Eugene Goostman is a 13-year-old Ukrainian who enjoys reading Kurt Vonnegut novels. He doesn’t speak the best English, but is very eager to make conversation.
According to about 20 percent of the people he’s spoken with, he’s not a computer.
Those 20 percent are wrong, though – Eugene is an artificial intelligence entered in the 18th Loebner Prize.
The competition, designed to pit AI against human intelligence, is a mass Turing Test. In a Turing Test, computers try to fool humans into thinking they are conversing with another human.
Named after Alan Turing, these tests are the ultimate determinant of a computer’s ability to reason – and, perhaps in the future, to create.
Turing believed computers would be able to fool 30 percent of humans into believing that they weren’t computers by the end of the century, which was the mark set for the Loebner Prize.
No program entered achieved that mark, but one came very close – 25 percent of all the human judges were fooled by a program called Elbot.
Anybody with a Web browser can experience the odd feeling of conversing with a machine at www.elbot.com, but don’t expect to have a deep philosophical conversation with your box. He’s able to answer basic questions, but doesn’t do much abstract thinking.
That is the eventual goal, though – computers need to be able to reason with abstract concepts at a human level to be accepted as truly "intelligent."
Right now, machines can’t function outside their databases – think of a politician whose script doesn’t cover every issue that may come up.
Similarly, they respond to things they don’t understand by asking questions and changing the subject. Some are subtler than others, but in the end, they aren’t going to be any good in a debate any time soon.
Humans have an amazing ability to piece together all sorts of information, which often doesn’t have any readily available connection.
When computers can call Cheddar "sharp," or when they think music can sound "flat" they will be able to string together disjointed ideas and bring coherence from chaos and converse on a genuinely human plane.
Humans have questions posed by the ancients that still go unanswered – computers can’t imagine the question, let alone the answer.
And why should they? There’s no logical reason for philosophy or emotion, both of which are crucial to the human experiment.
Then again, who is to say humans need artificial intelligences? After all, we got along just fine without them for millennia, right?
Well, yes and no – we got along just fine, but it wasn’t as comfortable or as convenient as today’s lifestyle, and there was far more strife.
Take, for example, a senior citizen home alone, with no family anywhere nearby. An artificial intelligence could keep this person company and monitor them – informing family members when something happens that needs human attention.
If Turing’s predictions are right and computers blend in with humans as seamlessly as is hoped, there is going to be a whole new society of silicon neighbors, and all the rules of personal interaction might change.
When machines can build themselves, humans will become second-class citizens in our own world.
I do want something to talk to, though. I’ll call up Deep Blue and get a game of chess going.
