How soon will AIs surpass humans? – 04/15/2024 – Tech

A series of advances in artificial intelligence over the past year has raised expectations for the arrival of machines that can outperform human experts in a variety of intellectual tasks.

With progress in AI appearing to accelerate, there is a near-unanimous consensus in Silicon Valley that skills that seemed unattainable just a few years ago are now closer.

But asking AI researchers exactly when they expect to reach that milestone will yield more arguments than precise dates.

Elon Musk, the Tesla and SpaceX billionaire who is investing ever-increasing amounts of time and effort into AI, said this week that so-called artificial general intelligence (AGI) could be achieved as early as next year.

Often used as a shorthand for machines whose intelligence exceeds that of humans, AGI is a loose concept that is characterized by those trying to achieve it as a moving target rather than a fixed destination.

Musk’s statement has drawn derision from rival researchers, including Yann LeCun, Meta’s chief AI scientist, who argues that the arrival of AGI is decades away.

As the race intensifies to achieve AGI — and with it supremacy over a technology predicted to create trillions of dollars in value — see how the industry’s top names are betting on when and how it might arrive.

Musk has emerged as Silicon Valley’s biggest AI optimist. Last year’s improvements in large language models — the systems that underlie apps like ChatGPT — and the computing power available to power them led him to bring forward his AGI forecast, he said this week.

“My assumption is that we will have AI smarter than any human by the end of next year,” Musk said. By the end of this decade, AI’s capabilities will likely be greater than those of all of humanity combined, he added.

AI that goes far beyond the level of individual human experts is often categorized as “superintelligence” rather than AGI.

“AI is the fastest-advancing technology I’ve ever seen of any kind, and I’ve seen a lot of technology,” Musk told Nicolai Tangen, CEO of Norges Bank Investment Management.

The South African-born billionaire has consistently been one of the most optimistic voices in the technology sector, with his predictions — such as for the capabilities of self-driving Teslas — often proving overly optimistic.

Some see a financial incentive behind his latest comments. Musk — one of OpenAI’s founders who left after board conflicts — is trying to raise billions of dollars for his own AI startup, xAI.

Others in the AI ​​industry are skeptical of such an imminent breakthrough. But many of Musk’s closest rivals don’t think AGI is very far away.

Dario Amodei, co-founder of Anthropic, a prominent figure in the industry, predicted that AI could match a “generally well-educated human” within two to three years. However, he refuses to predict when AI could surpass human capabilities.

Future developments will likely depend on factors that are difficult to predict, including advances in machine learning, continued increases in computing power, and a favorable regulatory environment.

Demis Hassabis, co-founder of Google’s DeepMind, recently speculated that AGI could be achieved by 2030.

“When we started DeepMind in 2010, we thought of it as a 20-year project, and I think we’re more or less on the right track,” he said in a podcast hosted by Dwarkesh Patel in February. “I wouldn’t be surprised if we have systems similar to AGI in the next decade.”

OpenAI CEO Sam Altman has consistently resisted putting a precise date on AGI. He told an audience at the World Economic Forum in January that it could be achieved in the “reasonably near future.”

Its investors seem confident in its capabilities: eight-year-old OpenAI is now valued at more than $80 billion.

But when asked to estimate when AI could outperform an individual in various domains, Altman said, “I don’t know how to put a precise timeline on that, or what AGI means anymore.”

Altman, who in a post last year defined AGI as “AI systems that are generally more intelligent than humans,” may have his own reasons for shying away. Elements of OpenAI’s business relationship with its largest backer, Microsoft, will dissolve once this threshold is reached.

Microsoft’s investment gives it a license to use OpenAI’s technology until the startup’s board determines that AGI has been achieved, according to a lawsuit filed by Musk against the company.

While founders of Silicon Valley’s most talked-about startups predict AGI by the end of the decade, some of the industry’s most prominent researchers are more cautious.

“We hear a lot of people saying, ‘Oh my God, we’re going to reach AGI next year,'” said Meta’s LeCun, who has built his own highly powerful AI models. “Very prominent people say this and it’s just not happening… It’s a lot harder than we think.”

LeCun, one of the most respected figures in AI, has consistently downplayed the idea that AGI is coming soon. Far from threatening humans, he said last year that the latest models were no better at learning than a cat.

“The emergence of smart or super-intelligent machines will not be a one-time event. Progress will be continuous over the years,” he said. “It’s going to take years, maybe decades… The story of AI is this obsession with people being overly optimistic and then realizing that what they were trying to do was harder than they thought.”

Some experts are even more skeptical. Gary Marcus, a cognitive scientist who sold his AI startup to Uber in 2016, bet Musk $10 million this week that “we won’t see better-than-human AGI by the end of 2025.”

Marcus has previously written that AGI will be achieved, “possibly before the end of this century.” But, he said, today’s models are far from AGI and won’t be until they can fill gaps in “semantics, reasoning, common sense, theory of mind.”

Additional reporting by Cristina Criddle and Madhumita Murgia