Google as AI

For a while now I’ve been hearing that Moore’s Law will eventually produce real artificial intelligence, which many people expect will result in the Singularity. Nevermind that last part for now; I claim that the first part has already happened. I call it google. Whenever I say this in conversation or when speaking, most people go “pshaw” and change the subject. Maybe they’re thinking about AI like Asimov’s robots, autonomous and independent. Google isn’t that.

But look at this:

A crossword-solving computer program yesterday triumphed in a competition against humans. Two versions of the program, called WebCrow, finished first and second in a competition that gave bilingual entrants 90 minutes to work on five different crosswords in Italian and English.

Crossword software thrashes human challengers Tom Simonite, NewScientist.com news service, 31 August 2006

WebCrow uses a dictionary and looks in a database of known solutions, as well as using some heuristics. But it has a special sauce:
WebCrow performs a search using key words extracted from the clue. It can usually find the answer by looking at the small previews that appear with the search engine results, but it can scan whole pages if necessary. Words of the right length that crop up most often in the results are taken to be possible answers.
When DeepBlue beat chess grandmasters, that was considered a form of artificial intelligence. How is this different?

Sure, WebCrow can’t do many things humans do, such as drive, compose a speech, etc., nor can DeepBlue. But humans can’t do what WebCrow and DeepBlue can do, i.e., win at croswords or chess. And humans don’t all have the same abilities, either, whether due to nature or nurture. Some can do math; some can’t. Some can solve jigsaw puzzles in a jiffy; some can’t.

Google and Yahoo! and all the other search engines are merely conveniences, many people say; they don’t actually think. Yet how much of so-called human intelligence is fast recall and association of information? Having watched several people decline due to Alzheimer’s, I’d say a lot.

On a more practical bent, earlier today I was having a discussion with someone about the meaning of the word situation in 19th century warfare (where something is situatied, i.e., its geographical position) vs. its broader meaning today. In a few minutes with google I found appropriate quotes from Jomini (famous 19th century military strategist) to demonstrate that he did use it in the 19th century meaning, but he appeared to also already be using it in the broader sense. Could I have done the same thing by going to a library? Sure. Would I have? No; not to investigate one narrow point off the cuff. Here convenience matters enough that the task wouldn’t have been performed at all without it.

Sure, that’s a trivial example, but think of how many small examples like that go into your every workday. And how would like to do any of them without google? It would be like losing part of your mind….

Which was the plot of Vernor Vinge’s short story, “Bookworm, Run!” published in 1965. The protagonist was a chimpanzee, and the computer was a mainframe that provided memory and database access, but the idea was much the same: intelligence amplification.

So google isn’t a standalone AI. But what about WebCrow? And one of the main interests people have in AI is in the effects it will have on humans, so in that respect I’d say intelligence amplification counts.

There are potential risks and benefits of the Internet that we may not anticipate, nor even recognize even after they have already occured.

-jsq