Elementary, my dear Watson

June 17, 2010 in Data,Technology

In a pleasant surprise, the NYT Magazine has published an excellent article on artificial intelligence. What's more, it appears to be the first in a series. The article is well-written and accessible; it doesn't delve with any of the math, just the inspirations for and results of the AI procedures. It really speaks to the volume of attention being paid to advances in machine learning.

The article contrasts the "product of experts" semantic analysis that IBM's Watson does with the "database + statistical analysis" that Wolfram Alpha does, concluding that

[Watson] can analyze texts and draw basic conclusions from the facts it finds, like figuring out if one event happened later than another. But many questions we want answered require more complex forms of analysis. Last year, the computer scientist Stephen Wolfram released “Wolfram Alpha,” a question-answering engine that can do mathematical calculations about the real world....

But this sort of automated calculation is only possible because Wolfram and his team spent years painstakingly hand-crafting databases in a fashion that enables a computer to perform this sort of analysis...

All [Watson] will do is look for source material in its database that appears to have addressed those issues and then collate and compose a string of text that seems to be a statistically likely answer. Neither Watson nor Wolfram Alpha, in other words, comes close to replicating human wisdom.

Watson's "knowledge" is the result of an unsupervised learning process in which data was fed into the machine and the brain was left to draw its own conclusions. Wolfram Alpha's "knowledge" is the result of a number of explicitly connected facts that researchers "told" the computer could be combined in interesting ways - a supervised learning process.

I believe that a computer mimicking intelligent thought will develop first in an unsupervised manner, leaning more heavily on supervised "fine-tuning" as it matures, similar to a human developmental pattern (think about how children perceive the world and draw conclusions about how it works, eventually graduating into schools where curricula are more rigidly designed).

The only question now is how soon until Watson fits in your pocket? Apple has introduced a search product called Sherlock in the past; perhaps this would be the perfect complement.

Leave a Comment

Previous post:

Next post: