创建了一套课程或结构化的方法来做这件事。所以这些都是针对 LLM 的,但 open coding、axial coding 这些术语是植根于社会科学的。
Lenny Rachitsky: 太棒了。好吧,你们做这件事有意思的地方在于,我也想找个地方去做这件事。我手头没有任何 AI 产品可以拿来练手,但就是觉得,“这太有意思了。“坐在那里找出所有遇到的问题,给它们分类,然后尝试修复。
Shreya Shankar: 我太喜欢这种想法了。
Lenny Rachitsky: Hamel 打开了一个视频。你这里展示了什么?
Hamel Husain: 对。我打开了一个视频,就是为了进一步印证 Shreya 的观点。我们并没有发明任何东西。你在屏幕上看到的是 Andrew Ng,世界上最著名的机器学习研究者之一,说实话,他教会了很多人机器学习。你可以看到这是一个八年前的视频,他在讲错误分析(error analysis)。这是一种用于分析随机系统的技术,已经被使用了很长时间了。这里用的就是同样的机器学习思想和原则,只是把它们搬到了这里,因为同样,这些也是随机系统。
实操演示:工具与流程
Lenny Rachitsky: 太好了。顺便说一下,我们正在邀请 Andrew 上播客,正在沟通,到时候——
Shreya Shankar: 太好了。
Lenny Rachitsky: ……那会非常有趣。另外,我很高兴我的播客节目今天刚发布就出现在了你们的推荐流里,而且在那个流里特别显眼,我真的很开心。
Hamel Husain: 非常好。是的。推荐算法相当不错。
Lenny Rachitsky: 没错。来吧,希望你们点进去。别搞砸我的算法。好的,酷。我们已经做了一些综合整理。我知道我们不会把整个流程走一遍——你们有一整套课程,需要好几天才能学完整个过程。关于如何开展这个流程,你们还想分享什么?
Hamel Husain: 好的。你可以用任何工具来做这件事,在 ChatGPT 里用完全相同的提示词也完全可以。你可以看到它生成了轴向编码(axial coding)。我个人很喜欢用 Julius AI,这是我最喜欢的工具之一。Julius 是一种使用 notebook 的第三方工具。我个人很喜欢 Jupyter notebook,所以这更偏数据科学的风格。不过现在很多产品经理也在学用 notebook,挺酷的。它就像一个好玩的游乐场,你可以在里面写代码、看数据。但我们不需要深入讲这个。只是想提一下,你可以用很多工具。AI 在做这件事上真的很擅长。
好,让我们进入有趣的部分。来了。现在我们有了这些轴向编码(axial codes)。我拿到这些开放式编码(open codes),以及我们在 Cloud 项目或 ChatGPT 中生成的轴向编码。我做的第一件事是先收集它们,然后看看:“这些轴向编码合理吗?“我会查看不同轴向编码和开放式编码之间的对应关系,然后做一个审视,说:“嗯,我喜欢这些编码吗?能不能做得更好?能不能更精细化?能不能让它们更具体?“与其使用笼统的编码,不如把它们做得非常具体和可操作。你们看我在这里提出的编码:行程调度、重新调度问题、人工转接或移交问题、输出格式错误、对话流程——我们之前在短信那条中看到了对话流程问题、做出的跟进承诺未兑现。
基本上,我现在能做的——你能做的——就是拿到这些轴向编码,把它们收集成一个列表。这是一个 Excel 公式,把这些编码收集成一个列表,现在我们就有了一个逗号分隔的编码列表。然后你可以简单地拿你的那些笔记——那些开放式编码——告诉一个 AI——这里用的是 Gemini,只是为了简便——再说一次,我们尽量保持简单——“将以下笔记归类到以下类别之一。”
Lenny Rachitsky: 给观看的人说一下,我很喜欢你分享的这些不同的提示词和公式。这是 Google Sheets 的 AI 提示词。
Shreya Shankar: 超级粉丝。
Hamel Husain: 基本上,你可以把你的 trace(追踪日志)分类到这些类别之一,这就是我们这里做的。我们把遇到的所有问题都分类到了这些类别里。
Shreya Shankar: 而且这是自动的,非常令人兴奋。我是说,AI 在做这件事。所以这也印证了一个要点——你的开放式编码(open codes)必须是详细的,对吧?你不能只写 janky,因为如果 AI 读到 janky,它没法进行分类。即使是人也做不到,对吧?它得回去回忆你为什么说 janky。所以在开放式编码中保持一定的详细程度是很重要的。
Lenny Rachitsky: 好的。所以避免使用 janky 这个词。这是一条很好的经验法则。
Shreya Shankar: 对。或者至少配上其他十个词。
Lenny Rachitsky: 哦,好的。那——
Hamel Husain: 我当时是在开玩笑。
Lenny Rachitsky: 哈哈,好吧。人们常用的、你觉得不太好的词还有哪些?
Shreya Shankar: 我觉得不是具体的某个词的问题。我认为只是人们在开放式编码中写得不够详细,所以很难进行分类。
Lenny Rachitsky: 明白。顺便说一下,之所以需要把它们映射回去,是因为比如说 Claude 或 ChatGPT 给了你建议,你做了修改和迭代,所以你不能直接回头说”好,随便放进哪个桶里”就完了,对吧?
Hamel Husain: 对,对。
Lenny Rachitsky: 好的。
Hamel Husain: 这个问题其实问得很好。去迭代、去想一想——“我喜欢这些开放式编码吗?这些对我来说真的合理吗?“——这很重要。就像 AI 做的任何事情一样,把自己放在中间稍微把控一下总是好的。
Lenny Rachitsky: 人还在回路中。仍然有我们的用武之地。好的。
Shreya Shankar: 如果我在用 AI 来做这个标注,这步中我喜欢做的一件事是,也设一个叫做”以上都不是”的新类别。这样 AI 可以实际回答”以上都不是”作为轴向编码,这就告诉我:“好吧,我的轴向编码不完整。让我回去看看那些开放式编码,搞清楚需要哪些新类别,或者如何重新措辞其他轴向编码。”
Lenny Rachitsky: 太棒了。而且这件事很酷的地方在于你不需要重复做很多次。
Shreya Shankar: 不需要。
Lenny Rachitsky: 对于大多数产品,你做一次这个流程,然后在它的基础上构建,我猜是这样的,之后就是随着时间推移不断微调?
Shreya Shankar: 完全正确。而且速度非常快。人们每周做一次,整个过程三十分钟就能完成,突然之间你的产品就比从不了解这些问题的情况好太多了。
Lenny Rachitsky: 对。想到你竟然不知道正在发生这些事情,简直荒谬。看到这个过程,我就想,“你怎么能不对自己的产品做这件事?”
Shreya Shankar: 很多人完全不知道。
Lenny Rachitsky: 大多数人。是的。我们会聊到这个。关于这些内容有一整场争论值得讨论。好的,酷。你有了这个表格。接下来是什么?
Hamel Husain: 好的。现在是重大揭晓时刻。这就是神奇的瞬间。我们已经把所有这些我们认可的编码应用到了我们的 trace(追踪日志)上。现在,你可以做那个”当当当当”了——你可以开始计数了。
从数据透视表到 AI评估(evals)
Hamel Husain: 这里有一个数据透视表,我们可以直接对这些编码做数据透视表,统计不同类别的出现次数。那我们发现了什么?在我们已经分类的这些 trace(追踪日志)上发现了什么?我们发现了 17 个对话流程问题。我非常喜欢数据透视表,因为你可以做很多很酷的事情。你可以双击这些数字往里钻取。你可以说,“哦,好吧,让我看看这些具体情况。“不过这是关于数据透视表有多酷的题外话了。
但现在我们有了一个很好的、粗略的分类——我们的问题到底有哪些?我们从混乱走向了某种有结构的思考,“哦,你知道吗?这些是我最大的问题。我需要修复对话问题,也许是这些人工转接问题。“不一定是最多的那个计数最重要,也许某个问题数量不多但非常严重,你想先修复它,但总之,现在你有了一种审视问题的方式,然后你可以考虑是否需要为其中一些问题编写 evals。
其中有些问题可能只是低级工程错误,你不需要为此编写 eval,因为修复方式非常明显。也许是输出格式错误,也许你只是忘了告诉 LLM 你希望它以什么格式输出,甚至在 prompt 里都没提过。那就直接去修改 prompt 好了。然后我们可以决定,“好吧,你要为此写一个 eval 吗?“你可能还是想写一个 eval,因为也许你可以用纯代码来测试。你可以直接检查字符串,看它是否具有正确的格式。不需要运行 LLM。
所以 evals 是有成本收益权衡的。你不想在这方面走火入魔,但你通常要立足于你实际的错误来行动。你不想跳过这一步。我之所以在这上面花这么多时间,是因为人们往往在这里迷失方向。他们直接跳到 evals——“让我写一些测试吧”——然后事情就失控了。
如何选择问题并构建评估
好的。假设我们想解决其中一个问题。比如说,我们想解决这个人工转接的问题,我们会想,“嗯,我不太确定怎么修复这个。这是一个比较主观的判断——到底应不应该转接给人工?我不能立刻知道怎么修复。这本身不是特别明显。是的,我可以修改 prompt,但我不确定。我没有百分之百的把握。”
那么,这可能就适合用 LLM 作为评判者来处理。evals 有不同类型。一种是基于代码的,如果你能做到的话应该尽量用这种,因为成本更低。LLM 作为评判者是另一种,它相当于一种元评估。你必须对这个 eval 本身再进行评估,以确保担任评判者的 LLM 确实在做正确的事,这个我们一会儿会讲。
好的。LLM 作为评判者,这是一种方式。那你怎么构建一个 LLM 作为评判者呢?
Lenny Rachitsky: 在我们进入这个话题之前,为了确保大家完全理解你刚才描述的这两种 evals——一种是你说的基于代码的,一种是 LLM 作为评判者——也许 Shreya,你帮大家理解一下基于代码的 eval 到底是什么?本质上就是一个单元测试?这是最简单的理解方式吗?
Shreya Shankar: 对。也许”eval”这个词在这里不是最准确的,但可以把它想成自动评估器。当我们发现这些失败模式后,我们想要的是,“好吧,现在我们能不能以自动化的方式检查这个失败模式的发生率——不需要我手动打标签、做所有编码和分组工作——我想在成千上万的 trace 上运行它,我想每周都运行。“这就是说,你应该构建一个自动评估器来检测那个失败模式。
当我们说基于代码还是基于 LLM 时,我们的意思是,“好吧,也许我可以写一个 Python 函数或一段代码来检查某个 trace 中是否存在那个失败模式。“这对某些事情是可行的,比如检查输出是否为 JSON,或者是否为 markdown,或者是否足够简短。这些都是可以用代码捕获的,或者至少可以近似地用代码捕获。
当我们在这里谈论 LLM 评判者时,我们说的是这是一个复杂的失败模式,我们不知道如何以自动化方式评估。所以也许我们会尝试使用一个 LLM 来评估人工转接这个非常非常狭窄、特定的失败模式。
Lenny Rachitsky: 所以让我试着复述一下你描述的内容——你想测试你的 agent 或 AI 产品的表现。你向它提一个问题,它返回一个结果。
一种测试它是否给出正确答案的方式是:如果它持续做同一件事,你可以写一段代码来判断结果是对还是错。比如,它会不会说有虚拟看房?你可以问它。
Shreya Shankar: 是的。
Lenny Rachitsky: “你们提供虚拟看房吗?“它回答是或否,然后你可以写代码根据那个具体答案来判断是否正确。
但如果你问的是更复杂的问题,而且不是二元的——一种情况下你需要人类来告诉你这是否正确。为了避免人类每次都要手动审核所有内容的解决方案,就是让 LLM 替代人类做判断,你把它叫做 LLM 作为评判者。LLM 作为判断这个结果正确与否的评判者。
Shreya Shankar: 完全正确。你总结得很好。
Lenny Rachitsky: 太好了。
Shreya Shankar: 所以人们总是觉得,“哦,这至少和创建原始 agent 的问题一样难。“其实不是,因为你让评判者做一件事——评估一个失败模式——所以问题的范围非常小,而且这个 LLM 评判者的输出就是通过或不通过。所以这是一个非常非常紧凑的范围,LLM 评判者在这种任务上非常可靠地执行。
Lenny Rachitsky: 这里的目标就是拥有一套测试套件,在你上线发布之前运行,告诉你一切是否按照你期望的方式进行?你的 agent 的交互方式是否正确?
Shreya Shankar: LLM 评判者的美妙之处在于,你当然可以在单元测试或 CI 中使用它们,但你也可以在线上用于监控,对吧?我可以每天抽样 1000 条 trace,运行我的 LLM 评判者——真实的线上 trace——看看那边的失败率是多少。这不是单元测试,但我们现在仍然获得了一个极其具体的应用质量指标。
Lenny Rachitsky: 酷。这一点真的很棒,因为很多人只把 evals 看作这种非真实世界的东西——你在它进入真实世界之前测试的东西。而你说的是,在真实世界中实际运行的东西,你也应该对它做同样的事情?
Shreya Shankar: 对。
Lenny Rachitsky: 测试你真实运行在生产环境中的东西?而且可以是每天、每小时运行的那种?
Shreya Shankar: 完全可以。
Lenny Rachitsky: 太棒了。好的,Hamel 这里有一个实际的 LLM 作为评判者的 eval 示例,我们来看看。
LLM 作为评判者的实际示例
Hamel Husain: 我很喜欢 Shreya 帮我把铺垫都做好了,非常感谢。所以我们这里有的是一个针对某一个特定失败的 LLM 评判者 prompt。正如 Shreya 所说,你应该针对一个特定失败来做,而且你要让它变成二元的,因为我们想简化事情。我们不想说,“嘿,给它打一到五分。有多好?“在大多数情况下,那只是一种不願做决策的模棱两可做法。不,你需要做一个决定。这够好吗?是或不是?
Hamel Husain: 要具体想清楚这个标准是什么可能很痛苦,但你绝对应该去做。否则这件事会变得非常难以处理,而且当你汇报这些指标时,没人知道 3.2 和 3.7 意味着什么。
Shreya Shankar: 是的,这种情况我们也一直看到,甚至在网上那些专家精心策划的内容里也是这样,比如”哦,这是你的 LLM 评判者 eval prompt。这是一个一到七分的量表。“我总是会发消息给 Hamel 说,“哦不,我们又得去对抗错误信息了,因为我们知道肯定会有人去尝试,然后回来找我们说,‘哦,我的平均分是 4.2,‘“我们就只能,“好吧。”
Lenny Rachitsky: evals 领域的戏剧性真的太多了。我们后面会聊到这个。天哪。
(以下为广告段落,已跳过)
构建评判者 prompt 并与人工标注对齐
Hamel Husain: 好的,所以这就是你的评判者 prompt。做这件事没有唯一正确的方法。用 LLM 来帮你创建它是可以的,但同样,要让自己参与到其中。不要盲目接受 LLM 的输出。在我们展示的所有案例中,我们都是这么做的。用轴向编码的时候,我们对它进行了迭代。你可以用 LLM 来帮你创建这个 prompt,但要确保你读了它、编辑了它,该做的都做了。这不一定是完美的 prompt。这只是一个保持非常简单的示例,只是为了给你展示这个思路。就像,“好的,针对这个交接失败,“我说,“好的,我要你输出 true 或 false,“这是一个二元评判者。这就是我们推荐的。然后我就一条条列出来,“好的,什么情况下应该做交接?”
比如,明确的人类请求被忽略或循环了,某种政策要求的转接,敏感的住户问题,工具数据不可用,当天预约看房或参观的请求。这些你需要跟真人谈,诸如此类。核心思路是,既然我从数据中知道这是一个失败,我就想对它进行迭代,因为我知道这种情况实际上一直在发生。正如 Shreya 所说,如果能有一种方式,不仅能在我已有的数据上评估这个,还能在生产数据上评估,那就更好了,这样可以了解这个问题的规模有多大。让我找到更多的 trace,让我有一种方式来迭代这个。我们可以拿这个 prompt,然后我再用电子表格。第一步是,当我做这个评判的时候……我写好了 prompt。
现在,很多人到这一步就停了,他们说,“好的,我有我的评判者 prompt 了。搞定了,上线吧,“然后如果评判者说是错的,那就是错的。他们直接把它当作真理接受,“好吧,LLM 说是错的,那肯定就是错的。“不要这样做,因为这是让你的 evals 和实际情况不匹配的最快方式。当人们对你的 evals 失去信任时,他们对你也失去信任。非常重要的是你不能这样做,所以在你发布你的 LLM 作为评判者之前,你要确保它和人类判断是对齐的。怎么做呢?你有那些轴向编码,你要把你的评判者跟轴向编码对比,说,“嘿,它同意我吗?我自己的评判者,它同意我吗?“直接去测量。
我们这里展示的是,好的,我说,“评估这个 LLM trace。“同样,我这里只是在用电子表格,“根据这些规则评估这个 LM trace,“而规则就是我刚才给你看的那个 prompt。我问它,“好的,是否存在交接错误,true 还是 false?“然后这一列,让我放大一点。H 列是,“好的,这个错误发生了吗?“G 列是我认为错误是否发生了。你可以看到——
Lenny Rachitsky: 你是手动过了一遍,你做了那个工作。
Hamel Husain: 是的是的,而且我们已经做过了。我们已经手动过了一遍。这不是说我们还得重新做,因为有了轴向编码这个”作弊码”,我们已经做过了。如果你需要更多数据,你可能需要再过一遍,这里面有很多关于如何正确操作的细节。你需要拆分你的数据,做所有这些事情,这样你不是在作弊,但我只是想给你展示这个概念。基本上,你可以测量一致性。现在有一件事你应该知道,作为产品经理,很多人会直接看这个一致性。他们说,“好的,我的评判者在某个百分比的时间里和人类一致。”
这听起来很诱人,但它是一个非常危险的指标,因为很多时候,错误只发生在长尾上,不那么频繁,所以如果你的错误只有 10%,那么你的评判者只要一直说通过,就能轻松达到 90% 的一致性。这说得通吗?90% 的一致性在纸面上看起来很好,但可能会误导人。
Lenny Rachitsky: 因为错误很少见,是的。
检查评判者的一致性矩阵
Hamel Husain: 作为产品经理,或者即使你自己不做这个计算,如果有人向你汇报一致性,你应该立刻追问,“好的,给我说说更多细节。“你需要深入去看。他们给你更多直觉的方式是这样的,这里是这个特定评判者在 Google 表格中的矩阵,这同样是一个数据透视表,保持简单笨拙就好。“好的,行上是人类怎么想的?我怎么想的?有没有错误,true 还是 false?然后我的评判者有没有报错误,true 还是 false?”
Shreya Shankar: 这里的直觉和 Hamel 说的一样,你需要去看每种类型的错误。当人类说 false 但评判者说 true 的情况,或者反过来,就是那些非绿色的对角线单元格。如果它们太大了,就去迭代你的 prompt,让 LLM 评判者更清楚地理解,这样你就能减少那种不对齐。你要达到一个程度,大多数情况下……你总会有一些不对齐的,这没关系。我们在课程里也会讲如何用代码来纠正那种不对齐,但在现阶段,如果你是一个产品经理,而构建 LLM 评判者 eval 的人没有做这一步,他们说”75% 的时候一致,我们没问题”,但他们没有这个矩阵,也没有迭代来确保这两种类型的错误已经降到接近零,那就是一个不好的信号。去要求他们把它修好。
Lenny Rachitsky: 太棒了。这是一个非常好的建议,知道当别人做错这件事时该注意什么。
Shreya Shankar: 是的。
evals 就是新的 PRD
Lenny Rachitsky: 其实,能不能带我们回到那个 LLM 作为评判者的 prompt?我想特别强调其中一个非常有趣的地方。最近我播客上的一些嘉宾一直在说,“AI评估(evals)就是新的 PRD”,你看看这个,就完全是这样。产品经理、产品团队,这就是产品应该是什么样,这是所有的需求,这是它应该如何运作。他们构建了一个东西,然后去测试它,通常是手动的。而这个做法的精妙之处在于,它做的完全是同一件事,而且在持续运行。它告诉你,“这个 agent 应该这样回应”,而且是非常具体的方式——“如果是这样、这样、这样,就那样做;如果是这样、这样、那样,就那样做。“这正是我一次又一次听到的东西,你现在可以直接看到。这就是产品需求文档最纯粹的形式——这个 eval 评判者准确地告诉你产品应该是什么样,而且是自动化的、持续运行的。
Shreya Shankar: 对,完全同意。它是从我们自己的数据中推导出来的,当然就是产品经理的期望。我发现很多人忽略的一点是,他们只是把自己在看数据之前就有的期望放进去,但随着我们看数据的过程,我们会发现更多一开始根本想象不到的期望,这些最终也会加入到这个 prompt 中。
Lenny Rachitsky: 这很有意思。你的建议是不是说,在构建产品之前不要跳过 evals 和 LLM 作为评判者的 prompt,仍然要写传统的单页 PRD 来告诉团队我们在做什么、为什么做、成功的标准是什么。但到最后,你可以从中提取内容,如果你在用这个流程迭代产品,甚至可以改进原来的 PRD。
Shreya Shankar: 我会再推进一步,说你会改进……它会变的。你永远不可能提前知道失败模式会是什么,你总是会发现自己认为产品应该有的新的感觉。在这些 LLM 面前,你不知道自己想要什么,直到你亲眼看到它。所以你必须保持灵活,必须看你的数据,必须……PRD 是思考这些问题的一个很好的抽象工具。但它不是终极答案。它会变的。
Lenny Rachitsky: 我很喜欢这个观点。Hamel 正在打开一份很酷的研究报告。这是关于什么的?
Hamel Husain: 如果你想了解 evals,这是你能读到的最酷的研究报告之一。它的作者是一位叫 Shreya Shankar 的人。
Shreya Shankar: 天哪。
Hamel Husain: 还有她的合作者。题目叫”Who Validates the Validated?”
《Who Validates the Validated?》研究与评判标准漂移
Lenny Rachitsky: 这真是最适合研究者的名字。
Shreya Shankar: 谢谢,谢谢。
Hamel Husain: 我应该让 Shreya 来谈这个。我认为这篇论文中最值得关注的一点是评判标准的漂移(criteria drift),以及她的发现。
Shreya Shankar: 我们做了一个超级有趣的研究,当时我们在对一些用户做调研,这些用户正在尝试写 LLM 评判者或者只是验证自己的 LLM 输出。我觉得那时候 evals 在互联网上还没有像后来那么流行。我们大概在 2023 年底开始这个项目。但作为研究者,我脑海中一直在燃烧的一个问题是,“为什么这个问题这么难?我们搞机器学习和 AI 已经这么久了,不是什么新东西,但这一次,突然一切都变得非常困难。“我们就对一群开发者做了这个用户研究,然后我们意识到,“好吧,新的地方在于你没法提前确定你的评判标准。人们对好和坏的看法会随着他们审查更多输出而改变,他们只有在看了 10 个输出之后才会想到一些一开始完全想不到的失败模式,“而这些人是专家。这些人之前已经构建过很多 LLM 流水线,现在又构建了 agent,你根本不可能一开始就把所有东西都想全。我认为这是当今 AI 开发世界中最关键的一点。
Lenny Rachitsky: 这个观点真的很好。这完全印证了我们刚才讨论的内容,所以我把这个拿出来,就是……好的——
Shreya Shankar: 这背后的研究支撑。
Lenny Rachitsky: 对,好的,很好。你还是要用同样的方式做产品,但现在你有了一个非常强大的工具来帮你确保你构建的东西是正确的。它不会取代 PRD 流程。好的。那一般来说,你们最终会有多少个 LLM 作为评判者的 prompt?我知道这显然取决于产品的复杂度,但根据你的经验,大概是什么数字?
Shreya Shankar: 对我来说,四到七个之间。
Lenny Rachitsky: 就这些?
Shreya Shankar: 没那么多,因为正如 Hamel 早先说的,很多失败模式可以通过直接修改你的 prompt 来解决。你只是没想到把它放到你的 prompt 里,所以现在你把它放进去……你不应该对所有事情都做这样的 eval,只针对那些你已经在 agent prompt 中描述了理想行为、但它仍然出问题的顽固问题。
Lenny Rachitsky: 明白了。假设你发现了一个问题,你修复了它。在传统软件开发中,你会写一个单元测试来确保它不再发生。你这里的洞见是,“如果问题已经解决了,甚至都不用费心为它写 eval”?
Shreya Shankar: 我认为如果你想写也可以,但整个游戏的核心是关于优先级排序。你的资源和时间都是有限的,你不可能为所有事情都写 eval,所以优先处理那些更顽固的领域。
Lenny Rachitsky: 大概是那些对你的业务风险最大的情况,比如说了什么 Mecha Hitler 之类的,Grok。
Shreya Shankar: 天哪。
Lenny Rachitsky: 好的。这很让人松一口气,因为这个 prompt 要把所有这些细节想清楚确实很费功夫。
Shreya Shankar: 但这是一次性的投入。从现在起,你可以永远在你的应用上运行它。
数据分析的进阶
Hamel Husain: 好的,数据分析是非常强大的,会非常快速地推动你的应用的大量改进。我们展示了最基本的数据分析,也就是计数,这对所有人都是可用的。你可以在数据分析上做得更精细。有很多不同的方法来抽样、查看数据。我们让它看起来很简单,但实际上要做好需要很多技能。要建立一种直觉和嗅觉,知道如何筛选这些数据。比如说,我发现了对话问题,这种对话流程的问题。如果我想要进一步追踪这个问题,我会想办法找到其他我还没有编码的对话流程问题。我可能会用多种方式深挖数据,有不同的方法来做这件事。这非常类似于,甚至几乎完全类似于你在任何产品上会做的传统分析技术。
Lenny Rachitsky: 简单给我们一个接下来内容的概览,然后我们来聊聊关于 evals 的争议以及其他几件事。
LLM 评判者的后续应用
Shreya Shankar: 建好 LLM 评判者之后接下来做什么?我们发现大家会尽量把它用在所有能用的地方——把它放进单元测试里,然后构建出这样的流程:“这里有一些我们观察到的出现故障的 trace 示例,因为我们已经标注过了。现在我们要把它们纳入单元测试,确保每次推送代码变更时这些测试都能通过。“他们还会把它用于线上监控。大家基于此制作仪表盘,我觉得这非常了不起。那些正在这样做的产品团队,对自己应用的运行状况有着非常敏锐的感知。但人们不会去谈论这些,因为这是他们的护城河。人们不会去分享所有这些东西,因为这是合理的。如果你是一个邮件写作助手,你在这方面做得很好,你肯定不希望别人也去做一个邮件写作助手,然后把你挤出市场。
我真的很想强调一点:尽量把你构建的这些产物在线上尽可能多的地方使用,反复利用它们来推动产品改进。很多时候,Hamel 和我会教大家一路做到这一步,然后大家就豁然开朗了,之后就不再回来了。要么是他们——我不知道——辞职了、不再做 AI 开发了,要么是他们从这一步开始就知道该怎么做了。我觉得是后者,而且我觉得这确实非常强大。
Lenny Rachitsky: 看你们演示这些,真的让我大开眼界,理解了这到底是怎么回事,以及整个过程有多么系统化。我之前总想象你就是坐在电脑前,“好吧,我需要确保哪些东西能正常工作?“而你现在展示给我们的是,这是一套非常简单的、循序渐进的流程,基于你产品中真实发生的事情,如何捕获问题、识别问题、排定优先级,然后在问题再次发生时捕获并修复它们。
Shreya Shankar: 是的,这不是什么魔法。任何人都可以做到。你需要像学习任何新技能一样去练习,但你是可以做到的。我觉得现在非常令人振奋的是,产品经理们也在做这件事、也能做到这件事,并且真的能凭借这套技能构建出非常赚钱的产品。
关于 evals 的争议
Lenny Rachitsky: 很好,这正好自然过渡到前几天我们在 X 上被卷入的一场辩论。我之前完全没意识到 evals 周围竟然有这么多争议和戏剧性。有很多人持有非常强烈的观点。Shreya,给我们大致讲讲围绕 evals 的重要性和价值的辩论双方立场,然后说说你自己的看法。
Shreya Shankar: 好的。我先缓和一下气氛——我觉得大家其实站在同一边。我认为误解的根源在于人们对 evals 有非常僵化的定义。比如,他们可能认为 evals 就是单元测试,或者认为 evals 只是数据分析部分,不包括线上监控,也不包括对产品特定指标的监控——比如实际参与的对话数量等等。我觉得每个人在谈论 evals 时脑子里想的定义都不一样。另外我想说的是,人们过去在 evals 上吃过亏。有些人确实把 evals 做得很差。一个具体的例子是他们尝试做了 LLM 评判者,但结果与预期不一致。他们后来才发现了这个问题,然后就不再信任它了,接着就说”我反对 evals”。
我百分之百理解这种感受,因为你也应该反对 Likert 量表式的 LLM 评判者。我完全同意,我们也反对那种做法。很多误解源于两件事:一是人们对 evals 的定义过于狭窄,二是人们没做好然后吃了亏,于是不想让别人犯同样的错误。而不幸的是,X(也就是 Twitter)这个平台,大家一直在互相误读对方的意思,于是你就看到各种强烈的观点——“不要做 evals,那东西很糟糕。我们试过了,根本没用。我们是 Claude Code”或者其他什么知名产品,“我们不做 evals。“这背后其实有非常多的细微差别,因为很多这些应用都是站在 evals 的肩膀上的。编程代理就是一个很好的例子,Claude Code。它们站在 Claude 基础模型……不对,是经过微调的 Claude 模型的肩膀上,这些模型已经在许多编程基准测试上进行了评估。这一点是无法否认的。
Lenny Rachitsky: 为了让你说的更清楚——Claude Code 的一位负责人,我想可能是主管工程师,上了一档播客,他说”我们不做 evals,我们就靠感觉。我们就看感觉”,所谓”感觉”就是他们直接使用它,凭感觉判断对不对。
Shreya Shankar: 我觉得那种做法是可行的。这里面有两点。第一,他们站在了同事为编程所做的 evals 的肩膀上。
Lenny Rachitsky: 就是 Claude 基础模型的那些。
Shreya Shankar: 当然对吧?我们知道他们会公布那些数据,因为我们能看到基准测试结果,我们知道谁在上面表现得好。第二点是他们实际上在错误分析方面很可能也是相当系统化的。我敢打赌他们在监控谁在使用 Claude、有多少人在使用、创建了多少 trace、这些对话有多长。他们内部团队可能也在监控,在做 dogfooding(吃自己的狗粮)。每当有什么不对劲的地方,他们可能有一个队列,或者把问题发给开发 Claude Code 的人,这个人就在隐式地做着 Hamel 谈到的那种手工错误分析。所有这些都是 evals,对吧?不可能存在这种情况——他们就这么说”我做了 Claude Code,我再也不看任何东西了”。而不幸的是,当你不去想这些、不去谈论这些时,我觉得社区……
社区中的大多数人都是初学者,或者是不了解 evals 但想学习的人,这给他们传递了错误的信息。当然,我不知道 Claude Code 具体在做什么,但我愿意拿钱打赌他们一定在以某种形式做 evals。
Hamel Husain: 我还想说的是,编程代理和其他 AI 产品有着根本性的不同,因为开发者本身就是领域专家,所以你可以省略很多步骤。而且开发者整天都在使用它,所以存在一种特殊的 dogfooding 和领域专业性——你可以把这些活动合并在一起,不需要那么多数据,不需要那么多反馈或探索,因为你自己就知道结果如何,所以你的 eval 流程理应看起来不一样。
Lenny Rachitsky: 因为你能看到代码,你能看到它生成的代码。你可以判断”这个很好,这个很糟糕。”
Hamel Husain: 对,对。我觉得很多人把编程代理的情况做了过度推广,因为编程代理是第一个真正发布到野外的 AI 产品,我认为把编程代理的做法推广到所有场景是一个错误。
Shreya Shankar: 另外一点就是,是的,工程师天生就有 dogfooding 的特质。而有很多应用场景,人们在某些领域尝试构建 AI,但那些领域没有 dogfooding 的条件——比如面向医生的场景,你不可能让医生整天去获取各种 AI 给出的错误建议还能保持包容和接受的态度。我觉得这些细微之处非常重要,需要牢记在心。
evals 与 A-B 测试之争
Lenny Rachitsky: 有趣的是,Shreya,我从你这里听到的是,如果团队中的人在进行非常细致的数据分析、错误分析(error analysis),疯狂地 dogfooding,本质上他们就是人工评估——你把这也归类在 evals 的范畴之内。如果你有时间和动力,可以这样做;或者你也可以把这些东西设置为自动化运行。
Shreya Shankar: 完全正确,这也涉及技能的问题。在 Anthropic 工作的人技能水平非常非常高。他们接受过数据分析、软件工程、AI 等方面的训练。当然,任何人都可以通过学习这些概念达到那个水平,但大多数人目前还不具备那种技能。
Hamel Husain: dogfooding 这件事有一个需要警惕的地方,就是很多人会说自己在 dogfooding。他们会说”对,我们 dogfood 了”,但真的是这样吗?很多人并没有真正在那种 visceral 层面上进行 dogfooding,而那是闭合反馈回路所必需的。这是我想补充的唯一一个提醒。
evals 与 A-B 测试的关系
Lenny Rachitsky: 还有一种感觉像是稻草人论证的说法,就是 evals 和 A-B 测试的对立。谈谈你们的看法吧,因为这似乎是这场争论中很重要的一部分。人们在讨论”如果你有 A-B 测试在检验生产环境的指标,还需要 evals 吗?”
Shreya Shankar: A-B 测试同样也是 evals 的另一种形式,我是这么认为的,对吧?当你做 A-B 测试时,你有两个不同的实验条件,然后你有一个量化某个成果的指标,你在比较这个指标。在我们的理解中,eval 就是对质量的系统性度量,也就是某种指标。你不可能在没有 eval 的情况下做 A-B 测试来比较,所以也许我们只是对此有一个不同的、有些奇怪的看法。
Lenny Rachitsky: 好的。我听到的是你们把 A-B 测试视为所做 evals 套件的一部分。我想人们说到 A-B 测试时,通常是指”我们在产品中改了某个东西,看看是否能改善我们关心的某个指标”。这样就够了吗?为什么还需要测试每一个小功能?如果它影响的是我们在商业上关心的某个指标,我们有大量的 A-B 测试在持续运行。
Shreya Shankar: 这一点提得非常好。我认为很多人过早地做 A-B 测试,因为他们从一开始就没有做过任何错误分析(error analysis)。他们只是凭假设提出了产品需求,然后相信”我们应该测试这些东西”,但结果表明,当你深入数据之后——正如 Hamel 所展示的——你看到的错误并不是你以为会出现的那些错误。而是一些奇怪的交接问题,或者,我不知道,短信那件事就很奇怪。我想说的是,如果你要做的 A-B 测试是建立在像我们今天展示的这种真正的错误分析(error analysis)基础之上的,那很好,去做吧。但如果你只是想做 A-B 测试——我们发现很多人确实是这样——只是基于你假设性地认为重要的东西来做,那我建议大家重新思考一下,把你的假设建立在扎实的基础上。
Statsig 被 OpenAI 收购
Lenny Rachitsky: 你对 Statsig 在 OpenAI 会做什么有什么看法吗?有没有什么有趣的东西?那可是一件大事,一次巨大的收购。一家 A-B 测试公司,人们都在说”A-B 测试就是未来”。怎么想?
Hamel Husain: 先补充一下上一个问题,为什么会有 evals 对 A-B 测试这样的争论?我认为,从根本上说,evals 是……人们在努力理解如何改进自己的应用,从根本上需要做的是——数据科学在产品中是有用的。看数据,做数据分析。有各种不同的工具套件,你不需要发明什么新东西。当然,你不需要数据科学的全部广度,而且在 LLM 场景下看起来会有些不同,但只是稍微有些不同。你的策略可能不同,所以本质上就是用分析工具来理解你的产品。现在人们说”evals”这个词,试图划分出一个新东西,说 evals 然后又是 A-B 测试,但如果你退后一步看,这和以前的数据科学是一样的。我认为造成困惑的原因就是——“我们需要数据科学思维”,AI 产品需要这种思维,就像任何产品都需要一样,这是我的看法。
Lenny Rachitsky: 这个观点非常好,我觉得就是”evals”这个词现在会触发人们的反应。
Shreya Shankar: 对。
Lenny Rachitsky: 如果你只是说”我们在做错误分析,做数据科学来了解我们的产品在哪里出了问题,然后设置测试确保我们知道——”
Shreya Shankar: 那太无聊了,听起来很无聊。不,不,不。我们需要一个神秘的术语,比如”evals”,才能真正推动起来势头。你关于 Statsig 的问题,我觉得很令人兴奋。说实话,我对这件事了解不多,因为我只是觉得他们是一家……很多人在用的一款工具的公司,也许恰好 OpenAI 收购了他们。我确定他们之前一直在使用 Statsig,我确定 OpenAI 的竞争对手也在使用 Statsig,所以也许这次收购有某种战略考量。我完全不知道,我对此没有任何内情,但我觉得这些才是更大的问题,而不是”这是否从根本上改变了 A-B 测试或让 evals 变得更重要”。我认为 evals 一直都很重要,我认为 OpenAI 一直在做某种形式的 evals,而且从历史上看,OpenAI 甚至走得更远——他们会去看所有的 Twitter 情绪,尝试做一些回顾性分析,然后将其与产品关联起来。当然,他们在发布新的基础模型之前肯定做了某种程度的 evals——
然后将其与产品关联起来。当然,他们在发布新的基础模型之前肯定做了某种程度的 evals,但他们做的远不止于此,比如”好吧,让我们找到所有抱怨的推文,所有抱怨的 Reddit 帖子,然后去搞清楚到底发生了什么。“这说明 evals 非常、非常重要。还没有人真正搞明白。人们在使用所有能获取的信号来源来改进他们的产品。
Hamel Husain: 我想说的是,我真的很希望这能在 OpenAI 内部推动或创造出一种关注点,希望如此。到目前为止,各大实验室理所当然地把重心放在像 MMLU 分数、human eval 这样的通用基准测试(benchmarks)上,这些对基础模型来说确实很重要。但这些和产品特定的 evals 关系不大——比如我们今天讨论的交接问题之类的,它们往往没有相关性。
Shreya Shankar: 对,它们和数学问题求解能力不相关,很遗憾。
Hamel Husain: 没错。如果你看看那些 eval 产品,比如最近之前一些大实验室所拥有的那些,它们没有错误分析(error analysis)功能。它们有一套通用工具——余弦相似度、幻觉评分之类的,但那不管用。作为第一次尝试,它是不错的。还行。至少你在做些什么,让人们去看数据。但最终,我们希望看到的是,在 eval 流程中融入更多的数据科学思维。这才是我们希望工具能达到的目标。
Shreya Shankar: 对,Pamela 和我不应该是这个星球上仅有的两个在推广面向应用特定 evals 的结构化思维方式的人。这让我百思不得其解。为什么全世界只有我们两个人在做这件事?到底怎么回事?我希望我们不是仅有的两个人,希望更多人能跟上。
关于 evals 的常见误区
Lenny Rachitsky: 你们在 Maven 上的课程是 Maven 平台上收入最高的课程,这显然说明有需求和兴趣,而且我觉得你们这边的人越来越多了。有趣的是,你一直在 Twitter 上分享的一个例子很有启发性——所有人都在说 Claude Code 根本不在乎 evals,他们完全凭感觉(vibes),然后大家就说,既然它是最好的编程智能体,那显然这种做法是对的。但最近又有各种讨论说 Codex,OpenAI 的 Codex 更好,所有人都在转向它,而他们是非常支持 evals 的。
Shreya Shankar: 我知道。
Lenny Rachitsky: 是啊。
Shreya Shankar: 每次都让我无语。互联网太不靠谱了。我最喜欢的一件事是昨天,我和几个实验室同事出去吃甜点什么的,有人问:“你更喜欢 Codex 还是 Claude?“另一个人说:“我喜欢 Claude。“然后又有人说:“但新版 Codex 更好。“第一个人接着说:“哦,但我上次看是两天前,所以我的想法可能——可能我没跟上最新动态。“我当时就想,天哪。
Lenny Rachitsky: 太真实了,太真实了。这就是我们生活的世界。天哪。好吧,我想问问关于 evals 人们最大的误解,以及成功的技巧和窍门。也许每人各分享一两个。先从误解开始,让我先问 Hamel。关于 eval,人们最常见的误解有哪些?
Hamel Husain: 排第一的是:“嘿,我买个工具,插上就能帮我做 eval 了。我为什么要操心这个?我们生活在 AI 时代,AI 不能自己 eval 吗?“这是最常见的误解。人们太想要这个了,以至于真的有人卖这种东西,但它不管用。这是第一个。
Lenny Rachitsky: 哎,很多人类依然很有用。我觉得这是个好消息。
Hamel Husain: 我经常看到的第二个是:“嘿,就是不看数据。“我做咨询的时候,人们总是带着问题来找我,我第一句话就是:“让我们去看看你的 trace(追踪日志)。“你能看到他们瞪大眼睛:“什么意思?“我就说:“对,现在就看。“他们会很惊讶,因为我要去看单条 trace(追踪日志)。而每次,100% 的情况下,都能学到很多东西,找到问题所在。我觉得人们就是不知道看数据有多么强大,就像我们在这期播客里展示的那样。
Shreya Shankar: 我同意这一点。
Lenny Rachitsky: 这就是前两个?好的。
Shreya Shankar: 是的。
Lenny Rachitsky: 还有别的吗,还是说解决这两个就够了?
Shreya Shankar: 哦,那两个确实是最重要的。那我想补充的第三个是,做 evals 没有唯一正确的方法。错误的做法有很多种,但正确的做法也有很多种。你需要考虑你的产品处于什么阶段,你有多少资源,然后找到最适合你的方案。它总是涉及某种形式的错误分析(error analysis),就像我们今天展示的那样,但你如何将这些指标落地操作化,会根据你的具体情况而变化。
实用建议
Lenny Rachitsky: 太棒了。好的。大家刚开始做 eval,或者想改进现有做法的时候,有什么技巧和窍门想留给他们的?
Shreya Shankar: 第一条建议就是,不要对看数据感到恐惧或害怕。这个过程,我们尽量让它结构化,但不可避免地会有各种问题冒出来,这完全没问题。你可能觉得自己做得不够完美,那也没关系。目标不是把 evals 做得完美,而是切实改善你的产品。我们保证,无论你做什么,只要你做了这个流程的一部分,你就会找到切实可行的改进方向,然后你会在此基础上不断迭代自己的流程。
另一条建议是,我们非常支持 AI。在整个过程中,用 LLM 来帮助你整理思路。这可以是方方面面的——从最初的产品需求开始,想清楚怎么为自己整理这些需求。想清楚怎么根据你创建的开放式编码(open coding)来改进那个产品需求文档。不要害怕用 AI 来帮你更好地呈现信息。
Lenny Rachitsky: 好的,所以不要害怕,在流程中尽量多地使用 LLM。
Shreya Shankar: 但不是用来替代你自己。
Lenny Rachitsky: 对。好的,很好。工作还在。太好了。Hamel。
打造自己的数据查看工具
Hamel Husain: 好。让我分享一下屏幕,因为我想展示点东西。在 Shreya 说的基础上补充一下——如果你在这期播客里听到了什么短语,你听到最多的可能就是”看你的数据”。这件事太重要了,以至于我们教学中会说你应该创建自己的工具,让看数据变得尽可能简单。在之前的实时演示中,我给你们展示了一些标注数据的工具。我合作的大多数人,他们意识到了这有多重要之后,就开始凭感觉(vibes)编程——我们不应该说凭感觉(vibes)编程——自己做工具,而且现在比以往任何时候都便宜,因为你有 AI 可以帮忙。
AI 非常擅长创建简单的 Web 应用,可以展示数据、写入数据库。很简单。以 Nurture Boss 的场景为例,我们想消除看数据的所有摩擦。你在这里看到的只是一些截图,展示他们创建的应用长什么样。就是这样——“好的,他们有不同的渠道,语音、邮件、短信。有不同的线程,系统提示词(system prompt)默认隐藏。“一些小的体验优化。然后他们确实有这个轴向编码(axial coding)的部分,你可以看到红色的不同错误的计数。他们把那部分自动化了,做得很好,而且几个小时就搞定了。很难有一个通用的万能方案来看你的数据。你不必一开始就走到这一步,但值得思考的是,让看数据变得尽可能简单,因为,再说一遍,这是你能从事的最强大的活动。这是 ROI 最高的活动。有了 AI,就是消除所有摩擦。
Lenny Rachitsky: 太厉害了。再说一次,我觉得 ROI 这一点太重要了。我们甚至还没有充分展开谈这一点。这里的目标是让你的产品变得更好,从而让你的业务更加成功。这不是什么抓 bug 之类的小练习。这是让 AI 产品变得更好的方法,因为用户体验就是用户与你的 AI 交互的方式。
Hamel Husain: 完全同意。甚至可以说,我们教学生的时候会说:“嘿,你在做这些 evals 的时候,如果看到有什么不对的,直接去修就好了。“重点不是拥有 evals,一套漂亮的 eval 套件,你可以指着它、编辑它然后说:“哦,看看我的 evals。“不是的,直接修复你的应用,让它变得更好。如果问题很明显,就直接干。完全同意你说的。
Lenny Rachitsky: 太棒了。我还有一个没问到的问题,但我觉得大家在想的——你在这上面花多长时间?第一次做通常需要多久?
时间投入与持续维护
Shreya Shankar: 我可以结合自己参与的应用来回答。通常我会花三到四天时间,和相关人员一起做最初几轮的错误分析(error analysis)。大量标注工作,觉得状态不错了,就建一个类似 Hamel 展示的那种表格,让大家都能认同和信服,甚至搭几个 LLM 作为评判者。但这属于一次性投入。一旦搞清楚怎么把它集成到单元测试里,或者写一个脚本自动对抽样运行,再设一个 Cron Job 每周自动执行,就好了。我发现自己可能花了不少时间看数据,因为我就是那种对数据如饥似渴的人,好奇心太强了。我从这个过程中获益匪浅,它让我在与他人的合作中远远超出预期,所以我想一直做下去。但并不是非得这样不可。之后大概每周花 30 分钟就够了。
Lenny Rachitsky: 所以基本上前期花一周左右,之后每周 30 分钟来持续改进和扩充你的 eval 套件?
Shreya Shankar: 对,真的不需要太多时间。我觉得大家只是被前期投入的时间吓到了,然后以为后面也得一直这么做。
数据驱动的产品洞察
Lenny Rachitsky: 太棒了。还有什么想和听众分享的吗?在进入非常令人期待的快问快答环节之前,还有什么想特别强调的观点吗?
Hamel Husain: 我想说,这个过程其实很有趣。你会觉得,好吧,你在看数据,听起来像是在标注东西。实际上,我昨天刚看了一个客户的数据,用的就是完全相同的流程。那是一个发送邮件的应用,发送招聘邮件来吸引候选人应聘。我们决定开始看追踪日志(trace),直接就上手了。“来,看看你的追踪日志。“我们看了一条 trace,第一眼就看到一封邮件,措辞是这样的:“鉴于您的背景,blah blah blah……”我立刻就问了对方——这就是你戴上产品帽子、用挑剔眼光去看的时候,也是有趣的部分所在。
我说:“你知道吗?我很讨厌这封邮件。你自己喜欢吗,‘鉴于您的背景’?“当我收到一条”鉴于您的背景,逗号”开头的消息时,我直接就删了。我心里想:“鉴于您在机器学习方面的背景 blah blah,这是什么?“这就是模板化的东西。我问对方:“我们能做得更好吗?这听起来就是千篇一律的招聘邮件。“他们说:“哦,确实,也许吧。“因为他们之前还挺自豪的,觉得”AI 做对了,它发送了正确信息的邮件,有正确的链接、正确的名字,一切都没问题。“有趣的地方就在这里——戴上你的产品帽子,深入思考:这真的够好吗?
Lenny Rachitsky: 在进入非常令人期待的快问快答之前,我想确保我们提到一点——这只是做好这件事所需知识的冰山一角。但我认为这是目前关于如何做好这件事最好的入门指南了。
Shreya Shankar: 太好了。
课程推荐
Lenny Rachitsky: 但我想我们已经做到了。不过你们两位教了一门课程,对那些真正想精通此事、认真对待的人会深入得多。请分享一下你们在课程里还教了哪些我们没涉及的内容,以及作为 Maven 上那门课的学生还能获得什么。
Shreya Shankar: 好,我来介绍一下教学大纲,然后 Hamel 可以谈谈各种附加权益。我们按照生命周期来讲:错误分析(error analysis),然后是自动化评估器,再是如何改进你的应用,如何为自己打造那个飞轮。我们还有一些几乎没人听过或教过的专题,这部分非常令人兴奋。其中一个专题是:如何构建自己的错误分析(error analysis)界面。我们会展示我们实际构建的界面,还会现场用新数据即时编码演示。我们展示如何使用 Claude Code、Cursor,或者当天想用的任何工具来构建这些界面。
我们还会广泛讨论成本优化。我合作过的一些人,他们的 evals 做得很好,产品也很好,但一切都非常昂贵,因为他们用的是最先进的模型。我们怎样才能在某些场景下用更便宜的 GPT-5-nano、4-mini 之类的替代最贵的 GPT-5,大幅节省成本,同时保持相同的质量?这方面我们也会给一些技巧。Hamel,到你了。我们还有很多附加权益。
Lenny Rachitsky: 好,讲讲那些权益。
Hamel Husain: 好,附加权益。我最喜欢的权益是一本 160 页的书,写得非常细致,我们精心打造的,完整走过了整个 eval 流程的详细步骤,作为课程的补充资料。你不必坐在那里辛苦记笔记,我们已经替你做了所有苦活,把一切详细记录并整理好了,非常实用。另一个很有趣的东西——这个灵感其实来自你,Lenny——就是,这是一门 AI 课程,教育不应该只是看讲座和做作业,学生也应该有一个 AI 来帮助他们。我们的做法是,就像你有 LennyBot 一样。
Lenny Rachitsky: dot com。
Hamel Husain: 对,lennybot.com,我们用同样的软件做了同样的东西,把我们所有关于 evals 说过的内容都放进去了。每一节课、每一次答疑、每一条 Discord 聊天、任何博客、论文,我们在公开场合和课程中说过的所有内容,全都放进去了。我们让一批学生测试过,他们说很有帮助。我们给所有学生 10 个月免费、无限制使用的权限,随课程一起提供。
Lenny Rachitsky: 太厉害了。那之后会收费吗?
Hamel Husain: 我也不知道。我走一步看一步,不知道后面会怎样。
Lenny Rachitsky: 八个月后再想办法。我刚才在想,这整场访谈其实可以让我们的两个 bot 互相对话。
Shreya Shankar: 那太搞笑了。我会看的,但只看 10 分钟,之后我就不知道它们在说什么了。
Lenny Rachitsky: 哈,也许 30 秒就够了。顺便问一下,你们有基于语音模式来训练它吗?那是 Delphi 产品里我最喜欢的功能。如果没有的话,你们应该试试。
Hamel Husain: 哦,我想想……我记不清了,我得去看看。
Lenny Rachitsky: 你确实应该试试。现在我们有了这期播客,你可以用这些内容来训练它。是 11Labs 驱动的,效果特别好。好,那他们怎么获取……我觉得这样就行,报名你们的课程之后就能用上了。
Shreya Shankar: 对,报名课程后会收到一堆邮件,一切都会说清楚,希望如此。
Lenny Rachitsky: 太好了。
Shreya Shankar: 我们还有一个 Discord,所有上过课的学生都在里面。那个 Discord 非常活跃,我度假时在飞机上都会收到通知。
Lenny Rachitsky: 甜蜜的烦恼。太棒了。好,我们现在进入非常令人期待的快问快答环节。我准备了五个问题。准备好了吗?
Shreya Shankar: 准备好了,开始吧。
快问快答
Lenny Rachitsky: 来吧。好的,我会在你们两位之间轮流提问。想分享就分享,不想答也可以跳过。第一个问题,Shreya,你最常向别人推荐的两三本书是什么?
Shreya Shankar: 我喜欢推荐一本小说,因为生活不只有 evals。最近我读了 Min Jin Lee 的《Pachinko》,非常棒的一本书。另外我还在读《Apple in China》,作者名字一时想不起来了,但这本书更像是一篇纪实报道,由一位记者撰写,讲述 Apple 过去几十年在亚洲如何开展大量制造流程,非常开眼界。
Lenny Rachitsky: 太棒了。Hamel?
Hamel Husain: 好的,我书就在手边。我是个书呆子。好吧,我没有 Shreya 那么酷。我推荐的都是教科书,这也是我最喜欢的。这本是非常经典的一本,Mitchell 的《Machine Learning》。它是偏理论的,但我喜欢的一点是,它真正让你深刻理解奥卡姆剃刀不仅适用于科学,也适用于机器学习和 AI。很多时候,最简单的方法——工程上也一样——泛化效果反而更好。这是我从那本书中深刻内化的东西。我还很喜欢这本,也是教科书。我说过我是个书呆子吧。这本也很老了,是 Norvig 的《算法》。我喜欢它是因为它展现了人类的智慧,里面有很多在计算领域巧妙而实用的东西。
Shreya Shankar: 他和 Berkeley 就在一条街上。
Lenny Rachitsky: 做那个研究的人?
Shreya Shankar: 对,教科书的作者。
Lenny Rachitsky: 太酷了。天哪,书呆子们,我爱了。好,下一个问题。最近最喜欢的电影或电视剧?我先问 Hamel。
Hamel Husain: 好吧,我是两个孩子的爸爸,没什么时间看电视或电影,所以我就看孩子们看的东西。上周我看了三遍《Frozen》。
Lenny Rachitsky: 才三遍?哦,是一周之内。好吧。
Hamel Husain: 这就是我的生活。
Lenny Rachitsky: 很好,Hamel。《Frozen》,我喜欢。好,Shreya。
Shreya Shankar: 我没有孩子,所以能给出精彩的答案。实际上,我丈夫和我最近在看《The Wire》。我们小时候都没看过,所以开始看了,真的很棒。
Lenny Rachitsky: 我感觉每个人生命中都会经历这个阶段——终于有一天决定”我要看《The Wire》”。
Shreya Shankar: 对,我们现在就处于那个阶段。
Lenny Rachitsky: 就像要花掉你一年时间。但确实很棒,太好看的剧了。不过集数太多了,每集还都是一小时。
Shreya Shankar: 我知道,我知道。
Lenny Rachitsky: 真的是一个巨大的 commitment。
Shreya Shankar: 我们一周只看两三集,进度很慢。
Lenny Rachitsky: 值得。好,下一个问题。你最近有没有发现一个特别喜欢的产品?先从 Shreya 开始。
Shreya Shankar: 说实话我真的很喜欢用 Cursor。还有 Claude Code。我解释一下为什么。我本质上更偏向研究者。我写论文、写代码、构建系统,什么都做。我发现一个工具……我对 AI 辅助编程非常看好,因为我总是要同时扮演很多角色。现在,我可以对我构建和撰写论文的东西更有雄心了,所以我对此非常兴奋。Cursor 是我进入这个领域的入口,但我现在发现自己一直在追赶各种 AI 辅助编程工具。
Lenny Rachitsky: Hamel?
Hamel Husain: 我也很喜欢 Claude Code,喜欢的原因是我觉得它的用户体验非常出色,能看出里面倾注了很多心血。作为一个终端应用能做到这个程度,真的令人印象深刻。
Lenny Rachitsky: 你俩都这么喜欢 Claude Code,而它偏偏就是靠凭感觉(vibes)构建的,真是讽刺。
Shreya Shankar: 我觉得这个说法不对,它不只是靠凭感觉(vibes)构建的。
Lenny Rachitsky: 这就对了。好,还有两个问题。Hamel,你有没有最喜欢的人生格言,在工作或生活中常常回想起来的?
Hamel Husain: 持续学习。像初学者一样思考。
Lenny Rachitsky: 很美。Shreya?
Shreya Shankar: 我喜欢这个。对我来说,是始终试着站在对方的角度思考。我有时会在网上看到各种争论,比如关于 eval 竞赛的辩论,我会真的去想:“好吧,把自己放在他们的位置上。可能存在一个善意的解读。“我认为我们团结在一起远比互相争斗更强大。我对 evals 的愿景不是让 Hamel 和我成为亿万富翁,而是让每个人都能构建 AI 产品,大家达成共识。
Lenny Rachitsky: 以及每个人都成为亿万富翁。
Shreya Shankar: 没错。
Lenny Rachitsky: 太棒了。最后一个问题。当我同时有两位嘉宾时,我总喜欢问这个问题,先从 Hamel 开始。你最喜欢 Shreya 的什么品质?我问她同样的问题,反过来。
Hamel Husain: Shreya 是我认识的最有智慧的人之一,尤其是考虑到她比我年轻那么多。说实话我觉得她比我智慧得多。她非常踏实,对事物有着非常平和的视角。这一点总是让我印象深刻。
Lenny Rachitsky: Shreya?
Shreya Shankar: 我最喜欢 Hamel 的是他的能量。我不认识任何人能像 Hamel 一样始终保持着那样的势头和能量。我经常想,如果不是因为 Hamel,我可能早就不再那么关注 evals 了。每个人生活中都需要一个 Hamel,真的。
Lenny Rachitsky: 好了,我们现在每个人都有一个 Hamel 了。这次对话太精彩了,完全是我期望的样子。我觉得这是我见过的最有趣、最深入、最容易消化的 evals 入门指南。非常感谢你们两位抽出时间。最后两个问题。大家在哪里可以找到你们?在哪里可以找到这门课程?听众怎样能帮到你们?先从 Shreya 开始。
Shreya Shankar: 你可以通过电子邮件联系我,地址在我的网站上。Google 我的名字,这是找到我网站最简单的方式。如果你 Google “AI Evals for engineers and product managers”,或者直接搜 “AI Evals course”,就能找到这门课程。之后我们会分享一些链接,方便大家找到。如何能帮到我们?对我来说永远有两件事。一是有问题时就来问我,我会尽快回复。另一件是告诉我们你们的成功案例。让我们持续走下去的动力之一,就是有人告诉我们他们实施了什么、做了什么,一个真实的案例研究。Hamel 和我看到这些会非常兴奋,这真的让我们坚持下去,所以请多多分享。
Hamel Husain: 找到我很简单。我的网站是 Hamel.dev,我可以把链接发给你。你也可以在社交媒体上找到我,LinkedIn、Twitter 都有。最有帮助的,是呼应 Shreya 说的——我们非常希望不是只有我们在教 evals。我们很乐意看到其他人也来教 evals。任何形式的博客文章、写作,尤其是你在这个过程中学到了东西想分享的,我们都非常乐意帮忙转发和推广。
Lenny Rachitsky: 太棒了,非常慷慨。非常感谢你们两位来到这里。我真的很感激,你们两位手头都有很多事情,谢谢。
Shreya Shankar: 谢谢 Lenny 邀请我们,也谢谢你的所有夸奖。
Lenny Rachitsky: 这是我的荣幸。大家再见。
非常感谢大家的收听。如果你觉得这期节目有价值,可以在 Apple Podcasts、Spotify 或你喜欢的播客应用上订阅。也请考虑给我们评分或留下评论,这真的能帮助更多听众找到这个播客。你可以在 Lennyspodcast.com 找到所有往期节目或了解更多关于节目的信息。下期再见。
术语表
| 原文 | 中文 |
|---|---|
| agreement | 一致性(指人类标注与 LLM 评判者之间的一致程度) |
| axial codes/axial coding | 轴向编码(axial coding) |
| benchmarks | 基准测试 |
| benevolent dictator | 仁慈独裁者(benevolent dictator) |
| bullish | 看好 |
| commitment | commitment(时间/精力上的投入承诺) |
| cosine similarity | 余弦相似度 |
| criteria drift | 评判标准漂移(criteria drift) |
| dogfooding | dogfooding(吃自己的狗粮,即团队内部使用自己的产品) |
| entry point | 入口 |
| error analysis | 错误分析(error analysis) |
| evals | AI评估(evals) |
| failure mode | 失败模式 |
| foundation model | 基础模型 |
| hallucinating | 产生幻觉(hallucination) |
| hallucination score | 幻觉评分 |
| human eval | human eval |
| janky | janky(混乱/粗糙) |
| lead management | 线索管理(lead management) |
| Lenny Rachitsky | Lenny Rachitsky(播客主持人) |
| Likert scale | Likert 量表 |
| LLM as a judge | LLM 作为评判者 |
| MMLU | MMLU |
| moat | 护城河 |
| Nurture Boss | Nurture Boss(产品名称) |
| observability tool | 可观测性工具(observability tool) |
| Occam’s razor | 奥卡姆剃刀 |
| open coding | 开放式编码(open coding) |
| Pamela | Pamela |
| pivot table | 数据透视表 |
| RAG retrieval | RAG 检索 |
| retrospective | 回顾性分析 |
| sample | 抽样 |
| straw man argument | 稻草人论证 |
| system prompt | 系统提示词(system prompt) |
| theoretical saturation | 理论饱和(theoretical saturation) |
| tool calls | 工具调用 |
| trace | trace(追踪日志) |
| vibe | 凭感觉(vibes) |
| visceral | visceral(切身/ visceral 层面的) |
| weasel way | 模棱两可的做法 |
此文档由 AI 分片翻译(translate_long_document)