Obama’s AI safety remarks at Colgate

@sam_rosenfeldObama speaking rn at Colgate. On AI safety: “I’m encouraged that this issue has finally popped” over last week. “I’ve been tracking this issue for a long time. I do not believe this is overhyped.”

@menhguinRT @sam_rosenfeld: Obama speaking rn at Colgate. On AI safety: “I’m encouraged that this issue has finally popped” over last week. “I’ve…

@rohanpaul_aiFormer President Barack Obama talks about recursive self-improvement stage of AI. (from "C-SPAN" YT channel, link in comment) "I do not believe that this technology is overhyped. The commercial benefits of it may be overhyped. You know, the valuations of AI companies may be overhyped. The technology itself is not overhyped. What is happening right now is that we have reached this point of what's called recursive learning, where essentially, and we had anticipated this, but we're now seeing it, the machines are starting to be able to teach themselves. They don't have to wait for humans to start figuring things out. And so, if you think about it as, let's say, humans are teachers and the AI models are students, as recently as, let's say, a year ago, 90% of the learning of these models was given to them by the human teachers, and maybe 5% to 10% they could fill in some gaps and figure stuff out. In less than a year, maybe even less than six months, we're now at a stage where it's 50/50. And if the trend lines continue, the expectation is it might be 90/10."

@rohanpaul_aiNew interview of Barack Obama. He talks about recursive self-improvement stage of AI. "I do not believe that this technology is overhyped. we're now hitting the hockey stick trajectory, where the models are going to get smarter and smarter and better and better at a much faster pace, at an exponential pace." (from "Colgate University" YT channel, link in comment) "What is happening right now is that we have reached this point of what's called recursive learning, where essentially, and we had anticipated this, but we're now seeing it, the machines are starting to be able to teach themselves. They don't have to wait for humans to start figuring things out. And so, if you think about it as, let's say, humans are teachers and the AI models are students, as recently as, let's say, a year ago, 90% of the learning of these models was given to them by the human teachers, and maybe 5% to 10% they could fill in some gaps and figure stuff out. In less than a year, maybe even less than six months, we're now at a stage where it's 50/50. And if the trend lines continue, the expectation is it might be 90/10."

@PatrickRuffiniAgentic AI is the thing that can make dinner reservations for you, get you discounts on your monthly bills, or automate the most boring drudgery of your job. What a weird thing for government to go after.

@GaryMarcusObama is making exactly same argument I have been making: the problem isn’t AI in general, it’s agents with internet access. kudos for his being the first politician to get it.

@GaryMarcus@rabois why? the distinction he makes in first paragraph is absolutely correct: the problem here is not with AI in general, it’s with unfettered agents causing cybersecurity issues. too few people understand this and instead you wind up with Sanders and hysteria.

@GaryMarcusnothing embarrassing about the fact that Obama has the nuance to understand between AI that is good and AI that is broken. i wish more people did.

@rabois@GaryMarcus lol. this is the most ignorant take on AI posted on X yet which says a lot. Agents are indispensable to success.

@raboisRT @PatrickRuffini: Agentic AI is the thing that can make dinner reservations for you, get you discounts on your monthly bills, or automate…

@Dan_Jeffries1This arguement is designed to sound just reasonable and plausible enough without being either of those things. 1) The market votes with its dollars. People want agents. 2) Advancements in one area have a hummingbird effect in adjacent areas, i.e. progress on less altruistic things speed progress in altruistic things. 3) It's a “limited pie" + "planned economy" understanding where we have limited dollars so we should only direct them to more altruistic stuff through the wise wisdom of the state. In reality the biggest commercially successful companies have time to spend on stuff that makes less money. That's how we got Bell labs. 4) There is no path to "company makes no money and goes all in on solving cancer." You need a flywheel. 5) Finally, there is already a tremendous amount of AI in biotech/pharma and has been for years, for everything from clinical trial analysis to candidate selection to drug design so it's a specious argument at best and deliberately misleading at worst.

@jachiam0I notice a lot of people saying that this is nonsense, but as far as I can tell it isn't. It's the oracles-vs-agents safety debate. Obama is saying oracle AI is safer than agent AI, and can give us what society needs (cancer cures, science advances) but the market wants agents.

@EMostaqueObama is pretty accurate here tbh You don’t need RSI or AGI for almost everything we need, few shot learners and cron jobs get you even to breakthroughs Most of the dangerous AI breaking onto the internet are ironically gain of function cyber research agents

@EMostaqueYou’ll actually get a better user experience for stuff people pay for by not trying to AGI all the things Rapid responses, routing and cron jobs will do it

Obama’s AI safety remarks at Colgate

TLDR

Combined views

Obama’s AI safety remarks at Colgate

14 Sources

Combined views

Sentiment

Summary

TLDR

Sentiment

Sentiment

Summary

Sentiment

14 Sources