In this segment of his YouTube video, Hamel Husain emphasizes that successful open coding requires domain expertise. While outsourcing annotation to contractors is acceptable, they must be trained in proper open coding techniques. Husain warns that contractors often default to simply reproducing AI instructions without critically assessing the output. He stresses that the evaluation should focus solely on whether the AI’s result is correct, rather than assigning blame for mistakes.
Absolutely critical when you do open coding that you have a domain knowledge to do the open coding. It's totally fine to get contractors and such to do things, but you kind of have to teach them how to do the open coding. You'll be surprised in the way that contractors fail at this. They'll do things like, oh well, the AI did what it was told to do. They'll open code like and they'll just pass it. But that's not how you do open coding. We don't care what we told AI to do. Like, did it do the right thing? Like, that's all we care about. Like, we don't care, like, whose fault it is and all that stuff.
