True, the computer system Watson, created by IBM beat all human competition in the US reality quiz show Jeopardy! But does it make Watson intelligent – is it an important step towards Singularity? Is it a significant step in the field of AI? Or, is the win meaningless and does not imply anything significant? What is the future of this technology? How far are we from understanding how the brain works?
If you are not familiar with Jeopardy! Then the best place to look is, where else, Wikipedia page. Let me just mention here that the open source community is quiet excited about this win as the software component of Watson relies heavily on open source software – on Apache, Linux, Hadoop…
However, if you want to know how Watson works this is probably the best post – Could Google Play Jeopardy Like IBM’s Watson? – written by Danny Sullivan. Also, here are some thoughts on what Watson can be used for in the future – Envisioning IBM’s Watson computer as a Smartphone app.
Is the core technology inside Watson significantly ahead of the others?
How well does it compare with Google? What about Microsoft? What about the semantic question answer engine – Wolfram? Here is an interesting post by Stephen Wolfram – Jeopardy, IBM, and Wolfram|Alpha.
… just feeding Jeopardy clues into it, and seeing what documents get matched. Well, just for fun, we tried that. We sampled randomly from the 200,000 or so Jeopardy clues that have been aired. Then we took each clue and fed it as input (without quotes) to a search engine. Then we looked at the search engine result page, and (a) saw how frequently the correct Jeopardy answer appeared somewhere in the titles or text snippets on the page, and (b) saw how frequently it appeared in the top document returned by the search engine…”
There is no way to tell if Watson is a giant leap forward or it is so tailored for Jeopardy! that it is of not much direct use elsewhere. The opinion in the blogosphere is divided. Personally, I find it difficult to believe that even if Google does not have the technology today, it will not take them more than 6-9 months to match or exceed it.
Why has Google not attempted something like this?
Here are the possibilities:
- The technology is indeed very advanced and Google cannot easily replicate it.
- Wait for few months and Google will have something similar in place.
- Google does not perceive any business value for doing this – after all it is already close to 70% accurate in retrieving the correct page.
- The extra computation needed to arrive at the answer will be too much and the speed of response will suffer.
- Giving an answer instead of the links will adversely affect the advertisement revenue.
- The technology inside Watson is too Jeopardy! specific and cannot be easily generalized.
Take your pick!
Is IBM only interested in selling hardware?
This TED talk gives you some insight into Watson. The participants are Steven Baker (author of Final Jeopardy!), Kerrie Holley (IBM Fellow looking for Watson’s next job), Dr. Herbert Chase (Columbia University Professor of Clinical Medicine) and Dr. David Ferrucci (IBM Watson Principal Investigator).
Dr. David Ferrucci talked sense. According to him about 20-25 people worked for 4 years to bring Watson to what it is today. The answer I really like was his reply to the question “Does Watson think?” His response was “Do submarines swim?”
However, I cannot say the same for what Kerrie Holley said. He was asked what “Watson can be used for?” His response left me speechless! How can somebody who is in charge of looking for Watson’s next job make such naïve statements? How can he be so misinformed about what Watson is? Read it for yourself – I have transcribed his reply here:
Business Analytics is a huge area. Improving decision making … making fact based decision … evidence based decision. This is something that C-suite executives are going to love. If you think what PCs have done to enable and make them more effective … think about what internet has done. I think this represents the next stage of evolution because suddenly they have the C-suite assistant and they can begin to ask questions like if I do this if I do that what impact it is going to have on my profit, what impact will it have on my revenue? If my company is under siege financially and I have to make some cuts in my marketing budget, do I cut what I spend on TV, do I cut the print. Suddenly I can ask these questions … not only do I get an answer and probability but also get the reason why the answer of why those answers are of 80% probability or 70%. So that is an example of C-suite assistant … this is going to be a powerful tool … based on analytics … based on all the technologies you see in Watson. The brilliance of human is the ability to ask questions … that is what the machine does not do right now … so it is a great complement.
If you look at city planners who are planning traffic who are dealing with congestion … dealing with weather … dealing with accidents … this could provide another tool to planners to make better decision dynamic real-time decisions of routing traffic. Look at it as a navigator on steroid … whether it is a smart phone … whether it is a navigator on a car … we also have this capability as well where we can ask questions about we need to get here in 5 minutes … there is an accident in a road right now … what do you recommend? I think that is a powerful capability as well. Also from a travel and entertainment standpoint our ability … may be I am stuck in Alaska and I need to get into London … whatever the situation is … now use this technology to get me out of the bind … but may be to recreationally to help me plan the most optimum trip based on my preferences … based on what is happening in the world.
Give me a break! Is Watson an analytics engine? Is it a GIS software?
IBM says the technology is available starting at a million dollar which includes hardware and software. Details are not yet available on what you will actually get for this money – let us wait and see. I am also curious to know if the push for healthcare industry is because of the federal spending.
Also will there be a new dot on Gartner Hype Cycle for Emerging Technologies next year?