OpenAI and the False Prophecy of “AGI”

I will be fascinated by how much feedback here splits tribally into griefers who are strong AGI proponents and don't like this write up, and maybe more my tribe who agree. I know, griefers is .. objectifying but I do expect pretty sharp retorts from pro-AGI people.

Culturally, publishing "manifestos" was more an artists pose, common from the 19th century onwards. But it's become something in niche corners in science too. Mostly in the "pledge for peaceful societal benefit" kind of sense. It's a word which gets up some people's noses. I suspect because of the big C manifesto which is a spectre hauting the world of manifesto. But also, it has America/Europe divide qualities: Europeans will happily spend 2/3 of the conference writing the manifesto which is agreed to in the plenary, The Americans on a tighter travel budget want to leave that to some sub-committee, do the real work, and move on.

If I felt AGI was close, I would be concerned we got the social dimension right. My friends from robotics say that its work well in hand, inside the field anyway. The Oxford Institute and like places go into ethics of CompSci Networks, AI, Expert Systems. We've had problems in the space since decision support systems started rei-ifying bad policy (medical school pre-admission checks was a big one 20+ years ago)

Manifestos tend to the mission statement side of things. Endless debates about every clause and semicolon.