I used a Parisian cab on November 16 and had the most interesting conversation with the driver. Upon learning that I live in the United States he promptly engaged me in a “Uber is the devil” conversation which I have heard time and again in France. The conversation took an unexpected turn though as we discussed the terrorist attacks of November 13.
He was driving around the East of Paris that night, close enough to one of the places assaulted by ISIS jihadists that he became involved indirectly in the ensuing mayhem. He ended up offering people fleeing the terror free rides to safer surroundings. He apparently was not alone, many cab drivers were similarly altruistic that night. They called one another on their closed loop radio system, and started helping until very late in the evening. The Mayor of Paris officially thanked them over the weekend for their solidarity and selflessness.
According to my cab driver, Uber did not quite behave in a selfless way. Many Parisians must have fired up their Uber app that night, more than for a usual Friday night in the East of Paris, and surge pricing kicked in he said. This would not have been the first time Uber kicked in surge pricing in a life-threatening situation – it happened already once during a hostage crisis in Sydney in 2014. Anyways, it seems Uber has denied surge pricing was in full swing on November 13 in Paris as per this article here. If that is the case, kudos to Uber and par for the course to taxi drivers hell bent on attacking Uber’s reputation.
I review hundreds of digital fintech startups that use an algorithm to better price credit, underwrite an insurance policy, invest capital, detect fraud, authorize payments. Most of these startups claim their algorithm performs a specific task much better than any human used to, can or will and do. I agree wholeheartedly with the promise of software automation, data analysis and advanced algorithms, even more so when machine learning is involved. What I have always wondered is what happens when an unexpected data point, a freak occurrence comes into play; especially one that has no connection to past data points; one that will result in the algorithm dismissing or misinterpreting said data point and where the contextual human behavior dictated by the algorithm or the decision taken by the algorithm will be out of sync with reality. One can even imagine worse case scenarios where algorithm-driven robots interact with human beings without the proper contextual understanding of a rapidly evolving real-life situation.
The financial services world obviously offers less heightened examples than a terrorist attack: a robot advisory algorithm run amok and making inappropriate investment decisions, an anti-fraud algorithm misinterpreting data and resulting in a batch of credit cards being systematically denied for every transaction in a given time period, a lending algorithm extending mis-priced credit.
Getting back to the Uber example, and if that company’s denial and actions are true then it means someone at the Uber Paris head office monitored the events and was able to override any potential algorithmically driven surge price. This is a perfect example of why “HUMAN TOUCH” is, and should be, a critical component of any well functioning algorithm. The human touch can complement an algorithm when faced with an ethical or moral argument, when a new and disconnected or uncorrelated data point emerges or even to facilitate how the new algorithm will be accepted and trusted by its users.
Getting back to a few Fintech examples:
– Several digital robo advisory wealth management models are starting to understand that human touch is essential for their business model and are including human advisors to their operations.
– Several startup digital banks realize they cannot be only mobile and only digital and are adding a human touch component via call center capabilities.
– At least one startup digital insurance wallet startup employs customer service agents as an adjunct to their automated service engine when advising their customers on renewal or changes to their insurance policies.
– A new debt collection startup with a “secret sauce” algorithm reaches out to delinquent accounts with personal messages and highly trained collection agents to provide an empathetic human touch to all interactions.
I could share many more examples where data analytics coupled with advanced algorithms delivered digitally fell short of building sustainable interactions with clients and where the HUMAN TOUCH became essential for a business to thrive. This is especially true for any direct to consumer business, even technology centric ones.
As I have explained in prior posts, fintech businesses focus on three things: a) technology, b) data and c) user experience. And user experience needs the HUMAN TOUCH.
Do you think that algorithm-centric business models still need a HUMAN TOUCH or is my thinking antiquated and will machine learning and AI will shine in a fully digital way?