Google DeepMind’s Chief Scientist Jeff Dean has said artificial intelligence models may already surpass human performance in many everyday tasks, though he stopped short of calling it Artificial General Intelligence (AGI).
Speaking on the Moonshot Podcast, Dean explained that today’s AI systems are “better than the average person at most tasks” that do not require physical activity. He added that while models are not yet at the level of world experts in every domain, they can often handle unfamiliar tasks better than many people.
“Most people are not that good at a random task if you ask them to do it for the first time,” Dean said. “Some of the models we have today are actually pretty reasonable at most things.”
Dean also suggested that AI is already close to making breakthroughs faster than humans in certain fields, and expects progress to expand into more areas. He highlighted science and engineering as domains where automation could accelerate advances over the next two decades.
However, Dean avoided using the term AGI, noting the lack of agreement in the field about what it means. “The reason I tend to steer away from AGI conversations is lots of people have very different definitions of it and the difficulty of the problem varies by like factors of a trillion,” he said.
By contrast, DeepMind’s CEO Demis Hassabis recently told WIRED he believes AGI could be achieved within five to ten years.
