I've had similar results in both mathematics and programming. For one paper I was writing I wanted to try them and it was a fairly straightforward problem counting the number of permutations. I spent much more time trying to get the AI to figure it out than it took to actually solve it. Couldn't get it to do it even after I solved the problem. Similarly in coding I've had it fail to find trivial bugs like an undefined keyword, which would have easily been caught had I turned on ctags (because the keyword was inherited from a class, which made it so easy to miss). But similarly I've had them succeed on non-trivial tasks and those have been of great help, though nothing has ever been end-to-end just the AI.

So I agree. I think these tools are very helpful, but I also think it is unhelpful that people are trying to sell them as much more than they are. The over hype of them not only legitimizes any pushback but provides ammunition. I believe the truth is that they're helpful tools, but are still far from replacing expertise. I believe that if we try to oversell them then we run the risk of ruining their utility. If you promise the moon you have the deliver the moon. That's different from aiming for the moon and landing in the trees.