Not your father’s analytics, search and AI….”What is Watson? for 2000 please”
I can barely wait for Feb 14th 2011, when the world will watch “Watson” play the best Jeopardy players ever on National TV. I don’t get to do anything as cool as the gang on Watson team, but being able to say that I work in the same company – IBM – as these folks do, itself is a matter of great pride and honor for me.
For those of you who don’t know of Watson, please check out http://www-03.ibm.com/innovation/us/watson/ and then please watch Jeopardy on 2/14/11
Some might ask – what is the big deal here? We know Deep Blue defeated Kasparov. Is there anything to be really excited about a computer that can play a game? And why on earth should any one in enterprise space care about this at all?
Not all SDN members might have seen Jeopardy. So, if you have not yet – then I suggest you first watch a short clip on Youtube to get an idea of how the game is played. And then picture Watson, the super computer, as one of the contestants competing with humans. Or watch an old episode of Star Trek, where questions are asked in English to a computer, who then answers in English.
To begin with, Watson needs to understand human speech – in this case English, spoken with slang, accent, and often with pun. Then it needs to search a huge database ( Watson does not cheat – so no searches on google). Unlike a search engine that returns many possible links where a user can find an answer, Watson has to pick the right answer. And finally, Watson has to frame a question, which in Jeopardy’s case – is the answer. And, all this has to be done in 3 seconds . And to prepare for competition, Watson has to learn a lot – and quickly too. And to top it all – Watson should be able to learn and use already revealed answers, and evolve a game strategy.
Compare this to the best enterprise tools we have today – even for a relatively staightforward question, it is hard to get a concise answer from our existing systems. Imagine the possibilities if Watson’s capabilities are available to a company. It is hard not to get a little excited, I would say.
Watson is built on extremely powerful IBM hardware and IBM Software – code named DeepQA. Please do spend some time on the Watson website, and read about the amazing people who are working on it, and the technology and research behind it. Of special interest to me was how Watson is designed to fire many different algorithms, and then also be able to assess with algorithm to trust for a given context. Massively parallel processing is taken to a level I never thought possible.
This is innovation at its best – and it is not just some vendor marketing talk. Watson already has done some trial matches – in public – and came out on top most times. Here is one of those http://www.youtube.com/watch?v=WFR3lOm_xhE
Winning is not the big deal here – and while I hope Watson wins hands down on 14th – I am just excited that such an amazing concept could be brought to life in just about 3 years or so. AI has been a topic out of the realms of the enterprise for the most part, and now we realize how close we are to reaping benefits from this.
It is only 8 more days to the show – did I say I can’t wait?
I'm excited about new technologies. What I'm not excited about is their relevance to the business world.
The answers for business questions are not always facts. It would look like a fact if you look at it from one angle. If you look at it from another angle, it would not be a fact. Sometimes in order to please the listener(Wall Street?), the companies may need to come up with a creative way to 'manipulate' data so Wall Street would be happy. The Quarterly/Yearly reports are the best examples. Could you please tell why the best-run companies take a few weeks to report quarterly/yearly results? IMHO, the results include an unmeasurable(scientifically) quantity of "political/legal" factor.
Data quality is a huge problem. Sometimes we don't capture all data; sometimes we capture incomplete/misleading data; sometimes we move data from one system to another with a team of people who either know the source system or the target system but not both. In addition, we need to consider "political/legal" factor. In general, we're not able to get a concise answer in business world not because we don't have technology but because we don't have good data.
Regards,
Bala
Vijay, that was a very exiciting piece especially since I have read about Watson & Jeopardy...
I second Bala on the thoughts put by him, honestly, in full faith to the 'political/legal' factor that acts as the garnishing factor in each of the reports for analysis in some or the other way. Unfortunately, this factor is not constant or does not have a pattern defined, else we could have defined in our systems/reprots 😛 So finally, the reports that reaches the decision makers are the one's on which 'massaging' of the data has already been taken and in an indirect way the decision making is deflected. As we all have studied, 'White Box' what goes in, should come out! Analyze! Decide! Prosper!
Regards,
Kunal Gandhi
The value proposition is data mining in exciting ways we have not been able to do before. Companies can still use the result in whatever way they want.
Are you saying that these technologies do not have a relevance in business just because people will try to manipulate final outputs?
The idea of Watson in Jeopardy is not to augment but to replace human intelligence. And it(replacement) will work in Jeopardy because the contest is based on facts. In business, as you said, the idea of Watson is probably to augment Human Intelligence. It all depends on how one chooses to use Watson in business. This is a big difference between Jeopardy and business.
In SCN, I read discussions on the effectiveness of LinkedIn. I believe the usefulness of LinkedIn or any other social media site depends on how one uses it. And what works for x may not work for y. I also believe we don't use what we've today in business and always looking for something new. When that new thing becomes available, we don't focus on mastering it but rather wait for another new thing. This is probably another behavioral issue.
In short, regardless of what technology we discuss whether LinkedIn or Twitter or Facebook or BW or Watson, the success or usefulness of that technology in business to a great extent depends on how one chooses to use it. And mastering it is an art. In scientific world(such as Jeopardy), the success or usefulness of a product depends to a great extent on the technology behind it.
Watson is probably going to win Jeopardy:).
Regards,
Bala