> What is your best argument against this possibility? People were laughing at the idea of AI doing math at all a few years ago, I feel that all those still secure in human superiority in any other domain won't be feeling the same way in a few years once robotics reaches a critical threshold.

Would you trust an AI or a humanoid robot to safely fly and land a commercial plane with 300+ human passengers and zero human captains?

We will always need humans for highly critical roles and tasks. Building yet another web app framework or building apps in Nextjs or React isn't one of them.

> I think the only thing keeping humans in charge at that point would be very effectively managed regulation which gives humans a decisive, if artificial, advantage.

It just means that someone has to take full responsibility for when it all goes wrong given that the cost might go into the millions or tens of millions of dollars.