Yesterday I watched Intelligent Week’s webcast “Beyond Jeopardy!: The Business Implications of IBM Watson”, the webcast of a panel discussion with Stephen Baker (author of “Final Jeopardy”, which he describes as “a sort of biography” of IBM’s Watson, the computer that was built to meet the challenge of competing on the TV show Jeopardy!), and David Ferrucci, Tom Rosamilia, Arvind Krishna, and Dan Pelino, all of IBM.
A few of the themes that came up were: (i) that Watson is doing so much more than search, in that it considers what evidence it has for a certain statement, and calculates confidence levels of different statements; (ii) that Watson can do a lot more in real life situations than it could on Jeopardy. On Jeopardy, it was not permitted to be connected to the internet, nor to receive any hints or direction from people. In actual situations, it may be desirable to use it in a way such that people can interact with it and it can interact with the internet. (iii) Health care is only one of the areas of application envisioned for it. Just as it might aid physicians in ways that no human assistant ever could, due to its ability to “digest” vast quantities of published material, so it might similarly help other professionals. It might aid (rather than replace) customer service representatives, by suggesting a reply to a particular customer that improves on the scripted response by taking advantage of databases and statistical information it can process.
I think the overall message was that, even though the Grand Challenges of chess and Jeopardy! put IBM machines in competition with a human, that Watson’s applications ought to be thought of as a machine and human cooperating rather than competing. And, that they complement each other: Watson’s capabilities exceed human ones on the tasks it is really good at, and a human’s capabilities exceed Watson’s on the tasks that humans are really good at. Rather than asking whether a machine can ever do what a human does, it seems to me, the topic of the day was that machines can help humans do much more than they can do without them. Not a new idea in industry, of course, but perhaps a new idea for work traditionally thought of as professional work that was beyond the reach of automation.
They did get around to the question I was interested in and mentioned in an earlier post: do you have any idea what the next Grand Challenge might be? No concrete answers, but the general feeling was that the next step would be being able to integrate, not just text, but a variety of sensory modalities, a variety of media.
It’s now available online: http://bit.ly/elpurE It might make you think.