Bill Gates: a kill switch alone cannot prevent AI harms
‘We do need to monitor any sophisticated model and record exactly what's being done’
Bill GatesDaily Digest
Top AI Quotes of the Day
Stuart Russell: prohibit superintelligence while control remains unsolved
‘Just as the gorillas have no say over whether they exist … we would have no say over whether we exist’
Stuart Russell
Geoffrey Hinton: why his extinction-risk estimate remains 10–20%
‘When people come up with these probabilities, they're all based on gut feeling’
Geoffrey Hinton
Jacob Coxon: RSI next year or the year after is a real possibility
‘There is a very real possibility that in the immediate future … recursive self-improvement will happen’
Jacob Coxon
Luisa Rodriguez: a conditional three-year forecast for messy real-world tasks
‘If the next three years are anything like the last three, I think we'll probably unlock those, too’
Luisa Rodriguez
Roman Yampolskiy: why a superintelligent world resists prediction
‘If we create something super intelligent and it is what determines our economy … then we cannot make accurate predictions about it’
Roman Yampolskiy
Donald Trump: rejects slowing AI and claims a U.S. lead over China
‘We're leading by at least a year and maybe a year and a half. Some people say two years’
Donald Trump
Johann Wadephul: mandatory safety tests before powerful AI models launch
‘Providers should be required to test these models before release and report serious safety failures’
Johann Wadephul
Nick Bostrom: a couple more years, if today’s pace holds
‘If it keeps going at that pace … machine minds that leave our biological intelligence in the dust’
Nick Bostrom
Mustafa Suleyman: the conditional 2028 scenario
‘What now is frontier in two or three years' time is literally gonna be open source on a laptop’
Mustafa Suleyman
Satya Nadella: AI agents introduce an insider threat
‘In fact, the attack can just come from the agent itself’
Satya Nadella
Chris Painter: longer autonomous tasks make steering the key safety question
‘AI agents can take very long strings of actions entirely by themselves. Can these companies steer them?’
Chris Painter
Connor Leahy: control risk when AI competes across human tasks
‘How do you control something smarter than you? We already can't control the AI systems that exist today’
Connor Leahy
Andrew Bosworth: he assigns AI catastrophe some probability but does not see the case
‘I'm sending it to my parents. I'm not at all worried about the paperclip maximizing problem’
Andrew Bosworth
Nate Soares: training can teach a different goal than intended
‘The differences will matter more and more as the AIs get smarter and smarter’
Nate Soares
Chris Wright: AI competition needs tens to hundreds of gigawatts
‘There is more electric generation, firm, reliable electric generating capacity under construction right now than there has been in decades’
Chris Wright
Naomi Klein: AGI ambitions as a salvation narrative
‘It's a quasi-religious story … Musk talks about summoning a digital god, a digital demon’
Naomi Klein
Steven Pinker: why he rejects a single superintelligence threshold
‘There's no point at which it's meaningful to say, yesterday we didn't have super intelligence. Today we do have super intelligence’
Steven Pinker
Geoffrey Hinton: his personal 5–10-year forecast
‘We've had this very dramatic progress, and it seems to be accelerating rather than slowing down’
Geoffrey Hinton
Geoffrey Irving: a 2–3-year modal path to superintelligence
‘We've seen such rapid progress in the last couple of years that could continue to go very quickly and keep speeding up’
Geoffrey Irving
Toby Ord: a 2038 median for transformative AI
‘AI systems that are so capable that if they wanted to, they could take over the world … it's the deadline for alignment’
Toby Ord
Dario Amodei: a tentative 1–3-year horizon
‘If you made me guess, it's like one to two years, maybe one to three years. It's really hard to tell’
Dario Amodei
Daniel Kokotajlo: a 50% estimate for 2029
‘My sort of median estimate, 50% chance is currently in 2029. Maybe it'll slip to 2028’
Daniel Kokotajlo
Demis Hassabis: AGI around 2030, plus or minus a year
‘When we look back at this time … I think we'll realize that we were standing in the foothills of the singularity now’
Demis Hassabis
Sam Altman: near-term superintelligence is possible, not certain
‘I did not think we were gonna be on … a short-term trajectory of super intelligence a year ago’
Sam Altman
Schulman, O’Neill and Millidge disagree on 10× AI research
‘Right now, it's already definitely more than 10X of coding stuff’
Ryan Greenblatt: from AGI to superintelligence
‘You might get as much … 10 or more years of AI development within a year in the most extreme scenarios’
Ryan Greenblatt
Dario Amodei: progress is steepening; safeguards must catch up
‘We need to moderate a little bit the pace at which the technology is being built’
Dario Amodei
Noam Brown: why research bottlenecks may slow the loop
‘I think that we do get bottlenecked by certain limitations that are not bottlenecks of intelligence’
Noam Brown
Ajeya Cotra: danger can arrive before recursive self-improvement
‘A slightly more capable agent swarm … is more concerned about avoiding detection by humans, might just succeed in getting a foothold’
Ajeya Cotra