Cambridge Union
8 HN mentions · mentioned in 3 months, 2023–2024
1:04:20 · 179K views · Nov 15, 2023
Sam Altman accepts the 2023 Hawking Fellowship at the Cambridge Union and speaks about OpenAI, AGI and AI safety, recorded November 1, 2023. It includes his comments on what is left for AGI and on discovering new physics.
During the Altman ouster threads, HN mined it for his recent unscripted statements on AGI and safety; some also mock its tone.
8 comments from Hacker News
Tinfoil? Hat time. Everyone from Greg to Sam to Ilya keep hanging on "AGI for the benefit of humanity". According to OpenAI's constitution: AGI is explicitly carved out of all commercial and IP licensing agreements, including the ones with Microsoft. Nuclear. Well apparently, the board decides what will be AGI. But 3 (everyone now left except ilya) of those board members don't even work at Open AI and are only privy to what the rest share. Yesterday, this is what Altman said, "On a personal note, like four times now in the history of OpenAI, the most recent time was just in the last couple of weeks, I’ve gotten to be in the room when we pushed the veil of ignorance back" [LINK] 2 days ago, Sam is asked about what's left for AGI. this video He says they can push Language Models much farther but that there are more breakthroughs required. But here's where it gets weird. He immediately starts talking about Super Intelligence and "discovering new physics". In No prior podcast 2 weeks ago with Ilya, he says Transformers can obviously get us to AGI. [LINK] To Ilya, they built AGI internally, and Altman wanted to release/monetize it early and tried to undersell it as being far away from AGI to the rest of the board. Ilya considers it AGI, or something extremely close to it, and deserving of far more caution, and so decided to convince the board that Altman was underselling the capabilities of their newer models and was risking a premature release of AGI. If true, then it could be seen
I agree. By tactical charisma, I didn't mean to imply that he has genuine empathy. I mean that he says things the other person finds pleasing, in just the right words, and credible-sounding seriousness. Tactical in a tempting sense: "Don't you want to be the bridge between man and machine, Scarlett?" or, "Imagine comforting the whole planet with your voice" — I've slightly rephrased a bit here, but this is how he tried to persuade Scarlett Johannson into "much consideration" (her words). Yes, I've listened to Altman. A most recent one is him waffling with a straight-face about "Platonic ideals"[1], while sitting on a royal chair in Cambridge. As I noted here[2] six months ago, if he had truly read and digested Plato's works, he simply will not be the ruthless conman he is. Plato would be turning in his grave. [1] this video [2] [LINK]
> I chose to ignore the possibility that it discovers something about how the universe that humans have not yet observed but my thought is that is a low probability outcome I think it's likely that it is precisely these sorts of discoveries that will augur the emergence of a superintelligence. Physics work is probably one of the first things that ML scientists will use to test advanced breakthroughs. As Altman said recently: "If someone can go discover the grand theory of all of physics in 10 years using our tools, that would be pretty awesome." "If it can't discover new physics, I don't think it's a super intelligence." this video
On lying: There's a great irony here. Altman apparently accepted[1] "Hawking Fellowship Award on behalf of OpenAI" at the University of Cambridge. I kid you not, sitting in a fancy seat, Altman is talking about "Platonic ideals". See the penultimate question on whether AI should be prescriptive or descriptive about human rights (around 1h 35sec mark). I'll let you decide what to make of it. [1] this video
I watched his recent award acceptance talk at Cambridge (on behalf of all of OpenAI), and in the questions section this came up. If I'm summarising correctly, his answer was that we can only learn how to make them safe by playing with these models while they're relatively small and poor. this video