Google's Brin Claims AI Works Better When Threatened

Google co-founder Sergey Brin shared his take on how to get the most from artificial intelligence (AI) during a recent podcast, All-In.

“AI models, not just ours, tend to do better when you threaten them with physical violence,” Brin said during the interview. “People feel weird about that. We don’t really talk about that. Historically, you just say ‘I’m going to kidnap you, if you don’t blah, blah, blah’.”

Brin’s response came from a question during the interview around calculations about Formula One driving, and the number of deaths that occurred per decade on the track. The AI model responded by suggesting the inclusion of practice miles.

Hosts of the All-In podcast, referred to as The Besties, are venture capitalists and entrepreneurs Chamath Palihapitiya, Jason Calacanis, David Sacks, and David Friedberg. During the podcasts they discuss a variety of topics, including economics, technology, politics, and poker. 

advertisement

advertisement

“The developments in AI are astonishing in comparison” to most advancements these days, Brin said, in a discussion of developments in smartphones and other technology. “The web’s breadth didn’t technically change so much from month to month and year to year, but these AI systems chance quite a lot.”

Google coders need to submit code to gain access to run basic experiments, Brin said, so that is something he did.

Post AI training, the thinking models, seem to be one part of the AI-stack that interests him most.

The exciting thing about AI, according to Brin -- although it’s not yet artificial general intelligence (AGI), which people are seeking -- is superhuman intelligence, but it’s pretty smart.

“The superintelligence comes in when the AI can do things in volume that I cannot,” Brin said. “If it sucks down the top thousand [search] results and does follow-on searches for each and reads them deeply, that’s a week worth of work for me and I cannot do that.”

The discussion turned to hardware, and the fact that Google had owned Boston Dynamics and bought or sold around eight robotics companies. The company built out the hardware. 

“Robots are cool, but the software wasn’t quite there to make them truly useful,” said Brin. When asked about humanoids, he said: “I’m probably the one weirdo who doesn’t believe in humanoids, but maybe I’m jaded because we acquired at least two humanoid robotics starts-ups and later sold them.”

 

 

Next story loading loading..