The difference might be people are actually held accountable for the results.

And, in most cases, they tend to learn from their mistakes.

Why is that better if in aggregate, the people who can be held accountable are worse anyway?

We are killing thousands on the road to be sure we can blame a driver instead of a computer as one example.

In the self driving case people are still held accountable. Now it’s the model developers and car manufacturers instead of drivers.

Yes, we also clearly see and upwards pressure on people.

People are way more reliable.

That's why we have code reviewing. If you review your AI generated code, understand it, and completely agree with it, that's fine of course. But the message of this article is that you still can't blindly rely on AI for even such simple tasks.