Watson may win Jeopardy, but does it matter?

Con­tin­u­ing a short-lived fas­ci­na­tion with Skynet’s immi­nent takeover, recent­ly I saw that IBM has designed a com­put­er to win at Jeop­ardy. The asso­ci­at­ed com­pe­ti­tion begins air­ing, odd­ly enough, on Valentine’s day. And hey, that’s neat. This par­tic­u­lar man vs. machine skir­mish is a lot more approach­able than Deep Blue vs. Kas­parov. But there’s some­thing odd here…

The algo­rithms are backed up by vast data­bas­es, though there’s no active con­nec­tion to the inter­net — that seems like it would be cheat­ing, in Jeop­ardy terms.”

The gift­ed peo­ple at IBM know a lot more about arti­fi­cial intel­li­gence than I do. But this quote reveals some­thing that seems… sil­ly. Wat­son has no con­nec­tion to the inter­net. That may be a require­ment for this game show. But does it make any sense for the future of arti­fi­cial intel­li­gence, or com­put­er design in gen­er­al? Isn’t our future going to be con­nect­ed all the time, with most of the hard com­put­er work done in the cloud?

On the oth­er hand, it seems search browsers are com­pet­i­tive right now, and that’s with­out any PhDs adapt­ing their intel­li­gence to offer ques­tions in reply to answers. Google could be com­pet­i­tive with Wat­son, giv­en that it scores high­er than most humans, and would cer­tain­ly be com­pet­i­tive on any ran­dom night of Jeop­ardy. And Google isn’t even try­ing yet.

The real take­away? Your Search brows­er is prob­a­bly smarter than you. Already.

No Comments

Leave a Reply

Your email is never shared.