Media

Wolfram/Alpha: Nerd-Powered Web Gadget of the Future?

The launch of Wolfram/Alpha could take us to an era of emerging artificial intelligence a self-organizing Internet -- or not.

After personal-computer technology took off in the 1980s, visions of artificial intelligence danced in pop culture's head.

Some brat in his bedroom pranked NORAD into War Games. The gamer reality of Tron booted up, while Hackers, Sneakers and other digital archetypes fought for advantage in The Matrix.

In the so-called real world, Microsoft, Yahoo and Google fought for primacy in the search game, where smart and dull surfers punched questions into their computers and waited for a response. Until now, the computer has only been able to communicate back using links, text, videos and ads.

But with the recent unveiling of Wolfram/Alpha, a computational knowledge engine powered up by physics, math and computer genius Stephen Wolfram, that communication and computational evolution has just accelerated. However, slowly.

"Our rather ambitious goal is to compute everything that is computable," explains Eric Weisstein, senior researcher in the Scientific Information Group at Wolfram Research, the company behind Wolfram/Alpha as well as Stephen Wolfram's groundbreaking computational software Mathematica. "We're only at the very beginning of the process, and we have ambitious plans for data, computation, linguistics, presentation and more."

Simply put, the query field on Wolfram/Alpha mashes input through its complex algorithms and heuristics and replies not with a set of links, like Google or Yahoo's search engines, but rather with stats, graphs and analysis.

Punch in "Lennon Lenin" and you'll get a breakdown of each cultural icon's age, place of birth and other comparative knowledge. Punch in "Microsoft" and "Google," and you'll get the current prices, financials, fundamentals, projections and more for both stocks. Get tougher and try to solve the integration Sin[(a-b) x]/2+Sin[(a+b) x]/2n, and you'll get enough math to give you a headache -- and an "A" on your homework.

In fact, if you ask Wolfram/Alpha the meaning of life, it will give you the only answer yet achieved: 42, from Douglas Adams' The Hitchhiker's Guide to the Galaxy.

But one thing Wolfram/Alpha will not give you is the type of general information the majority of the world expects when it asks its computer something. Punch in "Star Trek," and you get the crew and cast of J.J. Abrams latest film, and nothing else.

"As a search engine, it's weak," argues CNET's technology reporter Matt Asay. "But it's pretty interesting as a way to compute datacentric relationships between two things. Perhaps its biggest inhibitor is that it requires very different inputs from Google to be useful. You can't just 'search.' You have to have some idea of how to construct an inquiry."

In other words, it's for nerds.

"Wolfram/Alpha is not a search engine, and its functionality and goals should not be confused with traditional search engines," cautions Weisstein. "Search engines can only find information if it explicitly exists on a Web page somewhere on the Internet. Wolfram/Alpha can provide answers to infinite classes of questions based in its collection of facts and computational knowledge without them having to be written down somewhere. So it can answer many questions and perform computations that a search engine would have no chance to do."

This discrepancy between what the public currently does on the Internet and what it still can learn to do is damping down some of the unrestrained hype that accompanied the computational engine's debut in May.

"A new paradigm for using computers and the Web? Probably," Convergenceofeverything.com's Tom Simpson was quoted in The Independent's brazenly titled article "An Invention That Could Change the Internet Forever." "Emerging artificial intelligence and a step towards a self-organizing Internet? Possibly."

Of course, Simpson is probably right. If you throw in speech-recognition software, a talk-back program and even some sweet holographic projection, it's not too far a leap from Star Trek's own interactive computational knowledge engine. We're far from that momentous stage, but let's remember that personal computers were once for nerds, too. How ya like them now?

"Wolfram/Alpha does some things really, really well, like comparing stocks," Asay says. "But this type of inquiry comes prepackaged. What it needs to do a lot better is anticipate what people are trying to discover when they type words into it. To get something worthwhile out of it, you really have to coach it. Its biggest stumbling block is making the service easy to use."

Weisstein admits the learning curve for Wolfram/Alpha isn't exactly shallow, but he also blames the sorry state of knowledge computation on the Internet for part of it.

"In some sense, people are probably inadvertently trained to ask questions on the Internet in a form that search engines do well with," he says. "Using Wolfram/Alpha requires unlearning these habits. But it's pretty intuitive really."

Especially if you need to cheat on your math or science homework. "Actually, calculus students have already used the Wolfram Mathematica Online Integrator to help check their work for years," he adds, using the word "check" lightly, no doubt. "Wolfram/Alpha just greatly enlarges the scope of homework checking that can be done."

For now, at least, that kind of much-needed, most importantly free computational power will be Wolfram/Alpha's greatest achievement. If you're looking for the latest song, video or headline, then you must obviously look elsewhere. Probably for a while. But if you want to compute knowledge of any kind, from data sets and derivatives to finances and weather, Wolfram/Alpha is already a team to beat. Sure, it's riding on potential, but what isn't? Apple owned the world until Microsoft came along. Then Google owned the world until Apple came back with its iPods and iPhones. And so it goes.

"Google wins because it delivers great search results and is mind-numbingly easy to use," Asay concludes. "At this point, Wolfram/Alpha is still simply mind-numbing."

But as we venture into what that infamous American poet Donald Rumsfeld called the "unknown unknowns," that might change. Who knows?

We might someday boot up the retina-phone, ask for the climate, traffic, financial, entertainment and health reports and have them read back to us by a Wolfram/Alpha computational engine that sounds like Angelina Jolie or Brad Pitt. It's not beyond the realm of possibility, even now.

"It's been notoriously difficult to predict the future and relative importance of various Internet technologies," Weisstein says. "It has, however, tended to be the case that innovative and useful technologies thrive and even end up leading to more innovation not originally even dreamed up.

"Wolfram/Alpha is an ambitious, long-term endeavor intended to deliver increasing capabilities over the years and decades to come. There's nothing else like it out there, nor is there likely to be in the forseeable future."

Scott Thill runs the online mag Morphizm.com. His writing has appeared on Salon, XLR8R, All Music Guide, Wired and others.