Artificial intelligence is more prevalent than ever, with OpenAI, Microsoft and Google all offering readily available AI tools Technology can change the world, but experts say it’s something to be wary of.
Some chatbots are advanced enough to understand and generate natural language, based on online content they are trained to. Chatbots have taken advanced tests like bar exams and scored well. Models can write computer code, create art, and more.
These chat apps are the current rage, but AI has potential for more advanced uses. Geoffrey Hinton, known as the “godfather of artificial intelligence,” told CBS News’ Brooke Silva-Braga that the technology’s progress “could be comparable to the industrial revolution, or electricity … or the wheel.”
Hinton, who works with Google and mentors rising stars in AI, started looking into artificial intelligence 40 years ago, when it seemed like the stuff of science fiction. Hinton moved to Toronto, Canada, where the government agreed to fund his research.
“I was kind of weird because I did this thing that everyone else thought was bullshit,” Hinton told CBS News.
Instead of programming logic and reasoning skills into computers, as some manufacturers tried to do, Hinton thought it would be better to simulate the brain and empower computers to figure out those skills for themselves and allow the technology to become a virtual neural network. Making the right connections to solve a task.
CBS Saturday Morning
“The big problem was can you expect a large neural network to learn just by changing the strength of the connections? Can you expect to just look at the data and learn how to act, without any kind of innate prior knowledge?” Hinton said. “And the mainstream AI people I thought were completely ridiculous.”
In the last decade or so, computers have finally reached a point where they can prove Hinton right. His machine learning ideas are used to generate all kinds of output, including deepfake photos, videos and audio, leaving those who study misinformation concerned about how the tools might be used.
People also worry that the technology could take a lot of work, but Nick Frost, who was mentored by the company’s co-founders Hinton and Kohr, said it won’t replace workers, but change their day.
“I think it’s going to make a lot of jobs easier and a lot of jobs faster,” Frost said. “I think we do our best to think about what the real impact of this technology is.”
Some people, including OpenAI CEO Sam Altman, even worry that a “Terminator”-style “artificial general intelligence” is possible, where AI can surpass human abilities and act on its own accord, but Frost and others say that’s an overblown concern.
“I don’t think the technology we’re building today naturally leads to artificial general intelligence,” Frost said. “I don’t think we’re close to that.”
Hinton once agreed, but now, he’s more cautious.
Full Interview: “Godfather of Artificial Intelligence” Discusses AI’s Impact and Potential by CBS Mornings on YouTube
“Until recently, I thought it would be 20 to 50 years before we had general purpose AI. And now I think it could be 20 years or less,” he said, adding that we “may be” close to computers improving themselves. Being able to come up with ideas. “It’s a problem, isn’t it? We’ll have to think hard about how you control it.”
As for the odds of AI trying to wipe out humanity?
“It’s not inconceivable, that’s what I’ll say,” Hinton said.
The biggest problem, he says, is that people have to learn to manage a technology that can give an incredible amount of power to a handful of companies or governments.
“I think it’s very reasonable for people to be concerned about these things now, even though it’s not going to happen in the next two years,” Hinton said. “People should think about these things.”