Go to the Globe and Mail homepage

Jump to main navigationJump to main content

(Jason Logan)
(Jason Logan)

Ivor Tossell

Do easy digital answers put us in jeopardy? Add to ...

What a piece of work is man! How noble in reason; how infinite on Jeopardy. At least until someone invites a computer to play on the show. Then, there's hand-wringing.

We have come once again to one of those seminal weeks in human-computer competition. The battleground: Jeopardy, the long-running TV game show. The contest: a showdown between silicon and whatever it is that Ken Jennings's brain is made of.

Next month, Alex Trebek will host two former champions, including Mr. Jennings, in a match-up against an IBM-built computer called Watson. A preview is already available, and it doesn't look good for Team Organic: In a practice run, released online last week, the computer crushed its human opponents. Represented at its podium by a monitor with a pulsating little round icon, the computer sat implacably, knocking off answers in series like, well, a computer.

A drubbing is a drubbing, but I wouldn't call it great theatre: In 2011, a computer that answers trivia questions seems about as novel as a car that switches gears automatically. We already carry around wireless devices that augment our memories with trivia a-go-go, accessible as quickly as we can type. Mr. Jennings's saleable talent isn't that he knows so much; it's that he's faster than an iPhone.

For all that, Watson is hardly insignificant. Its ability to slice through Jeopardy questions reminds us not so much of a skill that computers are gaining, but of one the rest of us are letting slip away.

Trivia ain't want it used to be. Today, there's nothing Alex Trebek knows that Wikipedia doesn't, except perhaps the secrets in Alex Trebek's dark heart, and maybe even those. Evidently, answering trivia questions isn't a problem for computers - anything can be fished out of a database. Understanding the questions in the first place is another matter.

Watson's talent is the ability to make sense of Jeopardy's prosaic and punning categories and convoluted clues ("Potent Potables," "Things You Should Not Put In Your Mouth"). It's a natural-language machine, designed to figure out what humans are asking for. In a way, it's playing an entirely different game than the human competitors, for whom understanding questions is easier than memorizing an encyclopedia.

If it all seems a bit underwhelming, it may be because a more familiar artificial intelligence is already performing such tricks millions of times a minute: Google. Apart from the not-entirely-inconsequential feat of having indexed the total of humanity's digital knowledge and put little text ads next to it, Google routinely divines what its users are looking for, no matter how convoluted their queries.

It scans all sources; it catches misspellings and partial phrases. It recognizes that most people who search for "Shawn Connery" are really looking for "Sean Connery." It knew that, as I fumblingly searched for "What a noble thing is man," I meant "What a piece of work is man," which isn't the same at all. It promotes pages that are most likely to be useful and informative - even if, more often than not, that means a Wikipedia entry.

There was a time, in living memory, when there was even a knack to searching for information on the Internet. One had to have a feel for search terms, a knowledge of which site to visit for what purpose, a touch for finessing the sometimes-balky quirks of different search engines.

No longer. It's true that specialized digital research still requires specialized tools, but that's largely because the archives in question are in private databases, beyond search engines' reach. It's true, as well, that Google still has advanced search options, but when was the last time you truly needed to use them? Nine times out of 10, the delicate technique of modern searching comes down to "sometimes I use quotation marks and sometimes I don't."

In fact, Google and its peers have become so adept at returning the most useful links first that, without meaning to, I've fallen out of the habit of pursuing search results past the first page. Rightly or wrongly, I habitually put so much trust in Google's rankings that I've stopped thinking of Google as a research tool and started thinking of it as a question-answering machine. No longer do I "Google for information." I just "ask the Google" and go with what it tells me.

The Google, as it were, is increasingly answering with the same handiness with which Watson answers Alex Trebek. Not only has knowledge become something we've outsourced to the Internet; the strategies to search for that knowledge have, too. As information systems become more powerful, all the search strategies of yore are coalescing into one: "Just ask."

Is this shift a worrying one? The question of whether Google, as author Nicholas Carr puts it, "makes us stupid," is ripe for enquiry. I'm unconvinced. After all, many research strategies are only as useful as the systems they manipulate. The ability to navigate the Dewey Decimal System isn't much good in a world without card catalogues, just as a knack for AltaVista isn't all that useful today. And "just ask" is the exact approach one takes to the last great information system - the librarian.

The real question is whether tools like these, which readily answer the most obtuse of questions, will undermine the basic techniques of inquiry, and the ability to learn new skills where needed. We can hope. Only the Google knows the answer.

Report Typo/Error

Follow us on Twitter: @GlobeTechnology


Next story




Most popular videos »

More from The Globe and Mail

Most popular