Comment by avilay
15 days ago
+100 I tried to get Claude and Gemini to build me a course on probabilistic ML from scratch. They both failed. This was a few months ago. The best they could do was design a course curriculum. My learning since then has been through asking the model to explain a very narrow topic from the curriculum. And then go through the agonizing back-and-forth questioning and answering to actually build an understanding. No doubt the models are great at this and my learning has never been so through with such solid foundations. But zero-shot it was not!
Isn't prompting a full course in one shot similar to prompting a full fledged app? Aren't we supposed to break it down into plan mode vs execute mode?
But that fits with the point that you’d have to already know the domain to be able to break it down into discrete units for steering the LLM.