IBM’s Watson Joins Georgia Tech to Solve Complex Problems

When you have a quick question, how do you find the answer? Probably by consulting an Internet search engine like Google, right? It’s fast, and if you find a reliable source, provides tried and true answers. But when scientists and engineers face complex problems, they sometimes need more innovative solutions, something that’s never before been considered.

Some scientists are turning to nature for answers, a movement called biologically-inspired design or biomimicry, which seeks sustainable solutions by copying nature’s time-tested strategies. For example, studying flamingos’ beaks to design a better water filter.

Professor Ashok Goel of the School of Interactive Computing and his students are now working to combine biomimicry with a Google-like search engine to create powerful tool. And they’re doing it with the help of IBM supercomputer Watson.

Fans of Jeopardy! remember Watson as the computer that beat 74-time winner Ken Jennings in 2011. Watson is more than a game show champ, though. According to the IBM Watson website, the supercomputer is “built to mirror the same learning process we have – through the power of cognition.” In other words, Watson makes decisions the same way humans do – by observing, interpreting, and evaluating.

Goel and his Computational Creativity class have inputted 500 biology articles and over 1200 question-answer pairs into Watson.

WATCH: Videos from the Spring 2015 Computational Creativity Class - Bioinspired invention using Watson 

“They’ve trained Watson so that if you ask any question similar to the questions it knows about, it knows which article to find,” said Goel.

Unlike Google, this new class of search engine uses cross-domain selection. Goel uses the example of lifting water from the ground to the top of a tall building.

“If you input the problem of how to lift water 100 feet high, Google will give you all of the traditional designs,” Goel explained. “So imagine a new version of Google. You type in, ‘how do I lift water 100 feet high?’ then it goes to nature and the right systems and articles and brings them to you.“

One of those articles could reference the redwood tree, the largest and tallest tree in the world.

Watson and healthcare

Goel is also studying Watson’s potential to sort through a glut of information to help medical professionals stay on top of the latest cancer research. Experts say the amount of available medical information doubles about every three years, and the process of discovery – finding the right drug to block cancer – can take 20 years of research.

Goel wants to know, “Can we use computational technology to do it in two years?”

Watson would find similarities between patient cases and cancer articles to point researchers in the right direction much faster than humans can. Goel hopes to receive government funding to support this work in cancer research.

“Our vision is that we would use artificial intelligence to provide access to all of this information that’s out there in the world to really creative people – scientists, and engineers, and people who invent things - and discover things in a language that a human being understands.”