I’ve long been a fan of Douglas Hofstadter. His writings on creativity, math, music, consciousness and humanity (among other subjects) are often astounding. A lot of people get scared off at a quick glance, but his fascinations with what might seem maddening minutiae are mostly in service of a grander goal–figuring out nothing short of how humans think, and what it means to be self-aware. If that’s not interesting, well, you’re probably a fan of Lost.
Hofstadter came pirouetting through Thoughtland because of something I saw on this here Internet today. The NYT says that the little things called "captchas," those boxes with warped and obscured text you have to recognize and type into a little box in order to prove you’re human, are increasingly vulnerable to attack. (You’ll see just such a "captcha" at the bottom of this page, human.) In other words, computers are getting better at mimicking human abilities. Hofstadter and others have long been interested in that very thing, because computers mimicking human thought processes is, after all, the ultimate goal of the artificial intelligence crowd.
Enter Microsoft’s fairly inane new replacement for the text captcha, a series of pictures of cats and dogs which must be sorted into their proper categories (it said I was a bot–not sure what to do now). This presents a new problem for hackers. Though it’s possible to solve such a problem with lots of human effort, using it as an experiment in computer programming ought to provide for some interesting innovation. Perhaps an effort should be undertaken combining the efforts of hackers and AI scientists? Just a thought. A stupid one, probably, but a thought. Although I guess that’s pretty much what the NSA is.
How good will computers be at imitating humans after this one’s solved, as they all inevitably are? Do androids dream of electric sheep?
Is this a dog?
The world becomes more interesting all the time, despite the fact that the handbasket is getting mighty warm.
