Comment by tempestn

The point is you need to collaborate with the LLM, not just expect it to provide fully-formed answers. I've saved countless hours of work over the past few months using GPT-4 to help me solve problems. One general framework can be to use it to help you google a solution instead of a problem. Googling a problem gives you all kinds of results, which you then need to dig through and understand in order to evaluate whether there's something that can be adapted to your use case. On the other hand, you can ask GPT-4 for solutions to the problem, and then google those, which cuts the process down significantly. Of course that's only one example. The pattern though is that you need to play to its strengths. It's not particularly smart, but it's very knowledgeable. Just like a human though, you can't expect it to perfectly regurgitate all the details. You use it for direction, and then refine from there, through conversation with the LLM as well as external research.