The American government committed genocide against numerous native tribes, genocide is not complicated. China is actively committing genocide. I do think that an “unlimited terminal” will have less power than people imagine. But also things need to be structured so you don’t have to trust the person at the “unlimited terminal.”
If you accept that the safety angle is worthwhile, it’s very hard to tell where “don’t tell jokes about gingers” turns into “don’t put gingers in concentration camps.”
If I were on the board I would fire him for trying to commercialize a nonprofit. I am hoping that’s what happened but yeah I feel like it’s something else. Although it seems likely he has a financial stake in Microsoft that he’s been hiding.
I think most targeted models are going to be too targeted for anyone other than the organization that trained it to use. I can’t really see a 7B translation model being worth anything vs. GPT or Google Translate. (And for real translation, you really want something which can answer questions about connations/puns/whatever, which requires general understanding.)
I hope Microsoft gets some antitrust action on this, the whole thing seems like it was an antitrust violation, basically Microsoft acquiring OpenAI without doing it on paper to get around required regulatory approvals.