Gizmodo: turing test

Computer Nearly Passes Turing Test for Artificial Intelligence

Jack Loftus — Sun, 12 Oct 2008 16:00:00 EDT

Today, the machines became a little smarter, as a computer named Elbot managed to achieve a 25% success rate when convincing a human being that they were talking to another human. The experiment is called the Turing Test, after mathematician Alan Turing, and Sunday's saw six Artificial Conversational Entities (ACEs) trying to ace the exam. Word is there was one human dunce in the mix, as all six computers managed to fool at least one interrogator into thinking they were speaking to another person, but none of the machines could officially pass Turing's strict standards.

The Turing Test states that to be considered "sentient," an artificial intelligence must achieve a 30% success rate. That means Elbot's accomplishment, while noteworthy, does not an AI make.

Organizer Kevin Warwick from the University of Reading's School of Systems Engineering was excited anyway, and readily compared today's events to the time in the 1997 when IBM's Big Blue defeated chess master Gary Kasparov. "This has been a very exciting day with two of the machines getting very close to passing the Turing Test for the first time," he said.

Perhaps even more impressive was just how believable the computers were, even if the human speaking with them knew they were speaking with a machine.

"Today's results actually show a more complex story than a straight pass or fail by one machine. Where the machines were identified correctly by the human interrogators as machines, the conversational abilities of each machine was scored at 80 and 90%. This demonstrates how close machines are getting to reaching the milestone of communicating with us in a way in which we are comfortable. That eventual day will herald a new phase in our relationship with machines, bringing closer the time in which robots start to play an active role in our daily lives."

Note to Warnick: Read anything on robots written by Isaac Asimov, then get back to me. [Telegraph]

Punk Rock Robots Rock Out to Punk Rock ONLY

Elaine Chow — Sat, 05 Jul 2008 14:00:00 EDT

Making robots even more hardcore, British roboticists have designed machines that will only dance to punk rock music. Standing 2 meters tall, padded in leather and decked in various punk scene insignias, the pogoing robots wait until they hear the familiar strains of anti-establishment rock before they start dancing. Is it just me, or do these things sound like they'd be a minor threat in the mosh pit?

The machines were designed by a collaboration of artists and scientists from Queen Mary University in London. They use neural networks, a collection of computer processors that function like a simple animal brain, to differentiate between Black Flag and say... Bob Marley. The result: robots that can dance to a genre of music even if they've never heard the song before. Check out the BBC link to see a video of the robots gleefully pogoing to The Fumadores. [BBC]

Spambots Can Now Fool Yahoo CAPTCHA Tests: Yes, Worry

Wilson Rothman — Wed, 30 Jan 2008 10:39:50 EST

You know those anti-spam tests that make you enter funny characters to prove you're a human? Well, non-humans can finally fake their way into systems using the "Completely Automated Public Turing test to tell Computers and Humans Apart" too—even Yahoo's pretty secure system, according to new reports.

A Russian security researcher known only as "John Wane" (sic) says that his team has developed a system that correctly identifies the images from Yahoo's CAPTCHA system 35% of the time. According to one analyst, the irony is that the image recognition used to fight off the current generation of image-embedded spam will now be used to create the next wave of spam itself.

Yahoo apparently confirmed that this was the case:

We are aware of attempts being made toward automated solutions for CAPTCHA images and continue to work on improvements as well as other defenses.

This doesn't just finger Yahoo, since the verification technique is used by other online e-mail providers too. In the words of the analyst, the hack "could be used for spam...could be used for phishing...could create a fairly significant number of e-mail accounts." I'm thinking this also means I'm screwed next time I want tickets for a concert, too. [TMCNet via Slashdot]

Anti-Spam Turing Test Is Really Global Human-Powered OCR System

Wilson Rothman — Tue, 02 Oct 2007 09:25:00 EDT

You know the test you have to take on Digg or Facebook, the one that proves you're a human? You see a hard-to-read word or string of gibberish, and you type in the correct characters. Carnegie Mellon researchers decided to replace randomly generated words with actual words from ancient manuscripts, words that machines are having trouble deciphering. When you or millions of other users type in a word, you are beating a machine and helping to preserve an irreplaceable text.

The original test is called the Completely Automated Turing Test To Tell Computers and Humans Apart, or CAPTCHA. This is CMU-originated modification is called reCAPTCHA. Instead of seeing one word, you see two, one that is already verified as correct. If you think about it, that's the only way the authentication could work. Both words are further distorted to fight spammers who may well have better OCR than the libraries.

Sites like Facebook and Twitter have already started using reCAPTCHA, and right now it's processing one million words per day. That's still chump change, though. According to Luis von Ahn, a professor at CMU:

"There's no danger of us running out of words. There's still about 100 million books to be digitized, which at the current rate will take us about 400 years to complete."

[BBC News]