In a session, “Why Tomorrow Won't Look Like Today: Things that Will Blow Your Mind."  John Kelly III, senior VP and director of IBM Research was asked if it was possible for a computer to draw upon information from a large database like Wikipedia to reason and form context-based arguments in natural language.

“The real question is can a computer not just answer complex questions with simple answers and not just take massive amount of genomic data and come up with better treatments for cancer. But can a computer take raw information and digest and reason on that information and understand the context.” Kelly said.

In a demonstration of its prototype Watson Debater functionality, Watson was able to draw from a series of Wikipedia articles to form both pro and con arguments to the topic “The sale of violent video games to minors should be banned.”

“Completely unaided by humans, the system went through all of this knowledge and constructed and 'reasoned' on its own,” Kelly said. “Now think about what this means. It's no longer, a game [of] man versus machine it's man and machine reasoning together.

Of course this doesn't mean that Watson actually understands the arguments its making. And the system still relies on data and information generated by humans to draw its conclusions. 

Kelly immediately discussed the medical applications of Watson Debater.”Think about a group of oncologists coming together and creating protocols... and having Watson Debater at the table with all of the known medical information being brought in in real time and in natural language, in dialogue, into that debate, into that argument, into that decision process. I believe this is the future of computing.”

The demonstration begins at the 45:25 mark.

  -Chris Wiltz, Associate Editor, MD+DI
[email protected]