Luciano Floridi issued a challenge to scientists to the world in 2005: prove that robots can display the human trait of self-awareness through a knowledge game called the “wise man” test. It was a venture he didn’t ever see being achieved in the foreseeable future.
A decade later, the Oxford professor’s seemingly unattainable challenge has been met.
On July 9, a team of researchers led by Professor Selmer Bringsjord helped a robot solve the riddle, displaying a level of self-awareness and satisfying what had until then been considered “the ultimate sifter” test that could separate human from cyborg.
But the professor says there’s a bigger challenge he wants robots to accomplish: self-awareness in real time. If we achieve this milestone, he said, the way we interact with artificial intelligence and robots will drastically change.
“Real time” self-awareness means robots acting upon new situations that they are not pre-programmed for, and translating how to act into physical movements. This is a serious challenge that Bringsjord has not tapped into because self-awareness algorithms are still separate from a robot’s body. If robots could work in real time, mind-to-body, he says, we would break through major barriers that could result in scenarios such as droids that act as our personal chauffeurs.
“Think about a self driving vehicle for a moment, say one that’s been built by Uber or Apple or Google,” Bringsjord told Business Insider. “Three people get into it, and they’re all talking at the same time. If the AI doesn’t understand who’s talking and that it’s not him or herself talking — but only recognises when itself speaks — there is no way to engineer a system that would be able to understand this conversation that good old fashioned taxi drivers have no trouble with. That is, unless the driving AI has a concept of itself versus its occupants.”
According to the roboticist, this feat requires programming and maths so complex that it has not been invented yet. So what’s the next step? Bringsjord said there’s a lot of low hanging fruit that needs to be grabbed in the near term.
“We’re going to see more robots coming on the market soon that can make an intelligent discrimination of its machine body versus the body of humans and possibly animals as well, ” Bringsjord said. “You’re probably going to see this the most happening in cars, controlled home environments, and office buildings.”
If we ever wanted robots to save lives — give a soldier emergency medication, for example — this “discrimination” skill is absolutely crucial, Bringsjord said.
Bringsjord said he welcomes others to come up with new self-awareness tests for robots.
“If we just design the tests ourselves, then everyone would be sceptical,” Bringsjord said. “It’s up for other people to give us challenges; we want to be challenged by companies like Uber.