>“I don’t necessarily disagree with you about AI ‘art’, but in my field—which you do not understand, and in which the underlying build process is often not the creative output—AI presents very real productivity gains”

Being able to prove such gains in better products would be a start. And an emphasis on how it assists existing engineers/mathmaticians/researchers, not that any accomplishment made with AI assistance is "AI solves problem".

I don't know whatever happened to "words are cheap". I guess it literally made money to say words, so that adage is false for the time being.

>I am skeptical of there being sufficient data to build “ethical” training datasets

Well if all those scam job ads paying 100/hr to create AI training content was not a scam and instead the approach from the start, there may have been a chance to bridge that gap ethically. The industry chose to break things and is trying to act mad that people are mad at all the broken stuff.

These results are entirely a consequences of the actions chosen. And I don't believe there was ever an honest consideration of there being ethical training datasets. They just thought they could brute force society with fearmongering and bribes. The BOTD was already low in the beginning but completely gone now.