Why Some AI Leaders Say Artificial General Intelligence Is Already Here

Artificial intelligence is still a relatively new technology, but one that has been seeing seemingly exponential jumps in its capabilities. The next big milestone many founders in the industry have discussed is artificial general intelligence (AGI), the ability for these machines to think at the same level as a human being. Now, some of AI’s biggest names say they believe we could already be at that point. The recent Financial Times Future of AI summit gathered Nvidia CEO Jensen Huang, Meta AI’s Yann LeCun, Canadian computer scientist Yoshua Bengio, World Labs founder Fei-Fei Li, Nvidia chief scientist Bill Dally, and Geoffrey Hinton (often referred to as the “Godfather of AI“) together to discuss the state of the technology. And some of those leaders in the field said they felt AI was already topping or close to topping human intelligence. “We are already there … and it doesn’t matter, because at this point it’s a bit of an academic question,” said Huang. “We have enough general intelligence to translate the technology into an enormous amount of society-useful applications in the coming years. We are doing it today.” Others said we may not even realize that it has happened. While most forecasts for the arrival of AGI still put it at several years down the road, LeCun said he didn’t expect it would be an event, like the release of ChatGPT. Instead, it’s something that will happen gradually over time—and some of it has already started. AI companies are generally less bullish on the subject of AGI than the panelists. OpenAI has said if it chooses to IPO in the future, that will help it work toward the AGI milestone. Elon Musk, last year, predicted AGI would be achieved by the end of 2025 (updating his previous prediction of 2029). Last month, he wrote in a social media post that the “probability of Grok 5 achieving AGI is now at 10 percent and rising.” Not all of the AI leaders said they felt AGI was here. Bengio, who was awarded the Turing Award in 2019 for achievements in AI, said it was certainly possible, but the technology wasn’t quite there yet. “I do not see any reason why, at some point, we wouldn’t be able to build machines that can do pretty much everything we can do,” said Bengio. “Of course, for now … it’s lacking, but there’s no conceptual reason you couldn’t.” AI, he continued, was a technology that had “a lot of possible futures,” however. And that makes it hard to forecast. Basing decisions today on where you think the technology will go is a bad strategy, he said. World Labs founder Li straddled the question, saying there were parts of AI that would supersede human intelligence and parts that would never be the same. “They’re built for different purposes,” she said. “How many of us can recognize 22,000 objects? How many humans can translate 100 languages? Airplanes fly, but they don’t fly like birds. … There is a profound place for human intelligence to always be critical in our human society.” Hinton, meanwhile, opted to look beyond AGI to superintelligence, an AI milestone where the technology is considerably smarter than humans. There are several startups exploring this space now, including Ilya Sutskever’s Safe Superintelligence and Mira Murati’s Thinking Machines Lab. “How long before if you have a debate with a machine, it will always win?” Hinton posited. “I think that is definitely coming within 20 years.” BY CHRIS MORRIS @MORRISATLARGE