And the hyperbolic headline strikes again!

Linus actually has a fairly nuanced take on this. It is a tool, it shouldn't be blindly trusted but it has its place.

First, AI is not just a tool like other tools. It's much more random and complex, which makes it very different from other tools.

You can't simply say that it's useful. How useful it is depends on what you are trying to do. If you are making some software that you need to understand well and be responsible for, then I think AI could make that harder, because it doesn't force you to think through and understand everything, and having to do that probably makes it just as fast to write it by hand anyway.

Don't forget that part about understanding and being responsible. It's important. Even if AI got 100000x better, understanding and being responsible would still take about the same amount of time, because it has to be a very manual process, and it taking time is part of the point of it. Also, while a computer can't take responsibility, a human has to do it, so using AI could easily reduce the responsibility taken by the human without increasing it anywhere else.