General Discussion

This was one of FT's "Lunch with the FT" features, with this interview taking place at Altman's Napa Valley farm house, where he's spent much of his time since he and his husband welcomed their first child via surrogacy. Altman offered to cook lunch - a simple vegetarian meal - for the journalist instead of going to a restaurant where he'd be pestered by selfie seekers.

But he still reminded the journalist - FT editor Roula Khalaf - that as OpenAI CEO, he has the most important job in the world.

Not surprisingly, she writes that she found Altman "brimming with confidence" and "radiating ambition" and "convinced of his own destiny."

He's gone from thinking AI is as important as the Industrial Revolution to believing that the "explosion in creativity" from AI makes it more comparable to the Renaissance. (Note: Generative AI like OpenAI's is FAKE creativity via mashups possible only because OpenAI stole as much of the world's intellectual property as it could, and the theft continues, with OpenAI's bots scraping some websites as much as several hundred times a day.)

Altman, ever the company hypester,.tells her "people are saying" their most advanced models are "genius-level intelligence.". (People are saying otherwise, too.)

She asks Altman about recent questions on his company training AI on copyrighted intellectual property, including art from a Japanese anime studio whose founder has made his contempt for AI art clear. Altman tells her " compensation for artists may be required" (her words, not necessarily his) but he'd prefer to release the tools first and then "find answers to the questions that arise" (again, her words). And this is standard Silicon Valley "move fast, break things, steal things, pay the best lawyers whatever's necessary so you won't have to pay for what you broke and stole" business philosophy.

He compares AI art to the invention of the camera, a BS comparison. (Cameras capture reflected light from a real subject in front of them, and artistry and creativity are involved in choice of subject, lighting, angle, editing, etc. AI does a sort of shopping-with-keywords exploitation and mashup of stolen photos and artwork and their descriptions - often grotesque mashups - and the AI user and pretend artist can generate images of something with just a name, like a Latin species name, without having any idea what that object looks like, as long as the name and related image(s) were ripped off for the AI's training data.)

(and so forth . . . )

https://www.democraticunderground.com/100220304330