This is the final installment of a three-part series exploring the impact of artificial intelligence (AI) on investment management. I want to thank the speakers at the AI and the Future of Financial Services Forum, hosted by CFA Institute and CFA Society Beijing, for inspiring this series. The initial articles offered a primer on the AI technologies that are relevant to investment professionals and explored the potential threat AI posed to human portfolio managers.
Not all is lost, investment professionals.
Despite artificial intelligence (AI)’s significant and rapidly increasing “brain” power, the investment management business is not going away tomorrow.
But it is changing and the current model will one day be rendered obsolete. So what does the road map leading to that eventuality look like? And what’s our best strategy in the interim?
Contrary to conventional wisdom, the data scientists on the panel at the AI and the Future of Financial Services Forum in Beijing in December collectively reassured the investment managers in attendance that AI will not drive them out of business overnight. Why? Because in the words of Eric Chang of Microsoft Research Asia, “There is not yet enough data.”
Given the biological limits of the human brain, researchers believe that the threshold for an intelligent system to beat human beings at a goal, say playing the ancient board game of Go, is about 10-million rules.
To put all this in a layperson’s language: It takes time and effort to write the code and tag the data for the machines to process.
How much time? In the ImageNet image recognition competition, 50,000 people from 167 countries took more than three years to organize and label over 100-million pictures. And investment management may be a more nuanced subject matter than image recognition.
Moreover, although finance is a data-rich industry, financial markets are not controlled environments and encounter many unanticipated events that AI is not particularly well-suited to navigating.
“[Warren] Buffett can make split-second decisions on complicated M&A deals,” Chang said. “AI cannot do that yet.”
Neko Chen, the former CTO of Goldman Sachs China, agreed, pointing to instances of market flash crashes and the inability of machines to respond in a timely fashion.
Another practical challenge AI programs face is that they cannot explain themselves. By definition, a deep learning model is a black box. Quants are often blamed when their black boxes misfire, but not because they don’t know what’s in them. They just choose not to share their trade secrets. Data scientists, on the other hand, may really not know what’s in their black boxes.
A critical point to remember: This is not a race between humans and machines. As Chang put it: The AI plus human intelligence (HI) model holds the most promise. Our competition is not machines. It is the other people plus machine teams out there. We want the smartest machines working for us and our best chance of building them faster is to work with the best in AI.
This supports our May 2016 hypothesis that the optimal way forward for fintech is likely through the collaboration of powerful financial institutions and powerful technology innovators.
So what can investment professionals today take from all this? We will all yield to AI gradually over an extended period of time. In that process, investors will enjoy the benefit of AI-powered “assisted investing,” much like drivers today enjoy “assisted driving” en route to the moment when self-driving cars rule the roads.
What’s the road map for AI in investment management? This may be an oversimplification, but I believe (1) portfolio managers will have longer careers than analysts, and (2) investors in liquid markets will enjoy the benefits of AI sooner.
“AI’s advantage is in standardized and repetitive tasks,” Shu pointed out. In the near future, analysts will likely free themselves from such mundane chores as building basic financial models. Insights, rather than Excel skills, will rule the day.
It will be an iterative process: Analysts will help build intelligent systems to process more information to help analysts generate insights. Analysts will continue to fine-tune the system until it becomes the Siri that can answer all our questions.
The role of a portfolio manager involves more “dimensions,” however. Portfolio managers tend to cover more sectors or countries than analysts and make their decisions after considering a variety of “rules” that will take longer to code. Less liquid markets, such as convertible bonds or frontier markets, generally have less data with which to train the neural network models. In other words, our jobs are more secure if we work in the less efficient segments of the market.
But whatever our roles, nobody can afford to be complacent.
“Your destiny is in your hands,” Li Hongyu of ZhongAn Technology told the audience. The panelists all agreed. “Continuing professional development is a must even without AI.”
And obviously, the more we improve our investment skills, the harder it will be and the longer it will take for machines to catch us.
May the best team win!
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©Getty Images/ktsimage