08 June 2014

Computer Passes Turing Test

Yesterday at the University of Reading in London, a computer convinced human judges that it was actually a 13-year-old Ukrainian boy. By convincing one-third of the judging panel of its humanity, it became the computer ever to pass the famous Turing Test.
The Turing Test is a controversial test invented by Alan Turing in 1950. Turing believed that if thirty percent of humans could not distinguish a human from a machine in conversation, that would mean the machine is capable of "thinking." Until yesterday, a machine was never capable of convincing enough humans to be deemed artificially intelligent, though others have tried.
... the test was a five-minute keyboard conversation with someone or something on the other side. The questions are a free-for-all — no script is applied and there are no topics assigned in advance. It's meant to simulate a conversation with a complete stranger. The judges then determine if they believe they have been speaking to a machine or a human. As long as one-third of judges believe its human, the machine passes the test.
In 2012, a program nearly passed with 29 percent of judges convinced, but just barely missed the cut. Saturday's computer, who acted as a 13-year-old Ukrainian boy named Eugene Goostman, made the cut. Gootsman was created by a team of computer engineers, led by Russian Vladimir Veselov and Ukrainian Eugene Demchenko ... Goostman told the judges that he likes to eat hamburgers and candy, and that his dad works as a gynecologist.
The Turing Test has been a point of contention among researchers, and some argue that it is not a valid way to determine if computers can think. Professor Kevin Warwick, a visiting educator at the University of Reading, explained that this version of the test was actually quite vigorous: "The words Turing Test have been applied to similar competitions around the world [but] this event involved the most simultaneous comparison tests than ever before, was independently verified and, crucially, the conversations were unrestricted."
Goostman's age was crucial to passing the Turing Test. Developer Veselov explained that, "Our main idea was that he can claim that he knows anything, but his age also makes it perfectly reasonable that he doesn't know everything." So if the judges asked him something he was not programmed to know, judges might write that off as a factor of his age instead of his lack of humanity.
If you would like to talk to the first computer capable of "thinking," you can chat with Eugene here
The Wire

Perceptronium | A New Way To Look At Consciousness

... MIT’s Max Tegmark is championing a new way of explaining [consciousness]: he believes that [it] is a state of matter.
By “matter,” he doesn’t mean that somewhere in the deep recesses of your brain is a small bundle of liquid, sloshing around and powering your sense of self and your awareness of the world. Instead, Tegmark suggests that consciousness arises out of a particular set of mathematical conditions, and there are varying degrees of consciousness—just as certain conditions are required to create varying states of vapor, water, and ice. In turn, understanding how consciousness functions as a separate state of matter could help us come to a more thorough understanding of why we perceive the world the way we do.
Most neuroscientists agonize over consciousness because it’s so difficult to explain. In recent years, though, they’ve tended to agree that a conscious entity must be able to store information, retrieve it efficiently, process it, and exist as a unified whole—that is, you can’t break consciousness down into smaller parts. These traits are calculable, Tegmark says. A case in point? We put labels on the strength of our current computer processing power. While they’re not human, some of our computers can operate independently, and we can use our knowledge of artificial intelligence to push these machines to new limits.
Tegmark calls his new state of matter “perceptronium.” From the Physics arXiv Blog:
Tegmark discusses perceptronium, defined as the most general substance that feels subjectively self-aware. This substance should not only be able to store and process information but in a way that forms a unified, indivisible whole. That also requires a certain amount of independence in which the information dynamics is determined from within rather than externally.
So if consciousness is a state of matter, he concludes, we might be able to apply what we know about consciousness to what we actually see.
[...] the problem is why we perceive the universe as the semi-classical, three dimensional world that is so familiar. When we look at a glass of iced water, we perceive the liquid and the solid ice cubes as independent things even though they are intimately linked as part of the same system. How does this happen? Out of all possible outcomes, why do we perceive this solution?
In other words, quantum mechanics dictates that the world we see is just one of an infinite number of possibilities. But why? Tegmark doesn’t have an answer, but his ideas demonstrate that there might be a more dynamic relationship between consciousness and other states of matter—that our ability to perceive the world is both a means to an end and also an end (an “object”) in itself.
Read more here. | Abstract

07 June 2014

Bitter Clown Tears With A Hint Of Suspicion

A prankster replaced official descriptions on wine bottle displays at a Tesco store in Brixton, south London with his own irreverent versions.