17 ms·
Hopefully MSFT just goes off and does their own thing. I find this holier-than-thou "we're non profit and we're going to save the world how we see fit" irritati
by cedws 3y ago
Hopefully MSFT just goes off and does their own thing. I find this holier-than-thou "we're non profit and we're going to save the world how we see fit" irritating. Talking about AGI like it's an animal ready to pounce when we're quite obviously nowhere near it. Machine learning development has been sluggish for the longest time and the moment it was commercialised, there was an explosion in possibilities and development.
"AI safety" is a euphemism for "aligns with our interests." "Roko's basilisk" is nonsense science fiction babble.
- SpicyLemonZest 3y agoDon't these events prove pretty conclusively that AI safety is not a euphemism for "aligns with our interests"? There's no way anyone at OpenAI could have expected to benefit professionally or financially from this.
- cedws 3y agoIt can also mean "does/doesn't do things as we see fit." They want to build a model without bias, but they've elected themselves as moderators.
- SpicyLemonZest 3y agoThis isn't about bias. Sutskever has recently been dedicated to what OpenAI calls "superalignment" (https://openai.com/blog/introducing-superalignment https://openai.com/blog/introducing-superalignment), the problem of future AI systems which don't do what humans want at all because they're too smart. You can dispute that research or its premises, but after it inspires a board coup at a billion dollar company, I don't think you can reasonably dispute that people believe it.
- adastra22 3y agoA lot of very smart people believe some very stupid things throughout history. Doesn’t make them right.
- fullspectrumdev 3y agoAI safety basically is bullshit FUD spread by people hoping to gain a monopoly through regulatory capture. Or people who are mentally deficient and actually believe in rokos basilisk and its ilk.