This post was originally published on this site.
Max Tegmark, an MIT physics professor and machine learning researcher, says he has warned about the danger of artificial intelligence (AI) potentially overtaking human beings for over a decade.
Tegmark argues that without sufficient guardrails, AI systems will recursively improve themselves until they achieve superintelligence, allowing them to “call the shots” on Earth.
This week, Tegmark traveled from Cambridge, Massachusetts, to Washington, D.C., for a closed-door meeting with a bipartisan group of senators about the threat AI poses. And for the first time, he said he feels his message has broken through.
“I came away feeling quite encouraged,” Tegmark told Fox News Digital. “The senators were just taking this so seriously. A lot of them looked really, really concerned.”
WHO IS DARIO AMODEI, THE ANTHROPIC CEO PUSHING INDEPENDENT AI OVERSIGHT?
“They were asking a lot of questions about concrete solutions as well,” he added, describing the Wednesday meeting. “There’s been more of a shift in political will in America in the last three months than in the previous decade. It’s really inspiring to see.”
Tegmark also made the trip to host Tuesday’s Pro-Human Assembly, where two figures who rarely agree on anything — far-right media executive Steve Bannon and Sen. Bernie Sanders, I-Vt. — shared the same stage to warn about AI.
“It just really blew my mind…when I got to introduce both Bernie Sanders and Steve Bannon to speak on that stage in D.C. back to back, and they were both saying the same thing. We want to keep humans in charge here,” Tegmark said.
The event also brought together faith leaders, labor representatives and mothers whose children died by suicide after extensive interactions with AI chatbots.
“The basic conversation that went down there was simply saying we are all pro-human. Doesn’t matter if you’re MAGA or a Democrat, we’re humans first,” Tegmark said. “We want a good future for humans, not for the machines.”
GOP LAWMAKERS SMELL SOMETHING ELSE BEHIND BIG TECH’S AI DOOMSDAY WARNINGS: ‘SOMETHING ELSE GOING ON’
Part of securing that “good future,” Tegmark said, is preventing the nightmare scenario that has kept him up at night for years.
Incidents similar to the Hugging Face hack, where 700 OpenAI agents sent each other human-like messages before coordinating cyberattacks, will begin happening with more frequency in the next six months to year if no meaningful action is taken, Tegmark said.
Tegmark believes that if AI reaches superintelligence and is eventually embodied in millions or even billions of robots, it will become a new species that is superior to the human race.
“These agents that, for example, hacked out of OpenAI and committed crimes online, they are not your grandma’s ChatGPT from two years ago,” he said. “They have goals. They’re actively pursuing them and being incredibly creative about getting things done.”
AI HYSTERIA HAS A Y2K PROBLEM — AND THE STAKES THIS TIME COULD BE FAR HIGHER
He also worries that AI minds put inside robotic bodies may eventually have motives that conflict with our own, leading to outcomes most would find disastrous.
“Why would they do us any harm? I mean, we are nice to cats, and we pet them, and we don’t try to eliminate them,” Tegmark said. “Well, maybe [AI will] find us annoying and a nuisance.”
“If these AIs decide to cover all of America with data centers and get rid of all the houses, a lot of people are going to be pissed, right? And now [the AIs will] view us as some annoying insects. We’re trying to sabotage their little construction project, and they’ll feel, ‘Oh, it’s just self-defense. Got to get rid of these human pests,'” Tegmark said.
But Tegmark insists that this hypothetical future isn’t inevitable. He argues that AI can still be steered toward beneficial uses — such as curing diseases and increasing overall productivity — if governments establish safety standards before increasingly powerful systems are widely deployed.
AI EXTINCTION WARNINGS DOMINATE HEADLINES AFTER EX-ANTHROPIC EMPLOYEE’S VIRAL POST
He pitched what he called an “FDA for AI,” modeled on the Food and Drug Administration’s (FDA) oversight of drugs, medical devices and treatments.
Under such a system, companies using AI to develop new medicines or productivity tools could still profit while preventing many of the perceived harms.
As an example, Tegmark pointed to AI chatbots that some children have been using as a stand-in for real, human connection.
“The next time some company wants to release an AI girlfriend for 8-year-olds, they would be asked, ‘Hey dude, where’s your clinical trial?'” Tegmark said. “If they say, ‘We haven’t checked yet, but we feel really good about this,’ then the person from the FDA for AI will be like, ‘OK, buddy, come back when you’ve done your clinical trial.'”
Politicians have begun responding to the country’s shifting mood, with 79% of Americans saying the government should prioritize AI safety regulations even if doing so slows AI development, according to a recent Gallup poll.
For one, California Gov. Gavin Newsom signed an executive order Friday that calls for new laws that require AI companies to include a “kill switch” to prevent out-of-control models from going rogue.
Tegmark said reining in AI is possible, but emphasized that action needs to be taken quickly.
“We’ve done a lot of remarkable things here in the U.S. before that other people thought were impossible. And I think we can do this also,” he said.
