On Tuesday, 27-year-old AI researcher Jacob Coxon started a Twitter thread that ignited an online firestorm and soon dominated headlines and cable news. “I resigned from Anthropic today,” he began. “I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.” Coxon went on to warn of the existential threat posed by AI systems, saying that the people working on AI “earnestly” believe that the technology could “kill us all” by the decade’s end.
By Wednesday night, he was explaining to CNN’s Anderson Cooper that what scares him isn’t what the models can do now but the idea of AI being used to “make itself more intelligent.” Geoffrey Hinton, the so-called “Godfather of AI,” was on BBC Newsnight the same day, saying that a 10% chance of AI causing human extinction is a reasonable estimate.
Perhaps most remarkable were the replies filled with people who work at major AI companies right now. A long list of them, including some with job descriptions containing the word “safety,” agreed with Coxon’s warning as if they had long known the dangers and had been holding it in. Which raises the question: what took you all so long? And just how bad is it?
Here are 9 top AI researchers and executives, including Coxon, who, over the last 72 hours, have come out of the woodwork to sound the alarm bells that critics of AI have long been ringing. Just look at what they’re saying!
Evan Hubinger, Anthropic, Alignment Science Lead
As the alignment science lead, isn’t it his job to lead the creation of the plan for alignment? Perhaps something more important was taking up his work hours.
Anna Wang, Anthropic, Technical Staff
Thanks for the warning, but how are you writing this in your personal capacity when you are referencing both your current and former employers in the first sentence? And don’t you bear any “personal” responsibility for this mess?
Jason Wolfe, OpenAI, Technical Staff
Literal extinction is possible, but he is “heartened” by a company that is in large part responsible for the possibility of said literal extinction.
Drake Thomas, Anthropic, Technical Staff
If he and his colleagues would burn their equity to the ground, do you think they will do so when Anthropic’s planned IPO rakes in billions?
Alex Turner, Former Research Scientist at Google DeepMind
Evidently, he never figured out how to stop that.
Geoffrey Irving, Co-founder and Chief Scientist at Resolution (previously worked at OpenAI and Google DeepMind)
Presumably, by “we will have to act despite uncertainty,” he means “we’re gambling all of humanity on a coin toss.”
Marcus Williams, OpenAI, Technical Staff, Safety Oversight
He works for a company that spends millions lobbying against greater AI regulation.
Andreas Kirsch, Google DeepMind, Research Scientist
Will this “get me into (more) trouble”? Yeah, because that’s what’s at stake here, right?
Zeteo’s Mohamad Rimawi contributed to this piece.
Check out more from Zeteo:
"Evidently, he never figured out how to stop that." Describing it would be a good first step.
I'm sorry to say it, but this article is simply blatant fear-mongering. Where's the specifics? How is it going to kill us in the next decade? More nuclear power? Capriciously shut off people's water and electricity? Robot police throwing us all in jail? What? Did these people all sign NDAs?
That one guy saying he'd burn his equity to the ground for a 1% improvement in humanity's chances. I have a lighter, buddy, if you want to get started. But please, tell us why.
The rich, the poor and everyone in between all knows that climate change and resource depletion is going to kill us all. Yet we continue to dig scrape and burn our way through the planet. Why on earth would be concerned about AI killing us all when all we really want is to book a faster flight or get that Amazon delivery quicker.