Wired 2.08: The Schank Tank

Issue 2.08 - Aug 1994

Page 2 of 6
previous | start | next
Printing? Use this version

The Schank Tank (continued)

SEE ALSO

Archive Category:

Computers

Robots & AI

6.02 - Feb 1998
Freeman Dyson's Brain
Stewart Brand talks to the deepest futurist alive - and the most trustworthy.
By Stewart Brand

5.12 - Dec 1997
The Brain Builder
AI researcher Hugo de Garis on evolvable hardware, asteroid-sized "artilects," and the issues of massive intelligence and species dominance that will roil global politics in the 21st century.
By Roderick Simpson

5.12 - Dec 1997
Pattie
MIT professor Pattie Maes has created a stir by making agents a household word, Firefly a $100 million business, and Newsweek's "100 most important people to watch" list.
By Marguerite Holloway

5.05 - May 1997
In Search of the Electronic Brain
For decades, Al programs haven't stacked up to 2 billion years of evolution. But, as one backgammon-playing bot proves,they're coming close.
By Michael Gruber

Though education software is not normally considered a subfield of AI, the new focus wasn't that big a switch for him. Schank first made his mark on AI in the 1970s at Yale by working on the problem of getting machines to understand written English, one of the most sought-after goals in the field. Later, Schank turned his attention to enabling computers to reason their way through brief stories, under the growing conviction that storytelling is our primary tool for making sense of the world and for sharing useful information. This offbeat but noisily advertised premise, along with his penchant for publicly needling other researchers' work and his personal eccentricities, earned Schank something of a reputation as one of AI's genuine characters. (Besides his somewhat Buddha-like appearance and loud, cranky enthusiasm, Schank also likes to flaunt his deep and apparently unmatchable appreciation of travel and fine food and wine.)

"He's always been ornery," recalls Danny Hillis, an AI scientist and founding scientist of supercomputer manufacturer Thinking Machines. "But he's ornery in a useful way." Robotics researcher Hans Moravec, who crossed paths with Schank at Stanford, remembers him simply as "a loudmouth with fairly interesting ideas."

In any case, Schank saw that educational software would give him a chance to apply his storytelling ideas in a useful -- if not theoretically profound -- way. After all, he thought, a good teacher would tell stories, so why couldn't a computer? Though educational software might seem a step down for a leading AI figure, the fact is that even Schank's most ambitious AI systems had always been considered modest technical achievements at best by most of the AI community; it is his arrogant panache that makes him seem larger than life to those outside the community. The transition from relatively simple AI systems to relatively complex educational software was not a particularly sharp one.

Meanwhile, in early 1989, executives from Chicago-based Andersen Consulting were meeting with Northwestern's board of trustees, rich with captains of industry, to bemoan the fact that the city lacked the sort of outstanding university computer science department that could change Chicago's economic landscape in much the same way as MIT did for Boston's Route 128 or Stanford for Silicon Valley. Northwestern, which has always suffered from an academic inferiority complex, jumped at Andersen's proposal for establishing a leading-edge department connected to the university. A search committee set out after a top name to run the proposed department, and feelers were put out to Schank.

Funny you should ask, replied Schank. He hadn't had much luck in persuading Yale to help him pull together the resources he needed to develop educational software, nor had he gone far with the two companies he set up on his own. What's more, his somewhat imperious behavior with colleagues, and especially with graduate students and junior faculty, along with occasional lapses in sound fiscal management (at one point Yale ended up holding the bag for a chunk of money Schank spent in anticipation of a grant that never materialized), left Schank feeling increasingly alienated at Yale.

Today, some 160 people at the institute are dedicated to Schank's vision of educational software. It is an eager, eclectic crew. Many are high-powered computer science PhDs enjoying what appears to be the best of both worlds: the steady funding of a corporate research lab combined with the relative intellectual freedom of an academic lab. "We're not constrained the way we were in industry," says Assistant Director of Research Ray Bareiss, a big, clunky guy with a soft drawl who used to work at a graphics start-up in Texas that eventually failed.

But like most university enterprises, most of the real work is done on the backs of graduate students and other very smart, very young people willing to channel atrocious amounts of energy into offbeat projects for which they will get only modest credit, and even more modest money. Typical is pixieish Carolyn Majkowski, a "content analyst" who figured out how to link up video clips and questions on the Army program. "It's like playing Jeopardy! backwards," she explains. "Now I know 5,000 military acronyms. I'm a joy at parties." It makes for an unpredictable mix at the institute; in the hallways, people in three-piece suits pass people in leather flight helmets.