How can we ensure that we use the LLM to generate useful code/features and fix bugs? I don’t think we can put the same expectations on LLMs as on junior coworkers. You hire junior developers in order to have people that are expected to grow and learn. LLMs come pre-trained (i.e. there is no point in teaching them as they are fixed neural nets after training). But anyway let us take the idea of having a multitude of “agents” to the logical conclusion. Say that we have a bunch of teams working on producing[...]