Moving Beyond the Squiggly CAPTCHA

posted by Stal on 2009.05.04, under Stal

In 1996, computer scientist Moni Naor wrote an unpublished paper proposing 9 types of tests to filter bots from humans on the internet. His tests included everything from natural language processing (fill in the blank sentences) to visual recognition (circle the eye on a picture, deciding whether a photo is dressed or nude).

If only his suggestions had been implemented, the internet might be a much more interesting place today. Instead, the use of CAPTCHAs (those boxes of distorted text) has been widely incorporated by websites, in part due to their ease of implementation by programmers. But we have all been stymied at one time or another by distorted text, and bots have been advancing their ability to bypass these sorts of text processing tests. Ideally, it would be nice not to have any sort of hoops to jump through at all, and a number of non-interactive tests have been proposed. The “honeypot” defense, for instance, relies on the fact that bots navigate the web by viewing it through code. If you include a form field that is visually hidden to the human user, anyone filling in the hidden entry is most likely a bot.

The problem is that computers will eventually have the ability to perform all sorts of tasks in the future. As laid out in the classic Turing test, we want to measure whether an entity behaves like a human, rather than simply measuring ability. From Slate:

How might it be possible to measure behavior rather than ability? The other day, I was writing a note to company using the online form they provided for media requests, doing the usual amount of typing, backspacing, and retyping as I tried to phrase my note in a way that would make them respond quickly. It occurred to me that the random, circuitous way that people interact with Web pages—the scrolling and highlighting and typing and retyping—would be very difficult for a bot to mimic. A system that could capture the way humans interact with forms algorithmically could eventually relieve humans of the need to prove anything altogether.

I’m a Human, Computer, I Swear! [Slate]

Human-Robot Jazz Improvisation

posted by Stal on 2009.05.01, under Stal

The stage debut of Shimon, the robotic marimba player, designed and programmed by Guy Hoffman at Georgia Tech. The robot takes cues from the human player to improvise and match the music’s tempo and themes.
[ Hat Tip: MW ]

IBM Computer vs. Ken Jennings

posted by Stal on 2009.04.28, under Stal

Researchers from IBM have announced that they are nearing completion of a program designed to compete against human contestants on Jeopardy. The program is based on three years of work with aid from experts in natural language processing, machine learning and information retrieval.

The real difficulty, Dr. Nyberg said, is not searching a database but getting the computer to understand what it should be searching for.

The system must be able to deal with analogies, puns, double entendres and relationships like size and location, all at lightning speed.

In a demonstration match here at the I.B.M. laboratory against two researchers recently, Watson appeared to be both aggressive and competent, but also made the occasional puzzling blunder.

For example, given the statement, “Bordered by Syria and Israel, this small country is only 135 miles long and 35 miles wide,” Watson beat its human competitors by quickly answering, “What is Lebanon?”

Moments later, however, the program stumbled when it decided it had high confidence that a “sheet” was a fruit.

Computer Program to Take on “Jeopardy!” [NYT]