If you’ve ever had to sign up for an account or buy concert tickets online, you may have seen those wavy, distorted jumble of letters and numbers. By typing what you read into a box, you prove you’re a human and not a bot. It’s call CAPTCHA, for “completely automated public Turing test to tell computers and humans apart.”
Now, an artificial intelligence startup in California called Vicarious claims to have created a computer algorithm that can solve CAPTCHA with greater than 90 percent accuracy -- including Google's reCAPTCHA, as well as CAPTCHAs from Yahoo, Paypal and CAPTCHA.com.
“An image is represented in the computer as a bunch of pixels with colors that are numbered on a scale of zero to 256,” Vicarious cofounder Dileep George explains to Businessweek. “This is what our system uses to then spit out a semantic understanding of those pixels.”
Creating machines that can make sense of images like we do is one of the “hard problems” in AI, Science explains. And since CAPTCHA is the security system used across the entire internet, breaking it means everyone should probably start transitioning to a new security system.
But their purposes go well beyond CAPTCHAs, Technology Review reports:
Rather than bring a product to market, Vicarious will pit its tool against more Turing tests, according to New Scientist, to create intelligence.
Vicarious hopes to eventually sell systems that can easily extract text and numbers from images (such as in Google’s Street View maps), diagnose diseases by checking out medical images, or let you know how many calories you’re about to eat by looking at your lunch. “Anything people do with their eyes right now is something we aim to be able to automate,” says cofounder D. Scott Phoenix.
Their “Recursive Cortical Network” -- virtual neurons connected in a network modeled on the human brain -- could be a huge breakthrough, but the company hasn’t offered any proof, Science reports:
Vicarious has credibility, given the scientists working there, but its current offer of proof is little more than a press release sent out to journalists and a video (above). The company has released no software code and no technical explanation, and… there are no current plans to write a paper.