Tuesday, February 22, 2011

A Sampling of IBM Watson News over the Weekend

[Cross-posted on ST-AIRC Blog]

IBM has inked a deal with Nuance to apply Watson's Jeopardy-winning Deep Question Answering technology to health care.

IBM’s Watson Jeopardy Stunt Unleashes a Third Great Cycle in Computing

NY Times' Stanley Fish -- not very impressed with Watson

Stephen Baker, author of “Final Jeopardy: Man vs. Machine and the Quest to Know Everything” on the impact of IBM's Watson

TED-cast by Stephen Baker

Friday, February 18, 2011

More Follow-up From Watson

A sampling of four new articles following upon on Watson's victory in Jeopardy!

Professor Oren Etzioni, the director of the UW Turing Center, on the match:
It didn't surprise me that Watson won. In fairness there are some small things in the game that give it an advantage, like the time that it has to process -- it sees the clue instantly, it gets transmitted via text, and then it's told when it can buzz in. So it has a number of these small advantages and particularly last night it was often winning on time.

Watson as the beginning of a new kind of search engine?
Etzioni says he expects natural-language software to make a big dent in search applications over the next five years, although at the moment systems such as Watson aren't ready for 'prime time': he notes that Microsoft bought a natural-language processing company called Powerset in 2008 for US$100 million, "but you don't see Microsoft using it in any visible way". Kautz agrees that systems as broad and powerful as Watson could be available for general use "surprisingly soon. Let's say three to four years."

Ken Jennings on his Watson experience:
I expected Watson's bag of cognitive tricks to be fairly shallow, but I felt an uneasy sense of familiarity as its programmers briefed us before the big match: The computer's techniques for unraveling Jeopardy! clues sounded just like mine. That machine zeroes in on key words in a clue, then combs its memory (in Watson's case, a 15-terabyte data bank of human knowledge) for clusters of associations with those words.

Gary Kasparov on playing chess alongside - rather than against - a computer:
Having a computer partner also meant never having to worry about making a tactical blunder. The computer could project the consequences of each move we considered, pointing out possible outcomes and countermoves we might otherwise have missed. With that taken care of for us, we could concentrate on strategic planning instead of spending so much time on calculations.

Thursday, February 17, 2011

Coca-Cola's Secret Formula: its Brand

When an entrepreneur starts a new company, one of the first things he(*) worries about is how to protect his new idea. And usually the choices are two: file a patent, or keep it under lock-and-key as a trade secret.

(*-or she, but we'll keep the masculine only here)

The trade-off for patents is clear: in exchange for a 17-year monopoly, the inventor has to disclose his methodology (or other invention) to the world, so that others can build off his patent. Indeed, the entire patent regime is is like a giant Jenga puzzle, with almost each patent "built" in effect upon previously-issued patents.

Trade secrets are the opposite: the inventor makes a decision that protection of the idea would be better served by keeping it private. Although there are some legal protections for stolen trade secrets (for instance, claims for misappropriation or theft, or unfair competition law), but even if an inventor insists on a non-dislosure agreement, if the secret is valuable enough, he may not be able to recover full damages from a breaching party.(*)

(*-The breaching party, even if insured, may not have assets large enough to make the inventor whole, for instance.)

For years the gold standard in trade secrets was the formula for Coca-cola. Urban legend has claimed that the only copy was held in the vault at the SunTrust bank in Atlanta, and that only two Coke executives (at any one time) had access to the document.

But in 1979, a columnist for The Atlanta Journal and Constitution ran an article that identified a ledger book with a formula for "Merchandise 7x"; over the weekend, the popular NPR radio show "This American Life" re-awakened interest in the 1979 story -- and the formula that goes with it. It is currently a viral sensation on the web -- this year's Susan Boyle.

The reality is that the formula -- or something very much like it -- has been "out" in the public domain for years. Type in "coca-cola formula" into a search engine and you'll find a number of recipes that will probably come very close to the taste.

But after a century of making the soda, Coca-cola now relies on a different intellecual property tool to protect itself: its trademarked brand. And trademarks don't expire (if subject to continuous use).

A lot of drinks may taste like Coke.

But only one can call itself (legally) the "Real Thing."

Machines 2, Humans 0

Building on its defeat (through Deep Blue) of chess master Gary Kasparov in 1997, IBM's AI team was a big winner last night with Watson storming from behind to win for the third night in a row, and dominating lead total of $77,147 for the week, more than the two humans combined. Ken Jennings finished second with a three-day total of $24,000 and Brad Rutter was third with $21,600.

Watson seemed to be off-its-game early, trailing for much of the first half of the game. But perhaps it was a machine version of "Rope-a-Dope": when "Double Jeopardy" began, Watson answered 18 of the 29 questions, ensuring its win.

Jennings finished the evening with a message as part of his "Final Jeopardy"
written answer:
"I for one welcome our new computer overlords."

Some analysis of how Watson "heard" the answers.

What Watson's victory means: Link

And an interview with IBM engineer: Link

Wednesday, February 16, 2011

Man-vs-Machine (3)

In Day Two of Man-vs-Machine Jeopardy!, IBM’s Watson dominated much of the show, answering 13 of the first 15 questions correctly and leading with $ $35,734 to $10,400 for all-time money leader Brad Rutter and just $4,800 for Ken Jennings.

In fact, but for its answer on Final Jeopardy (when it answered “Toronto” even though the category was U.S. Cities, and the clue was “Its largest airport was named for a World War II hero; its second largest, for a World War II battle.” With its answer, Watson showed the slightest bit of, um, fallibility.), Alex Trebek may have had to “stop the fight”… as it is, Jennings and Rutter will have a final chance tonight.

As they say, check local listings…

IBM’s Watson team offered this analysis of the Final Jeopardy mistake from last night. Meanwhile, IBM’s General Counsel, Robert C. Weber, offered this view of why Watson matters to the legal profession:

And here’s a PBS Newshour piece on the development of Watson, which also covers some of the history of AI:

Tuesday, February 15, 2011

Man-vs-Machine (2)

From an upcoming film on Uncontacted Tribes in the Amazon:

Also brings to mind a recent best-seller, "The Lost City of Z" by David Grann, about one of the last "romantic" explorers in the early part of the last century: Percy Fawcett. Fawcett, having explored much of the Amazon, left in 1925 on a search for El Dorado -- the Lost City of Gold. He, his son, and a friend disappeared into the Amazon jungle, and were never heard from again.

The book chronicles a latter-day effort to retrace Fawcett's footsteps, which brings the author all-too-close to repeating Fawcett's experiences. The result is a compelling read.


IBM’s Watson was tied for the lead after the first day of “Computer-vs-Human” challenge on this week’s Jeopardy, including long-time champ Ken Jennings (who has won 74 in a row).

One of the questions Watson answered correctly:

Clue: "Iron fitting on the hoof of a horse or a card-dealing box in a casino."
Watson: "What is shoe?"

But in other areas, Watson struggled:

Clue: "From the Latin for end, this is where trains can also originate."
Watson: "What is finis." Confidence level: 97%.
Trebek: "No. Ken?"
"What is terminus," Jennings answered correctly.

It will be interesting to see what the cultural impact of a Watson victory will be, especially in comparison to IBM's Deep Blue vs. Kasparov in 1997.

Watson is not connected to the Internet, and although Alex Trebek (and others) repeat that fact, it appeared that Watson did best when the answers were those that would be found by a search engine:

The questions that it did best at are ones that if you entered into Google or Bing, you can get the same answers. For instance, if you input one of the questions asked in the Jeopardy! tournament into Google, "Bang, bang, his silver hammer came down upon her head" one of the first results is "Maxwell's Silver Hammer" which Watson correctly answered. You get the same results with Bing. It's as if Watson is using the same sort of search algorithms, except not culled from the Internet, but a manually compiled, ginormous database of song lyrics, history, literature and other concrete, indisputable bits of information.

Computer-vs-human Jeopardy continues tonight (2/15/2011) and Wednesday (2/16/2011).

Friday, February 4, 2011

Bollywood's Vision of a Robotic Future

[Cross-posted on ABA ST-AIRC Blog]

Last fall, Bollywood (India's version of Hollywood; by some measures (for instance, number of films produced per year, it is larger)) released its most expensive film ever: "Robot" (or "Endhiran"). Great special effects, and -- it appears -- a "Terminator"-type story-line. Here's the official site.

Here's the official trailer:

And here's a compilation of the "best" scenes involving robots, spliced together (overdubbing is believed to be Russian):

Why Does Japan Emphasize Robotic Advances?

[Cross-posted on ABA ST-AIRC Blog]

From the BBC, interesting article about why Japan has taken such an active role in promoting "humanized" robots -- necessity being the mother of invention. And why some of the elderly in Japan prefer the human touch:

No, Robot: Japan's Elderly Fail to Welcome Their Robot Overlords

By Michael Fitzpatrick BBC News, Tokyo

"In Japan robots are friendly helpers not Terminators. So when they join the workforce, as they do often in factories, they are sometimes welcomed on their first day with Shinto religious ceremonies.

But whether the sick and elderly will be as welcoming to robot-like tech in their homes is a question that now vexes a Japanese care industry that is struggling with a massive manpower shortage. Automated help in the home and hospitals, believe some, could be the answer. A rapidly aging first world is also paying close attention to Japan's dalliance with automated care. It wants to know whether it can construct the nursing-care and medical-care needed in a future with fewer younger people to take care of the elderly.

Japan could show us how.

"The country sees it as an imperative to build carer robots and systems that can monitor health in the home. Because without them the nation's health care system won't cope," says carer Yasuko Amahisa.