What's Next for Watson after beating humans in Jeopardy?


So what do we do with this new
technology?  It's nice that we have a
computer that can understand natural language questions, but how do we use
it?  Gerry Moony, the General Manager of
IBM's Global Government and Education division, gave a talk on how Watson can
be used in higher education.  He talked
about how higher education was essential for a flourishing economy now, and how
people with less schooling put a drag on the workforce.  He said that what higher education did for
the work force was produce people that were capable of being the "expert" in
the previous analogy.  Thorough education
takes years to create a talent base like that utilized by companies like IBM,
but a single person with a high school diploma working a system like Watson
could do the work of a college intern. 
By using Watson, one could expand the effective talent base, improving
the economy, and thereby the local government. 
With a stronger and well-funded government there would be more support
for utilities, healthcare, and public education, starting the cycle again. 

One such use of Watson is in the
health care system.  Medicine is the
ultimate information-based profession. 
Med students are required to memorize thousands of facts and terms.  Especially in the field of differential
diagnosis, the ability to quickly access any of those thousands of facts can
make or break a case.  IBM is currently
working on a program to assist doctors in the differential diagnosis
process.  Not only does it take input of
symptoms from the doctor, it looks through the patient's file for important
details like age, trips out of the country, family history, or critical things
like pregnancy, and prompts further questions to ask the patient.  With this data it can suggest a list of
possible diagnoses with their relative confidences and justifications.  This is especially important in differential
diagnosis because Watson would not return one answer, but a group with varying
statistical probability.  After all, the
most statistically likely disease is not necessarily the one the patient

More details on IBM's plans for
Watson's future can be found at: http://www-03.ibm.com/innovation/us/watson/watson-for-a-smarter-planet/industry-perspectives.html

The problem with Watson is that,
like any human, his decisions are based on the content of his resources.  The data he draws from must remain objective,
or evenly populated by opposing perspectives. 
Achieving this balance is a challenge, so IBM is doing the next best
thing - calculating the reliability of the source into the confidences.  This, too, is worrisome because the
reliability of a source is opinion based as well. We can only hope that when
Watson is released on the internet its reliability software will kick into
overdrive.  I can see Watson, Google, and
Wolfram Alpha (www.wolframalpha.com,
the computational knowledge engine) teaming up to do amazing things for
internet searching, but some serious steps will have to be taken to account for
the variety and quality of information on the web.  I fully expect to see Watson for public use
on the web in 15 years, for use in schools in 10, and for use by medical professionals
in 5.  To convert it from a game-bot to a
useful tool will take every second of those 5 years.  However, I feel that Watson, in the right
form, could become invaluable to our knowledge and our culture, very similarly
to how Google changed how we used the internet.

Posted in: