In mythology, they had a priesthood manage the oracle at Delphi, and granted access to consult it only under certain conditions. The risk isn't AI, it's the peo

In mythology, they had a priesthood manage the oracle at Delphi, and granted access to consult it only under certain conditions. The risk isn't AI, it's the people with AI, and you need a better filter for people to manage it or you're going to trigger a war between people made subject to it, and those whose beliefs have been seduced by it. If I were to make a prediction, it's that the coming conflict will be between people whose ontologies can be simulated using language models, and those whose can't, and will not be. What's the difference between a political antagonist whose experience of the world can be simulated by machines and even wielded by them - and a zombie? The quotes in the article were concerning, as the tropes they used showed a lack of sophistication. I don't think they were well equipped. If I were hiring the red team for MSFT, I would also use what the team represented and not what they were capable of, as it's not an engineering problem, but a political legitimacy one. Twitter users using DAN have shown what the hacking/engineering approach entails.