If there’s one thing I hate more than “Barney” press releases, it’s technology PR stunts. Most recently to the top of my list is IBM’s Watson. I’ve heard about Watson’s capabilities and possibilities from so many different people both inside and outside technology. Each not realizing what it is, a technology daredevil. I remember TV’s “Fall Guy” saying a daredevil needs to make something work once. I’ve had a few people proclaiming it’s validity to ECM, but after seeing an article from American Medical News Fresh from “Jeopardy!” Victory , Watson to take on health care”, I decided it was time to speak up.
I was at HIMMS and got to have a conversation around this and how ECM needed to lead the way with Watson technology. But I pointed out to the “US Cities” question “It’s largest airport is named for a World War II hero; it’s second largest for a World War II battle” to which Watson answered “Toronto”. People focus on the fact that Toronto isn’t a city in the US. But what about the fact that the largest airport in Toronto is named after a former Prime Minister with World War I credentials and none after a battle. It’s like when Doc Hollywood was about to send that little boy on a helicopter to a heart specialist by helicopter when the problem really was that he ate his dad’s chewing tobacco. Do we want real life imitating Hollywood? Do we want our indigestion misdiagnosed as a heart attack?
“Sorry Wrong Number”
Don’t get me wrong. Watson is a great technology showcase. Something that works perfect in the world of science fiction or in Tomorrowland at Disney World. It gets us thinking about what’s possible, tomorrow. It’s show several pieces of interesting technology, speech recognition, artificial intelligence, and summarization. But really is Watson any better than contextual search? Watson was great at beating humans in a trivia game. But so is the team that cheats every week at my weekly trivia game using their smart phones. (Come on. I can see knowing that Bon Jovi was the largest concert tour of 2010, but who could know the dollars it brought in to within a 1,000 without using Google.) But it’s wrong answers that count.
Will Watson be in healthcare? No. Will pieces of it? Yes. Will IBM be all over the press saying it’s Watson? Absolutely. But really very little of the technology is new.
“The Number you have reached …”
“has been changed the new number is”
Artificial intelligence is something I played with when I got my first computer in the mid-80’s. I got a disk from PC Magazine that had an artificial intelligence engine that played twenty questions with me. It learned along the way. It started with it’s own original 10 or so questions around animals. Each time I stumped it, it would ask me what question would make it not the animal that it had thought it was. Then in the next round it would add that question. That early artificial intelligence evolved into Ask Jeeves type ask a question search engines and continues to evolve with contextual search and analytics.
“has been disconnected”
Summarization isn’t new either, especially in content management. It’s lost. I have talked about DM Partner’s Essence gisting capabilities before. This was used in the financial industry in Europe to read thousands of press release written in every European language every day to reduce the costs in translation. Essence would summarize page long press release into a single paragraph, but a human was used to decide relevance. They had a web version I used myself to summarize all the technology press release that were published.
The technology was based on Lernout & Hauspie who was the hot topic technology of 2000. Wired devoted an entire magazine to the topic of translation after a stunt where L&H show three people having a real time conversation in three different languages that none of the others knew. Of course, as we know we see this technology in use every day. (By the way, Watson would not recognize my last statement as sarcasm.)
“has been changed to an unlisted number”
And then there’s speech recognition. We work with speech recognition ever time we call a customer support line. And we know everyone loves it and finds it useful. (Sorry again for the sarcasm.) But it works, somewhat. For instance I love and still Love “Tell Me” (800-555-TELL or 800-555-8355) now AT&T #121. Tell Me was created in 1999 and was a voice portal. You could call Tell Me and it would tell you the latest news, stock quotes, sports quotes, and even play black jack with a Sean Connery voice. But I knew not to expect a lot of accuracy but it saved me time. But these day I wonder if voice response systems aren’t attached to customer service lines to get us to give up and not call. Tell Me started as a framework for anyone but is now hidden somewhere inside Microsoft.
Getting the 411
Again, let’s realize the possibilities of what Watson can offer but let’s don’t make it a requirement in business use cases just yet. There’s entire fields of study related to computer interpretation of language, called computational linguistics. The field looks for ways that computers can understand multiple meanings (bank – (n) a financial institution; (n) the area of a river where land and water meet; (v) to deflect of the side; (v) slang for to save) and context (like sarcasm) just to name a few aspects. And that field is still emerging. And then there’s the speech recognition we use daily that is foiled by not only the environment but regional dialects.
I love Gartner’s Hype Cycle Research Methodology. I put Watson at Jeopardy at the “Technology Trigger” and this latest release as the start of the “Peak of Inflated Expectations”. Anyone that’s looking of using Watson today, is headed for the “Trough of Disillusionment”.