Could IBM’s Watson replace Siri? That’s an interesting question and IBM’s answer appears to be yes. Big Blue is working to turn the supercomputing solution that made news when it beat Jeopardy! champions Ken Jennings and Brad Rutter into an app that could run on a smartphone. If successful, IBM will turn Watson into a supercharged version of Apple’s digital assistant.
Bernie Meyerson, IBM vice president of innovation, sees a voice-activated Watson as being something that everyone could use. In talking to SFGate reporter Sarah Frier, Meyerson described one scenario of how Watson might act like a much smart sibling to Siri.
A farmer could stand in a field and ask his phone, “When should I plant my corn?” He would get a reply in seconds, based on location data, historical trends and scientific studies.
That’s an incredible synthesis of information, but Watson is proving that it is capable of making such predictions. It’s certainly well beyond Siri’s current ability to tell you whether you need an umbrella and it’s even beyond Siri’s enhanced functionality in iOS 6, although from a user perspective, it’s essentially an extension of Siri’s functionality.
Siri, for example, will be able to tell you the score in a face-off between the Green Bay Packers and Minnesota Vikings. Watson would be able to tell you the current score and predict the final score and the winner based the starting lineup, injury reports, past experiences of each team and each player, and could even incorporate the weather conditions as a factor.
Currently, Watson is still be used in pretty specialized ways by IBM customers. Citigroup is using Watson to crunch financial data and WellPoint has Watson analyzing cancer data. Both are extremely complex and they are also very specific.
Crunching data for the everyday needs of smartphone owners might seem like simpler calculations, but the fact is that Watson needs time to learn about the data that it’s processing. IBM’s deal with WellPoint was announced last September and the companies don’t expect Watson to be fully aware of oncology as a field of medicine until sometime late next year. Imagine that kind of learning curve applied to NFL teams, farm data, and thousands upon thousands of random questions that individuals would ask Watson if it was immediately accessible in the way that Siri is on the iPhone 4S – would Watson understand how to respond to the first few people who asked it open the pod bay doors?
Beyond understanding the data, there are other questions. Some are easy, like giving Watson speech and image recognition capabilities. Some are a bit harder. Watson already consumes a huge amount of energy at its home in a Yorktown Heights, N.Y. data center. Adding the capacity to handle queries from users around the world would require significantly more power. There’s also the question or priority – Watson’s taking questions about a football game, for example, should not be able reduce its capacity to analyze potentially life-saving medical data. Beyond all those questions is another important one – how would IBM distribute Watson 2.0 to a global community and how would the company monetize it?
Those questions aside, one iPad application of Watson already exists – a health care program developed with a Columbia University professor is being used to demonstrate Watson potential medical capabilities to prospective IBM customers. Those capabilities, according to IBM’s chief medical scientist Martin Kohn, could include diagnosis and treatment options for various forms of cancer once Watson has finished learning about oncology.
Ultimately, IBM may face more criticism over a smartphone version of Watson than Apple has over Siri’s inability to understand many questions. Watson, after all, has proved itself to be more capable and more intelligent than Siri. That means users would expect or demand a much higher level of accuracy and success.