Watson Wrap-Up: “Puny” Humans Bested by Big Blue

by Michael Mullaney on February 17, 2011

There’s quite a buzz this morning about the third and final installment of Jeopardy! The IBM Challenge. And with good reason: Our new favorite friend Watson handily defeated his carbon-based opponents, the two most-celebrated Jeopardy! champions of all time, Ken Jennings and Brad Rutter.

Of course, given that Rensselaer graduates at IBM led and worked on the Watson project for four years, there was little doubt here in Troy that the now-famous supercomputer would prevail.

If Tuesday evening was the challenge’s Empire Strikes Back, filled with pitfalls and peril, then last night’s game was undoubtedly Return of the Jedi. Determining the heroes and villains, however, was a matter of perspective. Underneath his Final Jeopardy answer, Jennings wrote “I, for one, welcome our new computer overlords.” In his mind, he was co-piloting the Falcon with Rutter but they epically failed to destroy the Death Star.

While I was sitting in EMPAC last night with 800+ students, faculty, staff, and alumni, it was clear the audience had a different idea. To us, Watson was the hero of the story, our Luke Skywalker. The newsclip above ran last evening on Channel 13, which gives a nice overview of the viewing party. Below is a great observation by Times Union blogger Kevin Marshall, who was also at EMPAC last night:

“Aspiring scientists and engineers in the crowd treated the contraption with an eagerness and endearment that, though sometimes masked through attempts at ironic humor, was clearly affection for the machine. It was all in fun and a good sign for those of us who recognize our country’s quiet crisis in the maths and sciences, but I still felt slightly disheartened that people were rooting for the inorganic to triumph over the human underdogs.”

Jennings, who played well on Wednesday and at one point posed a very real threat to Watson, later came around to our way of thinking. “I wasn’t the hero at all. I was the villain,” he wrote last night in an essay at Slate titled “My Puny Human Brain.” He goes on to pay the IBM Watson team a sincere compliment rife with geek cred. “Watson has lots in common with a top-ranked human Jeopardy! player: It’s very smart, very fast, speaks in an uneven monotone, and has never known the touch of a woman,” he wrote.

It was a treat to watch the panel discussion leading up to and following the Jeopardy! broadcast. You can watch the whole thing here. Rensselaer graduates and IBM Team Watson researchers Chris Welty and Adam Lally shared great insight and really connected with the audience. My favorite anecdote was about Watson’s betting strategy. On Daily Doubles and Final Jeopardy, the computer wagers highly particular numbers that initially seem (but are not) somewhat random. This caught the attention of many people in the early testing days of the system, prompting the researchers to tweak algorithms so Watson always rounded its bet to the nearest 100. After instituting this change, people said they missed the quirkiness of the odd bets, so the Watson team ditched the round number command.

Another favorite part of the discussion, which was moderated by Rensselaer Vice President for Research Fran Berman, was hearing Rensselaer professors Deborah McGuinness, Sanmay Das, and Selmer Bringsjord connect their own research streams to the technology behind Watson.

The panel remarked that Watson doesn’t truly understand the significance or meaning of Jeopardy! answers or questions, it’s more about making connections and probability. Well, McGuinness is a world leader in using ontologies to empower computers to semantically “understand” language. She mentioned her wine app, now available for download on iDevices, that uses personal preferences, info from the cloud, and other factors to help users pair a wine with their meal in real time. Looking forward, we should expect to incorporate more kinds of Watson-like assistance in everything from making medical decisions to choosing a good pinot, she said.

Financial modeling expert Das spoke about our natural inclination to mistrust computers, simply because they’re computers, but how Watson can help peel back those biases. This is a good thing, he said, as in a few years we’ll all be best buds with our cell phone, which will be able to speak with us and convey truly useful information.

Welty went one step further, and noted he expects our cars to be self-driven and operated in the near future, which will reduce the total number of accidents and make our roadways considerably more efficient.

Bringsjord, who was Welty’s faculty advisor at Rensselaer, is a global thought leader in the field of artificial intelligence. He agreed about the vast potential of Watson technology as a way to augment, but not replace, the human mind. Bringsjord said he’s doubtful that Watson or other computers will be able to become self-aware or pass the Ada Lovelace Test and sincerely surprise their programmers with original, creative thought. However, he said we should take every precaution to make sure AI and cognitive systems – particularly those used in defense applications – have stringent and robust failsafes to counter any possibility of computers acting in an evil or immoral fashion. This harkens to his well-established research thrust that calls for vigilance today to hedge against potential AI hazards in the future. Overall, Bringsjord was bullish on Watson and decidedly proud of his former students:

“It’s an amazing thing to ponder how fast [IBM], with its brilliant minds, has taken us from Deep Blue – at a time when people said, ‘Meh, [chess] is a solvable game on an 8×8 grid, where’s the language?’ And here we sit. Personally, it’s amazing to me, as I knew these three guys here, and I knew they were brilliant. But what they’ve done here, with their colleagues, is amazing.

All in all, if Ken Jennings were in the audience last night, I’m confident the panelists’ discussion would have quelled all of his concerns about “computer overlords.”

Below is some grist for the mill:

The Associated Press quoted Rensselaer Professor Jim Hendler in a story that appeared online and in newspapers across the globe this morning:

“A human working with Watson can get a better answer,” said James Hendler, a professor of computer and cognitive science at Rensselaer Polytechnic Institute. “Using what humans are good at and what Watson is good at, together we can build systems that solve problems that neither of us can solve alone.”

Additionally, Rensselaer graduate and Watson team leader David Ferrucci was quoted in the New York Times this morning:

Nevertheless, Watson, which took 25 IBM scientists four years to create, is more than just a trivia whiz, some experts say.

“I see human intelligence consuming machine intelligence, not the other way around,” David Ferrucci, IBM’s lead researcher on Watson, said in an interview Wednesday. “Humans are a different sort of intelligence. Our intelligence is so interconnected. The brain is so incredibly interconnected with itself, so interconnected with all the cells in our body, and has co-evolved with language and society and everything around it.”

“Humans are learning machines that live and experience the world and take in an enormous amount of information — what they see, what they taste, what they feel, and they’re taking that in from the day they’re born until the day they die,” he said. “And they’re learning from all the input all the time. We’ve never even created something that attempts to do that.”