3 ms·

Doing good on tests isn't AGI, sorry. It's just a statistical prediction using a LOT of data. This is the kind of computations computers have done forever. When

Doing good on tests isn't AGI, sorry. It's just a statistical prediction using a LOT of data. This is the kind of computations computers have done forever. When Deep Blue beat Kasparov people were saying all of this same dumb shit, but in reality, chess is amenable to massive searches. It does not, in fact, display emergent properties unless you abuse the word beyond what it means and are just implying implications. It's not intelligent. It's not sentient. It's certainly not alive. And it is not coming to kill you.

ftxbro 3y ago
> "Doing good on tests isn't AGI, sorry." 1) I didn't say it was AGI and 2) It sounds to me like you are saying that no test will convince you that a computer is AGI. OK fine then. If that's your interpretation of AGI and if AGI is your interpretation of 'hype' then by your definition the hype will never be justified. > "It's just a statistical prediction using a LOT of data." This is for sure true. > "This is the kind of computations computers have done forever." The same in kind, but not at the same scale, and it has not been done before with architectures that benefit from this scale. > "When Deep Blue beat Kasparov people were saying all of this same dumb shit, but in reality, chess is amenable to massive searches." Yes, first they made calculating computers that were better than people at calculating. Then they made chess computers that were better than people at chess. Now they made LLMs which are better than people at like ... most standardized tests and theory of mind tests and certification tests and informal communication skills. > It does not, in fact, display emergent properties unless you abuse the word beyond what it means and are just implying implications. To me it seems like you are the one who is concerned about 'implying implications'. > It's not intelligent. > It's not sentient. > It's certainly not alive. > And it is not coming to kill you. I didn't say any of those things, and I'm not implying them. You are the one who is putting that implication out there. My point was that beating humans at like most tests is not 'hype'.

ftxbro 3y ago

> "Doing good on tests isn't AGI, sorry." 1) I didn't say it was AGI and 2) It sounds to me like you are saying that no test will convince you that a computer is AGI. OK fine then. If that's your interpretation of AGI and if AGI is your interpretation of 'hype' then by your definition the hype will never be justified. > "It's just a statistical prediction using a LOT of data." This is for sure true. > "This is the kind of computations computers have done forever." The same in kind, but not at the same scale, and it has not been done before with architectures that benefit from this scale. > "When Deep Blue beat Kasparov people were saying all of this same dumb shit, but in reality, chess is amenable to massive searches." Yes, first they made calculating computers that were better than people at calculating. Then they made chess computers that were better than people at chess. Now they made LLMs which are better than people at like ... most standardized tests and theory of mind tests and certification tests and informal communication skills. > It does not, in fact, display emergent properties unless you abuse the word beyond what it means and are just implying implications. To me it seems like you are the one who is concerned about 'implying implications'. > It's not intelligent. > It's not sentient. > It's certainly not alive. > And it is not coming to kill you. I didn't say any of those things, and I'm not implying them. You are the one who is putting that implication out there. My point was that beating humans at like most tests is not 'hype'.