Reid Hoffman dismisses AI doomers: 'Look at me, I'm Cassandra'

LinkedIn cofounder Reid Hoffman Bloomberg/Getty Images Reid Hoffman says the existential risks of AI are "remote and controllable." Hoffman said instead of slowing down, frontier labs need to "steer" better.

What do you think about the AI safety debate? Reid Hoffman isn't here for your take about AI is going to kill everyone.

"I actually think most of that stuff, if you actually really think about it, is so like remote and controllable that it is mostly where people are kind of like, 'Look at me, I'm Cassandra, I'm saying it's the end of time," Hoffman said during Politico's Decoded Summit in Washington on Wednesday.

Hoffman said there are benefits to continuing frontier AI research, including how the technology can help address biosecurity and cybersecurity risks.

"I don't think that slowing down is important, because when you get to the defense against biorisk, the defense against cyber risk, and creating economic prosperity is a reason, I think, to keep going," he said.

The LinkedIn cofounder said it's a legitimate criticism to question why the leaders of Anthropic and OpenAI are now calling for a slowdown in frontier AI research at a moment when both companies appear to be leading the world.

"You're in the lead, you can slow down anytime you like," Hoffman said.

Instead of slowing down, Hoffman stressed that "it's really important" that frontier labs properly manage recursive self-improvement.

Some in the AI industry are concerned that recursive self-improvement, the process by which AI can learn to build itself and eventually its successors, could unlock a path to exponential growth that model makers might not be able to fully control.

Jacob Coxon, the former Anthropic researcher whose viral resignation helped catalyze the current conversation , wrote on Tuesday night that Anthropic "largely initiated" the race to recursive self-improvement.

Over the weekend, Anthropic CEO Dario Amodei called for frontier AI labs "to pace" their developments and to use the time to better align AI models.

SpaceX CEO Elon Musk and OpenAI CEO Sam Altman later expressed agreement with Amodei's broad aims.

In an essay he published on Wednesday, Hoffman said that Amodei and Altman's calls were steps "in the right direction." Not everyone is pushing for a unified response.

Meta CEO Mark Zuckerberg wrote on Tuesday night that companies can slow down individually without some sort of industry-wide pact that Amodei wants.