IBM's Watson: Banking's Savior?
Banking has been in the dumps since the financial crisis of 2008. The road ahead doesn’t look much more promising. What with increased regulation, fewer significant profit centers ( prop trading, for example) and an extremely challenging commercial and residential real estate market, banks don’t seem to have a lot to look forward.
Few will shed tears for bankers as they struggle through tough times. The banking industry needs a superhero.
Enter IBM‘s supercomputer Watson. The proposition: use advanced technology to tap into the mountain of data coming into financial services firms to get closer to customers and become more profitable.
Last week I moderated a panel discussion at Forbes NYC Galleries called “The Power of Advanced Analytics for Smarter Banking.” My panelists included IBM’s global banking and financial services chief Boxley Llewellyn, financial analyst John-Patrick O’Sullivan, who happens to be associate director of SNL’s Financial Institutions Group, and Duke Chang, program director for Watson Solutions within IBM’s Software Group. We had a candid discussion about where banking was in terms of using its data and information efficiently and productively, and what the future might hold for the industry. [See the full videos of the panel discussion below]
Watson, as you may know, is IBM’s artificial intelligence supercomputer that has the ability to process hundreds of millions of data-points in seconds. Most computers require structured data, things that can be collected neatly into spreadsheets or database software. IBM’s Watson goes a step further because it is masterful at making sense of vast amount of unstructured data. Things like e-mails and online Twitter chats, video and recorded conversations you might have with a customer service rep. Believe it or not some 90 percent of the world’s information was created in the last two years and, according to Chang, 80% is in the form of unstructured data.
IBM’s Watson can answer questions posed in natural language. Apple iPhone’s Siri can do the same thing but I’m told it is in a much less sophisticated manner because Apple’s system is based on keyword matching. Watson seems to be more of a deep thinker. Watson, you may remember, trounced Jeopardy game show champions in an exhibition show. [ See Watson vs. Jeopardy]
Duke Chang told the group that Watson is now focused on a few real life applications. Health care is a major focus for Watson. My colleague Bruce Upbin recently reported how Watson is working to help doctors with medical diagnosis. According to Bruce, answering the question “What’s wrong with this patient given these set of symptoms and this family history?” is a near perfect challenge for a supercomputer like Watson. Indeed Watson has been deployed at Memorial Sloan Kettering Cancer Center to help with diagnosis and treatment of cancer.
Financial services, with its vast amounts of structured and unstructured data would seem to be another perfect fit for Watson’s skills and that is what the panel discussion focused on.
Apparently, banking systems are prone to human error resulting in poor data quality. They call it “data fuzziness.” Fuzzy data fouls up assumptions that go into financial models that might say, help a wealth management division model out a financial plan. This data fuzziness is also a big problem when banks assess the risks of certain market changes or product innovations. Garbage in- garbage out. We all witnessed the consequences of that in 2008 and 2009 ( Remember this bit of fuzzy data? “It’s not statistically probable that all the mortgages in this pool will default. So rate it AAA”).
Watson to the rescue, or at least that is what IBM hopes.
“Most financial services firms are dying of thirst in an ocean of data,” said Watson’s Chang.” “There are so many questions that they want to ask about their customers or about the things going on in the market and they know the data exists inside their walls.” The task at hand is to collect and organize all of this unstructured data and make it useful so that banks can make better decisions, mitigate risk and earn more money.
During our discussion JP O’Sullivan from SNL Financial made some great points about how most banks were not really ready to take full advantage of advanced analytics because they were still recovering from the effects of the financial crisis. Sullivan spoke about how most banks were still not doing a good job pushing pertinent information they had gathered to either senior managers or to front-line customer facing employees.
Comments
What is the cause of the next financial Collapse, Alex?
How about a predictive analytics system that will tell us what the economy would look like after stabilizing the housing markets and mortgage financing system. The financial system will continue to be a drag on the economy for several more years. Without proprietary trading, there will not be profits for the banks like they were getting before. We’ve got to go back to Glass-Stegall days. If a bank wants to invest for itself, then it should shed its commercial deposit accounts. There is no reason to allow banks to make risky investments based on capital not available for absorbing loses from those investments.
The housing and mortgage markets can be stabilized, restructured, recapitalized and put back into the economy as functional contributors realitively quickly compared to the revival of the financial sector. It appears the government is committed to retaining Freddie and Fannie in the mortgage mix. The use of these non-government organizations needs to be limited as does the payment of senior executives. They should not be involved in any mortgages that don’t meet decent credit criteria. The failure of these organizations during the past decade or so is not the organizations place in the mortgage system but the failure of the board and management. The management of these organizations needs to be focused on the mission to serve the private mortgage system instead of competing with it or trying to suck up some of the huge profits eing made. Get housing fixed and keep the financial system from sucking up too many more of the nations financial resources.
I suspect Watson may give bankers another chance to screw up royally as they once again start to trust models of human behavior.
We can model physical universe phenomena very accurately, because atoms and molecules behave according to laws and don’t have incentive to cheat and game the system. Humans however do have every incentive to figure out the rules and bend or break them to their own advantage. All you need is one or two to do it and the whole thing can collapse.
Your understanding of physics is out-dated by a century or so.
Anyway, if you assume humans will seek to maximize utility (as Austrian economists do), you won’t have such a hard time predicting human behavior.
I would use Watson to play Pong.
Watson has some great chops on the unstructured data, for sure.
But ask it “Hey, remember the flash crash, what happened there?”, and it’s likely to go quiet.
Lots of Big Data in markets is very structured, and very time sensitive, to the point of worry to many. Getting this into the mix would be a big win for Watson.