First many people can be that guy? If that is 5% that means 95% of the rest should go.

Second, just because a good engineer can have much higher throughput of work, multiplied by AI tools, we know the AI output is not reliable and needs a second look by humans. Will those 5% be able to stay on top of it? And keep their sanity at the same time?

Do not assume constant demand. There are whole classes of projects which become feasible if they cash be made 10x faster/cheaper.

As for maintaining sanity… I’m cautiously optimistic that future models will continue to get better. Very cautiously. But cursor with Claude slaps and I’m not getting crazy, I actually enjoy the thing figuring out my next actions and just suggesting them.