In an attempt to take the first step towards building a machine that works like a human brain, technology startup Vicarious claims to have created software with an artificial intelligence sophisticated enough to crack the Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA).

CAPTCHAs are short strings of text, numbers or symbols that are often used by websites as a means of authentication alongside a username and password. Invented by students at Carnegie Mellon University, it is a basic type of "challenge-response test," designed to determine whether or not the user is real or an automated bot.

Vicarious' co-founder, D. Scott Phoenix, said that the AI wasn't created for any malevolent or financial reasons; the San Francisco-based company wanted to prove to the world that it was possible and that "we are the best place in the world to do artificial intelligence research".

Vicarious providesa demonstration of its method on its website, which boasts cracking the CAPTCHAs of Google, eBay and Paypal, but doesn't go into significant detail as to how this was achieved.

Luis von Ahn, associate professor of computer science at Carnegie Mellon University and one of the developers of CAPTCHA, was less than impressed. He reminded Vicarious that CAPTCHA has been in use since 2000, and since 2003 there have been stories every six months with similar claims. One website provides 28 examples of apparent CAPTCHA hacks. According to Ahn, whilst text-based CAPTCHAs might be breakable, digitally distorted images can currently only be comprehended by humans.

Nonetheless Vicarious stands by its assertion that by leveraging core insights from machine learning and neuroscience, its AI achieves success rates of cracking up to 90 percent of modern CAPTCHAs. A CAPTCHA scheme is considered broken if an algorithm is able to reach a precision of at least one percent.

Vicarious's proclaims, "Solving CAPTCHA is the first public demonstration of the capabilities of Vicarious' Recursive Cortical Network (RCN) technology. Although still many years away, the commercial applications of RCN will have broad implications for robotics, medical image analysis, image and video search, and many other fields."

Similar AIs are currently in development, such as IBM's Watson. However, unlike Watson -- which uses a brute force approach by connecting massive computing power to massive datasets in an attempt to achieve similar AI goals -- Vicarious uses a "distinctively human act of perception", boasting that its algorithm requires miniscule amounts of data and computing power.

Dileep George, Vicarious' other co-founder, says the AI system is trained by showing it a series of letters and images: "It needs just a few examples of letters to learn about them. Previous work would require in the order 10,000 examples of a letter even to understand minor variations."

The company has many high-profile supporters, such as Facebook co-founder Dustin Moskovitz who said, "We should be careful not to underestimate the significance of Vicarious crossing this milestone," adding, "this is an exciting time for artificial intelligence research, and they are at the forefront of building the first truly intelligent machines." Other supporters and investors (which have helped Vicarious achieve over 9 million in funding) include Jaan Tallinn, founding engineer of Skype and, before that, Kazaa and Adam D'Angelo, co-founder of the online knowledge market Quora.

See the original post:

Artificial intelligence startup claims to have cracked CAPTCHA