IBM’s Watson Expands Commercial Applications, Aims to Go Mobile

25 19 Loading

Watson’s nerve center in Yorktown Heights, New York.

In the past couple years, we’ve watched IBM’s supercomputer Watson mature at an alarming rate. A mere concept birthed five years ago, the cybernetic prodigy is now Jeopardy! champ and doctor in training. Up next? Cross-industry and consumer applications in the cloud and ultimately your pocket.

This is the age of big data and, with Watson’s help, big answers—lightning quick.

If you’re a Singularity Hub regular, you know Watson well. The supercomputer defeated two human champions on Jeopardy! in early 2011. And later that year, Watson gave his first practical demo in medicine and inked his first big contract with insurance firm, WellPoint.

Sibling of Deep Blue, the Kasparov killing computer chess master, Watson is a collection of immensely powerful servers in Yorktown Heights, New York. Watson parses queries submitted by humans, runs numerous parallel algorithms on tens of millions of pages stored in his “brain,” and spits out an evidence-based response. All that in no more than two to three seconds, and at the relatively low cost of $3 million (per Watson-like setup).

But time moves swiftly in the world of technology, and those early accomplishments are now verging on ancient history. AIs beating humans at Jeopardy! is so 2011! What has our favorite cyber-prodigy done lately? And more to the point, what will he do in the future?

See here for an in-depth talk on Watson by IBM’s David McQueeney at Edge 2012 (skip to 22:30 for the future of Watson):

Turns out the world is this AI’s oyster.

In March, IBM added to Watson’s health applications when the firm agreed to work with Memorial Sloan-Kettering Cancer Center (MSKCC) on cancer diagnosis. Watson could potentially reduce the gap between new cancer discoveries and their implementation in the larger population.

Broadly, medical information is doubling every five years, and yet it takes some 15 years to enter mainstream practice. Martin Kohn of IBM notes even if doctors “read one article every night, all year, that was relevant to [their] practice, at the end of the year, [they’d] be 10 years behind current information.”

Cancer research, specifically, has advanced rapidly in recent years—and in the process become that much more complex to diagnose.

Watson would leverage MSCKCC’s databases including “molecular and genomic data and [a] vast repository of cancer case histories.” Non-specialists will instantly tap into the extraordinary knowledge and expertise of MSCKCC’s oncologists to aid symptom research and patient assessment.

But IBM is an ambitious parent. The firm hopes Watson has applications in industries spanning the economy.

The machine’s first foray outside medicine looks to be in finance. It’s a logical step. Financial professionals may face even greater data challenges than doctors.

Investors confront a barrage of scholarly research papers, news opinions, financial statements, press releases, weekly, monthly, quarterly and annual economic reports…and more. Reuters alone produces 9000 financial news pages each day—never mind the thousands of other financial publications.

That’s, of course, where Watson comes in. Now that Watson has learned one subject—he can conceivably learn any subject.

At least one financial powerhouse concurs. In March, Citigroup entered into an agreement with IBM to explore how Watson’s deep content and evidence-based analysis “can help accelerate and assist decision makers in identifying opportunities, evaluating risks, and exploring alternative actions that are best suited for their clients.”

What else might Watson be used to accomplish? The University of Rochester hosted  a competition to come up with new applications for the technology. The winning proposal suggested Watson analyze weather and census data to better prepare populations for storms. (Might Watson one day collaborate with Liquid Robotics to perfect tsunami and hurricane prediction?)

Beyond potential applications, IBM hopes the next generation (Watson 2.0) will be cloud compatible with smartphones and include voice recognition. Imagine having 24/7 access to one of the most advanced supercomputers in the world! Apple’s Siri will have a tough time matching wits with Watson.

Though exciting collaborations and agreements are rolling in, and optimism is running high—the age of AIs isn’t here just yet. (Granted, closer than ever!)

While Watson worked wonderfully well on Jeopardy!, he was servicing but one user and one question at a time. Real world applications would need to service many users and compound questions simultaneously. In other words, take the prototype model and scale it up.

For businesses, $3 million is affordable, but they’ll need more rigorous proof of concept before committing. That’s why we’ve seen agreements and collaborations not full-blown orders.

As for consumers, we may have to wait a little longer. Whereas one Watson might be sufficient for a small community of doctors or financial professionals, many more would likely be needed to service thousands or millions of retail customers. Perhaps the numbers work, if folks are willing to pay subscription fees. But for now IBM is focused squarely on commercial applications.

Also, machines like Watson are only as powerful as their data, and the world is not fully digital yet, particularly when it comes to historical documents. New medical research may be digital, but what about older and as yet relevant white papers? Or what about patient medical records?

Finally, Watson is powerful and fast but still takes a significant chunk of time to learn new subjects. The supercomputer won’t be an oncology expert until 2013, roughly two years after enrolling in “medical school.” Adding other subjects to its portfolio promises to be a multi-year process.

All that said, there is no reason Watson’s learning curve won’t accelerate as the IBM team becomes a more proficient teacher, and Watson adds to his already prodigious computing powers. And although the world’s information may not be digital just yet—we are closing in on that goal daily. Not to mention the shocking statistic that “90% of the world’s data is less than two years old” and therefore mostly digitized already.

Given a few years and a few versions, we might see a handheld supercomputer capable of understanding, analyzing, and answering our every (or almost every) query. That is the promise of Watson—a fruitful human-machine collaboration of immense proportions.

 

Discussion — 19 Responses

  • Gauss156 October 14, 2012 on 12:18 pm

    This is good news. I was hoping Watson could come back up for some other uses. Having worked some with IBM, I can honestly say that they are pumping full steam ahead on research in cool areas: neuromorphic chip architectures, practical AI, process automation, and so on. Many companies have gotten sluggish in their research budgets, but IBM continues to bite the bullet and put in the due diligence in terms of research. Good stuff!

  • Tracy R. Atkins October 14, 2012 on 2:41 pm

    Watson is one of those pieces of technology where I just can’t help but feel like we have “finally made it.” It’s hard to explain, but I can’t help but think that we have finally “broken the seal” on the future that was promised long ago. Every article I read or video I see with Watson just fills me with a sense of awe and little bit of emotional magic.