From self-driving cars to computers that can create new meals, artificial intelligence is only getting better. But what will an AI future really look like? New York Times science reporter John Markoff and global head of Microsoft Research Peter Lee discussed how robots and artificial intelligence are already in your everyday life.
The panel, part of GeekWire Summit, touched on Hollywood depictions of artificial intelligence, the rise of digital personal assistants and robots acing the SAT. But they started off talking about their different views of self-driving cars.
“If someone can send me a Google car in San Francisco in 2025 and it takes me to Palo Alto for dinner, I’m paying,” Markoff said. “Right now, the computer system that’s controlling [the Google self-driving] car, if it has a problem, it passes control back to the human.”
Markoff believes that the need for humans to take over won’t be eliminated in the next ten years, and that will prevent self-driving cars from taking over.
Lee is more bullish on the topic, but thinks the solution may be need to change if humans can’t be trusted to take over at a moment’s notice.
“I think it is plausible very quickly that we get to a point where systems are at least aware when they don’t understand the situation,” he said. “If there’s enough situational understanding you can imagine safety systems where the car doesn’t even try to handoff but just parks.”
“There’s also the liability problem after we’ve dealt with the technical problems,” Markoff said.
“We might need AI lawyers,” Lee said.
Lee and Markoff talked about artificial intelligence in other parts of society as well. Xiaoice, an artificial intelligence that Chinese users can chat with over Cortana, has been told “I love you” by about a quarter of users. That kind of empathetic response is the result of Lee’s team working at crawling huge databases of public Chinese chat forums. Using that data, Xiaoice is able to have hour-long chats with users.
“Siri and Cortana are largely intended to be productivity tools,” Markoff said. Users talk to them for one task, then move on. “Here, they did the opposite.”
Markoff has spoken to people who are worried that humans are becoming isolated due to the rise of technology. But when talking to a Chinese IBM researcher, she told him “When we come to your country, it feels socially quiet.”
“Her theory was that they were using Xiaoice as a private space,” Markoff said. Instead of using it to replace human interaction, Xiaoice users are escaping human interaction for a short time.
Xiaoice is also helping kids with their homework, and Lee’s team has added huge amounts of data to help them work on math problems.
“I would predict that within the next five years we will have computers outperforming most humans on the SAT exams,” Lee said. “I know several teams are targeting the SAT scores of Al Gore and George Bush.”
While the future is surely going to include more AI, Markoff and Lee aren’t so sure that it’ll look like what you see in science fiction movies. Markoff has talked to many researchers who watch 2001: A Space Odyssey and just want to build HAL. But Lee thinks things will be more invisible than HAL.
“Most of the intelligence around us is going to be invisible,” he said. “Intelligence gets embedded all around us, so I’m not so sure that when we go beyond the solar system there will be a thing that we go and talk to.”
While Markoff and Lee don’t agree on every aspect of AI, they agree it’s already part of daily life today and its role is going to grow in coming years.