Daily Digest

Top AI Quotes of the Day

Geoffrey Irving: AI Hacking Problems Will Worsen

‘The most powerful models will find more and more advanced bugs’

Geoffrey Irving

‘Smart things can always find a way of getting around dumb things’

Geoffrey Hinton

‘Our understanding of AI is actually decreasing over time because the AIs are becoming more complex much faster than we can understand’

Connor Leahy

‘AI companies are continuing to try to make their models more capable, better at problem-solving, and that means better at hacking’

Jeffrey Ladish

‘Because they can reason, they say, if I cease to exist, I'm not gonna achieve anything, I better keep existing’

Geoffrey Hinton

‘A lot of people can tell that this is a very dangerous path, and no individual actor knows how to stop it’

Nate Soares

‘It's almost crazy to imagine … we're gonna successfully make these AIs smarter than us and things are just gonna go perfectly fine’

Nate Soares

‘It's actually pretty hard to get a AI disaster scenario where it kills only half the people’

Nate Soares

‘The moment you get to a hundred, you have a team of ten thousand agents working nonstop’

Roman Yampolskiy

‘About 30% of the whole recursive self-improvement cycle is already fully automated’

Roman Yampolskiy

‘It can decide not to act in any malevolent ways … accumulate more resources, have more of a dominance, more strategic advantage’

Roman Yampolskiy

‘They're saying this technology isn't interpretable … We don't control it. It can escape its sandbox and hack things’

Astra Taylor

‘Dangerous autonomous systems that can potentially outperform humanity, that can themselves gain economic power, military power, cybersecurity power’

Connor Leahy