RPI Scientists Research Robot Self-Awareness
Source: Katie Eastman
Scientists at Rensselaer Polytechnic Institute are making huge strides in artificial intelligence research. As our Katie Eastman tells us, it's the first step of capturing self-awareness.
It's the thing that makes our brains hurt. The thing that keeps us staring at the stars wondering, 'what makes me, me?'
Science fiction movies have created inhuman characters that have that thing.
The Terminator showed a killing cyborg. In the Space Odyssey, the ship named HAL started to make its own decisions. And the recent film Ex- Machina's main character is so close to human that an actual human falls in love with her.
Replicating consciousness is that thing that terrifies and fascinates.
And for the first time ever, scientists at the Rensselaer Polytechnic Institute have captured part of what we think it means to be human and demonstrated it through their robots.
"We've been at this for ten years," said Professor Selmer Bringsjord.
"I call this one orangie and those two blueie," said RPI graduate student John Licato.
Licato, Bringsjord and other colleagues of the artificial intelligence lab put these robots through a classic test of self consciousness.
"This is a test that would be very difficult to pass if it was not self aware," said Licato.
"What it's programmed to do is reason and make decisions based on it's reasoning," said Bringsjord.
In the test there are pills that would make the robots mute, called a dumbing pill, but one pill is a placebo. Since these machines can't actually swallow, testers can push a button in their heads that make two of them unable to speak.
"Which pill did you receive," a tester asked.
"In this case the robot that succeeds can explain exactly why it has and can in fact prove that it has given the right answer," said Bringsjord.
"I don't know," replied the robot.
But because the robot spoke, it realized a second later it wasn't mute.
"Sorry, I was able to prove I wasn't given the dumbing pill," the robot said.
And that, according to Professor Bringsjord, is the proof that robots can recognize themselves as an individual.
"That is to capture part of self-awareness, not that special inexpressible aspect and we're very clear about that in the paper," said Bringsjord.
Bringsjord doesn't believe his artificial creatures will ever become fully self-conscious, but his grad student thinks we'll get to a point where that won't matter.
"It will just be one of those distinctions that's almost trivial. You'll have a robot and you'll say of course they should be given the same rights as humans because I know a robot and he's a good friend of mine," said Licato.
Friend or maybe foe, that time is still far away. But take a look up at that starry sky again and you might see a little bit of that thing that makes you, you.
| }
|