These LLMs are confabulation machines. They're as good as the knowledge of the person who is driving the current session. For coding problems this makes them

These LLMs are confabulation machines. They're as good as the knowledge of the person who is driving the current session. For coding problems this makes them like a very good teddy bear debugger, but do not expect independently produced creative work from them.