By Yasuki Okai, Executive Fellow, NRI
The ongoing stream of news about advancements in artificial intelligence enabling machines to do what had previously been the exclusive domain of humans has prompted a fundamental rethinking of what capabilities are uniquely human.
I am reminded of a line from The Knowledge-Creating Company (1995, Oxford University Press), a seminal work co-authored by Hitotsubashi University Professor Emeritus Ikujiro Nonaka, the father of knowledge management. Paraphrased, the line in question says that humans are essential to knowledge creation.
Knowledge management was very much in vogue 20 years ago. At the time, many companies were working on installing groupware and/or setting up data warehouses for the sake of knowledge management. Such initiatives failed to bear fruit in most cases. The people involved in the failed initiatives presumably saw knowledge as information contained in internal documents or databases, but knowledge is fundamentally more dynamic than static documents or data. It is a deep concept that should be recognized as companies’ source of added value, as noted in The Knowledge-Creating Company.
If I were to venture a simplified definition of “knowledge” without worrying about it being misunderstood, “problem-solving ability” seems a close-enough alternative. While this definition might seem questionable at first blush, it has the advantage of drawing a distinction between knowledge and information, the latter of which is often conflated with the former. As an example, financial advisors who provide financial advice to clients make personalized investment recommendations based on comprehensive judgments that synthesize information about not only the client’s family, job and income but also available financial products and market performance. In this case, knowledge is not merely information about the client and financial products but also the ability to figure out what investments are most suitable for the client based on such information.
Do humans play an essential role in giving rise to such knowledge? Nonaka asserts that knowledge within a company initially exists in the form of tacit knowledge inside employees’ heads. Going back to our financial advisory example, are machines incapable of recommending personalized asset allocations based on information on the client and financial products?
When posed in such a manner, some can easily disagree, given the fact that robo-advisors can provide advice within certain predetermined parameters.
My answer, however, is that machines are indeed incapable of truly providing personalized investment advice because I believe knowledge resides in humans only. A financial robo-advisor does not create knowledge. It has merely learned through iterative training how to answer questions based on data input by humans or, in the simplest case, to answer questions with output from an algorithm. If the value of its pre-programmed parameters change, it must relearn. Moreover, it cannot learn without assistance from humans. Whatever knowledge robo-advisors possess is implanted in them by humans.
Human advisors, by contrast, can respond far more flexibly to clients’ reactions, though they may take cues from training manuals or guidance from their superiors. While conversing with clients, they sometimes may perceive a need to completely change the subject. For example, certain situations may call for recommendations that incorporate not only securities investment products but also insurance. Or the client may reveal worries about having enough income to afford to invest. Such situations with no obvious all-encompassing solutions are where knowledge comes into play. Only humans possess the requisite problem-solving capabilities.
Some may still argue that computers can be adequately trained through deep learning to derive such solutions. This argument implies a belief that computers can learn anything if fed the right data. However, in cases where an advisor identifies the problem to be solved by talking to the client, the range of possible problems and solutions would be so broad that training a computer to fulfill the advisor’s role is not a realistic option. Expressed conversely, one cannot help but be awed by the human brain’s amazing ability to heuristically and robustly come up with solutions to problems without any prior learning about the problem in question.
In the US, completely automated robo-advisors’ growth prospects have started to dim as a hybrid model that combines advice from humans with robo-advisory services has become the mainstream approach. Digital strategies that directly tap into humans’ value by optimizing collaborations between humans and machines will become increasingly important going forward in not only financial advisory but a wide range of other services also. Focusing on knowledge as a gauge of human value could resurrect knowledge management as an integral component of digital strategies.