Since the poem that rocked ChatGPT half a year ago, expert Gary Marcus has warned against the lightning-fast development and deployment of artificial intelligence.
But to AI’s apocalyptic prophets, a New York University professor emeritus told AFP in a recent interview that threats to the technology’s survival may now be “overstated.”
“Personally, I’m not too concerned about the risk of extinction, at least at this point, because the scenarios aren’t that specific,” Marcus said in San Francisco.
“The more general problem I am concerned with is … that the AI systems we are building are not very well controlled, and I think that poses a lot of risk (but) perhaps literally into existence. It will have nothing to do with it.”
Long before ChatGPT, Marcus designed the first AI program in high school (software that translates Latin to English), and after years of studying child psychology, he built a machine learning company, later acquired by Uber. Founded Geometric Intelligence.
“Why AI?”
In March, alarmed by OpenAI, the creator of ChatGPT, jointly releasing a newer and more powerful AI model with Microsoft, Marcus joined more than 1,000 people, including Elon Musk, in AI development. Signed an open letter calling for a global moratorium.
But last week, he did not sign a more concise statement by controversial business leaders and pundits, including OpenAI boss Sam Altman.
The signatories argued that world leaders should work to reduce the “extinction risk” from artificial intelligence technology.
The statement said that tackling risks with AI should be “a global priority alongside other societal-scale risks such as pandemics and nuclear war.”
The signatories also included people building systems with the goal of achieving “general-purpose” AI, a technology that retains human-like cognitive abilities.
“If you really think there’s an existential risk, why are you even doing this in the first place? That’s a pretty good question,” Marcus said.
Rather than focusing on fanciful scenarios where no one survives, society should focus on where the real danger lies, Marcus speculated.
“People may try to manipulate markets by causing all sorts of mayhem with AI, but then we’ll blame the Russians, for example, and say, ‘They Look what they’ve done to our country’,” he continued.
“This escalation[could]lead to a nuclear war or something like that.
threat to democracy
In the short term, psychologists worry about democracy.
Generative AI software creates increasingly convincing fake photos, and eventually videos, at almost no cost.
As a result, “elections will be won by those who are good at spreading disinformation, and it can be very difficult for them to change the rules and advance democracy.”
He added, “Democracy is predicated on having reasonable information and making good decisions. How can democracy proceed if no one knows what to believe?”
However, the author of the book Rebooting AI believes there are still “many improvements to be made” and thinks we shouldn’t give up hope.
As-yet-uninvented AI could definitely “help science, medicine, and care for the elderly,” Marcus said.
“But in the short term, I feel like we’re not ready yet. There will be some harm along the way and we really need to step up our game and come up with some serious regulation.” he said.
At a U.S. Senate hearing in May, Marcus, who sat next to OpenAI’s Altman, advocated the creation of a national or international body responsible for AI governance.
The idea is also endorsed by Mr. Altman, who has just returned from a tour of Europe, urging political leaders to find “the right balance” between security and innovation.
But be careful about delegating power to corporations, Marcus warned.
“The past few months have been a real reminder that the big companies that have the lead here aren’t necessarily interested in the rest of us,” he warned.
