Hamel Husain and a colleague discuss whether non-data scientists can write AI evals. They argue it's teachable when people own a feature and care about customers. The work involves binary labeling, spreadsheets, and writing SOPs, but it's a muscle any team can develop with practice.
And my question is, do you find that people can be effective in the analyze stage if they're not data scientists? And how do they learn that skill? Or what do you find?
That's a good question. I think they can be taught. I think there are some people, if they feel a sense of ownership over the feature and they're able to put themselves in the shoes of the customer, they can try to understand this is bad, this is good. And you know, we have good methodologies to do this. And you know, Shreyas mentioned a bit, a few of them, like later in the chapter, like, you know, binary only or preference. Selection, I think those become reliable enough. What I find that, okay, I could label some data in a day and I could get a couple hundred samples, but then I'll spend a lot more time than a day trying to calibrate my auto—
my LLM evaluator to match those samples.
Actually very hard. And I've been through this exercise dozens of times.
I find that like very data science-y though, even the method.
It sounds very data science-y, right? But I find that the analysis part, if you can put it in a spreadsheet, And you can just tell people to say 1 or 0, or left or right. You can actually decompose that. Yes, it takes a bit of time. It takes— it's a muscle that the team needs to learn. And it's a muscle also writing SOPs, writing evaluation annotation criteria is a muscle.
Okay, I agree with you, you can get people to label, you know, binary labels, go through spreadsheets, things like that. One thing I found that I took for granted is the skills involved in studying that data afterwards, like aggregating it, slicing and dicing that data, whatever. You know, we have a lot of skills as data scientists to just do— we think it's like breathing air or something very simple, but I found that people struggle with—
I can't— part of categorized failure modes, that's slightly more tricky that not everyone can do as part of the analysis box.
