Artificial intelligence (AI) expert Geoffrey Hinton has issued stark warnings over the development of chatbots after he quit his job at Google this week.
Hinton, whose research into neural networks and deep learning preceded the development of programmes such as OpenAI’s ChatGPT, warned it was only a matter of time before AI became more intelligent than humans.
“Things like GPT-4 eclipses a person in the amount of general knowledge it has and it eclipses them by a long way,” he told the BBC.
“In terms of reasoning, it's not as good, but it does already do simple reasoning.
“Given the rate of progress, we expect things to get better quite fast. So we need to worry about that.”
75-year-old Hinton left his job developing AI programmes at Google after a decade this week, he told the New York Times, explaining he could now speak freely about the dangers of the tech.
Though he warned it was “a nightmare scenario,” Hinton suggested “bad actors” could use AI for “bad things,” such as incorporating the programmes into robots.
Hinton added Alphabet Inc (NASDAQ:GOOG)-owned Google had been “very responsible” so far when it came to AI, but heated competition with Microsoft Corporation (NASDAQ:MSFT) over its Bing search engine could hinder this.
Google chief scientist Jeff Dean responded the company remained “committed to a responsible approach to AI,” adding it was “continually learning to understand emerging risks”.
However, Hinton, who also received the Turing Award in 2018 for his work on AI, said he no longer felt “the idea that this stuff could actually get smarter than people” was “way off”.
“I’ve come to the conclusion that the kind of intelligence we’re developing is very different from the intelligence we have,” he added.