In 1983, I was approached by Index Systems, a Kendall Square software developer of financial services applications, apparently the result of a referral from a consultant working with the Corporate Planning Department at the Traveler's Insurance Company in Hartford. Index Systems, closely associated with MIT's Artificial Intelligence Laboratory, was working on an assignment for Traveler's to determine whether the insurer's Agency force could be upgraded from a conventional insurance distribution network to a consultative-sales organization. Traveler's long-range planning staff thought that converting their product distribution network to agents-financial planners might develop expanded cross-selling opportunities and upgrade the efforts of its organization, as well as build a closer longer-term relationship with the owners of its life and property-casualty insurance policies.

Travelers Selects TFC as Domain Experts

Although it was still in its infancy, Index and Traveler's had become intrigued with the idea of using the newly emergent Artificial Intelligence (AI) technology as a platform for a comprehensive personal financial planning system to run on desktop minicomputers (think Digital Equipment Corporation and Xerox). I was engaged as the domain expert around whom to form an ancillary firm, Applied Expert Systems (Apex), which would employ TFC's approach as the logic structure for an AI-based expert system running on a Xerox 960 Series minicomputer¹. Warner Henderson, Fred Pryor, and another independent financial planning advisor, Bob Wegner, joined in this pioneering effort from time to time. Two years later, the result was the most elaborate financial services expert system of its time — based on 6,600 "if-then" rules - and focused on the personal financial planning community.

The Apex saga became a classic start-up venture in its own right, growing from 3 employees to 128 in 18 months and back to 12 a few years later. The story became the subject of a PhD thesis and a Harvard Business School case study², focusing on the introduction of disruptive technologies in the workplace. It continues to be taught at business schools throughout the country — for example, in 1992 I sat in on my daughter Charlotte's HBS section class on Disruptive Technologies, where the case was taught by its author, Professor John Sviokla.

Computer Hardware Speed Constrained AI's Potential

The Apex program produced a thorough 70-80-page client plan, but the minicomputer's speed and computing power were inadequate for the task. The Traveler's agents using the system faced unpredictable and long waits, but perhaps worse, users were never convinced that the system was producing correct results. In many cases, the Traveler's agent was not made aware of the assumptions driving the AI logic, and had trouble accepting the resulting conclusions. Further, it turned out that the Traveler's agents did not necessarily wish to become consultants responsible for longer-term retainer customer relationships and resisted the home office's blandishments to become more involved in their customer's personal financial affairs. In the end, the project generated a write-off for Travelers of just under $40 million, and, for those of us involved in the RIA world, provided further confirmation that the delivery of effective targeted financial advice requires face-to-face meetings between the client and a trusted advisor who can provide context and continuity.

Today, with virtually unlimited computing power, AI-based robo-advisory programs accessed via the internet are touted as the next phase in the evolution of financial planning services for HNW clientele. The reality is that in some segments of the trusted financial advisor-client equation, AI-based analytical software can be helpful as a tool or technique to analyze complex data and test possible future scenarios. Coupled with a series of Monte Carlo simulations, AI software provides a reasonable means of generating a range of outcome probabilities, but as usual, the "garbage-in-garbage-out" inconvenient truth remains. Regardless of those limitations, robo-advisory programs continue to be promulgated, and even the Vanguard Group has announced a completely automated online app to be targeted to a younger audience which is accustomed to online solutions.

AI's Challenge

Global financial markets are complex, vast, interconnected, and volatile systems. When investors offline and in a contemplative rational mode select the parameters that define their needs and risk profile, the result is a sensible policy outcome. In moments of stress, reactions are often emotionally driven. The outcomes of an AI system using those original rationally derived parameters can be useful and modulate the knee-jerk need to "just do something." But in this context, AI struggles to narrow its focus and has a long distance to go.

Investing combines luck and skill, and emotions play a significant part in explaining, never mind predicting, individual investment performance over short and intermediate periods of time. When creativity enters the equation, AI may never replicate the investment wisdom of a Warren Buffet or a Peter Lynch, or the technological innovation intuition of a Steve Jobs. Achieving human sentience in a "thinking machine" through AI programming today would seem a stretch. In a global mean reversion investment performance environment generating millions of data points daily, expectations that AI algorithms will be able to better their relevant investment return benchmarks are bound to be disappointing.

The real challenge AI poses for the individual investors and advisors, and certainly for robo-generated internet interactive services, is not so much whether such platforms will replace in-person rendered advice, but whether users (both advisor and client) can learn to work in a complementary fashion as the science evolves, offering clients and advisors alike the best of both worlds. Conversational AI bots will increasingly interact with their human counterparts and improve their spontaneous reactions to a broadening array of questions and verbal requests. Regardless, the in-person, face-to-face creative personal exchanges in the HNW marketplace will be difficult to replace as the preferred mode of communications.

____________
¹ See the Business Week article in the Appendix
² See HBS case study 2/87 in Appendix

Proceed to Managing Other People’s Money >

< Return to Fashioning A Useful Personal Financial Plan Prototype