The answer is, “This super computer beat a bunch of guys on Jeopardy and will now take its talents to Wall Street.” The question, of course, is, “Who is Watson?” That’s right, according to Bloomberg, Citibank has just contracted with IBM to have Watson do portfolio analysis, risk analysis, and consult on customer needs. Watson’s talent for digesting natural language may be just the breakthrough banks need in analytics to get it out of the doldrums of the financial crisis.
Of course, data mining and analytics are nothing new to banks, but Watson, Google, and Microsoft are going to change the way it is done. We’ve been talking about monitoring data on social networks for quite some time, but mostly this is centering around monitoring keywords, and providing reports to social media and business experts. This amounts at best to a snapshot of what people are thinking over a brief period.
However, Watson can read 200 million pages in three seconds. In other words, Watson isn’t just going to read a few tweets that mention Citibank, but every tweet. Then the information about what people are thinking about, struggling with, and or need can be digested quickly and reported to the right staff. No one has ever been afforded such a wide picture of their business environment and the world that affects it. It has to feel like an enormous safety blanket.
And speaking of safety, we all know risk assessment is one of the places that financial companies failed at most before the recent financial crisis. One of the great things about this type of analytics is that it can combine the analysis of traditional SEC filings with loan documents, newspaper reports, and social networking. It isn’t just about facts. Those can be discovered through traditional means. It is also about judging mood and social signals about industries, products, and companies, and doing it in near-real-time. A blue-chip stock may go from a “buy” to a “sell” just from what appears to be an innocuous tweet or article that goes viral.
Precisely what Watson will be doing is still up in the air. The first thing that has to happen is that Watson has to “learn” the financial business. It needs to grab as much data as it can to better contextualize what it reads. The great thing about technology like this is that it improves over time as it can make more connections to previous data. Watson has already helped WellPoint Health with cancer diagnosis and treatment.
If that experience is an indicator, what Watson and other natural language analytics computers can do best is simplify incredibly complex tasks. Diagnosis of cancer and treatment options is getting more diverse and difficult for single doctors to track. Watson was able to distill those complicated decisions without “dumbing down” the process or getting rid of vital information.
The other advantage is that other types of data mining don’t provide “learning engines” the way natural language analytics can. You can add more data to a traditional database without making it smarter. Watson and its competitors get better as their data grows.
The Bloomberg article says that we should expect it to start out analyzing portfolios for unknown risks. But the real revenue growth for banks is in making new products that respond to the needs of the market. Sounds like a great place to start, since I’ve never felt my bank was particularly aware of my needs. In any case, if computers can keep us out of another financial crisis, it will be well worth it.