Machine learning, human Investing

This is a great article for weekend reading. The future, by the time it arrives, is seldom as compelling as when first envisioned. Artificial intelligence (AI), especially as a tool for investment, is undergoing extremely swift technological advancement, even as some would-be adopters are hesitant. Part of that reticence is due to cybersecurity concerns, and uncertainty about whether autonomous systems create data risk greater than that of current systems. Likewise, a fund generating alpha on a consistent basis sounds good to clients, but if it is AI-based, it will likely prompt some investor discomfort about the “black box” nature of its decision making.

The London-based Man Group Plc hedge fund is among the most prominent early adopters of AI and its “machine learning” dynamic, yet managers have fretted openly about the fact that their engineers can’t fully grasp or explain all those profitable trades the software has generated for one of Man’s largest funds.

Either way, AI, as it reshapes fund management and other areas of the capital markets, has broad appeal, but the trick is to not lose sight of AI’s transformative power over the long term. This situation strikes a chord with Pippa Malmgren, an economist and robotics-industry entrepreneur whose career has connected her closely to scientific communities, both as a U.S. government policy analyst and in her business forays.

“I go back 60 years to C.P. Snow’s famous essay ‘The Two Cultures,’ and the dichotomy between science and humanities that Snow warned about,” says Malmgren. “Academia was divided into people who could quote Shakespeare and people who understood the Second Law of Thermodynamics, but the two sides couldn’t talk to each other.”

The scientific specialization she refers to now extends to financial investing, with data-crunching computers empowered to steer vast sums of money from one asset to another. The computers, however, don’t have thousands of pension checks to send out each month, or lease payments due on their G450s, so they’re off the hook when valuations tank.

Illustrating a partial solution, Malmgren cites a friend who was hired by NASA with the job title of Chief Storyteller and given the mission to make clear how scientific advancement leads to progress for society. “The people at NASA realized it was no good if you can’t explain it to non-science people,” Malmgren says. “We have an intense need for storytellers, because that’s how we’ll continue the kind of human inventiveness that spurs hope and imagination.” This is a pivotal moment, in her view, for anyone who works with institutional investors and high-net-worth individuals – one that calls for lots of non-financial skills.

“I’ve spent my whole life dealing with asset managers, and they are beginning to realize how skillful they’ll need to be as they interface with human beings in the AI age”, says Malmgren. It’s her belief that the investor-manager relationship will become progressively more layered. “They’ll want a multifaceted relationship with you,” she predicts. “They’ll say, ‘Here’s the money, and of course you have to perform, but along with that, what else will come out of this relationship? What will the fund manager teach me? What networks will I gain access to through knowing a fund manager? Some of the answers will be AI-led and some will be human-led.”

Chris Duggan, Vice President, Dart Enterprises, and Director, Kenneth B. Dart Foundation, agrees that human managers will want to double down on the functions that can’t be performed by some machine plugged into a wall. “There will always be a role for humans in the investment management industry,” says Duggan. “Instead of spending countless hours pulling data and building financial models, human traders and analysts can focus their energy on more value-added activities, like meeting with investors and building the business.”

Duggan believes that AI’s effect on the investment game is already extensive, including its disruption of an old barrier. “Where before, the giant firms could just outmuscle smaller players, today the playing field has leveled,” Duggan asserts. “Any firm with a computer and a sophisticated algorithm can compete.”

Especially regarding alternative assets, it’s important to ponder the value of patience and discipline for money management going forward, compared to in pre-AI times. Does non-human decision making about how to best deploy capital factor this in? Anthony Cowell, Head of Alternative Investments for KPMG in the Cayman Islands, appreciates the machine version of prudence and patience.

“Discipline will be built into machines,” explains Cowell. “Bots are a first phase, but with time, what we’ve seen and will continue to see is deep learning generating its own version of discipline. A very nuanced balance of restraint and risk will be part of machine learning’s contribution.” He speculates further to depict a machine hierarchy of sorts, envisioning “the machine that will keep its head when around it all other machines are losing theirs, so to speak, in events like flash crashes or momentum trading that keeps accelerating.”

Naturally, that type of breakdown will be traceable to human error in a machine’s strategic architecture, according to many AI experts. For Cowell, it’s no stretch to imagine how “a machine will copy a machine that will copy a machine, so that all arbitrage is suddenly gone from a market, at least at a certain given moment.”

Tom Chatfield, the British author and tech philosopher, believes AI could and should, in fact, have the effect of bringing far greater clarity to humans’ emotional life and behavior. As machines swiftly improve their human-like powers of reason, logic, and judgment in the name of utilitarian productivity, the distinctly human characteristics that won’t ever be built into software—what Chatfield calls our “broiling biological pot of emotion, sensation, bias, and belief”—will be exposed to a completely new method and level of analysis.

Partly with the help of AI, he believes, humans can “start talking far more richly about the qualities of our relationships,” and study “how precisely our thoughts and feelings and biases operate.” An intriguing paradox emerges from Chatfield’s analysis: When the artificial form of intelligence is fully developed, humanity’s chance to perfect the natural version may finally arrive.

This article was originally published on InstitutionalInvestor.com