What is artificial intelligence?
I’ve had nine cups of coffee today, is my intelligence artificial? What if I took Adderall to (finally) meet a deadline, or microdosed LSD to deliver laughter with thoughtful insights? Would that be an artificial form of intelligence? No?
What people tend to mean when they invoke “AI” is a sort-of non-human, robot intelligence. Robots — or “bots,” in virtual spaces — are machines that automatically perform complex actions in response to environmental stimuli. They are everywhere: As a society, robots have been changing our lives for centuries. Robots put horses out of work. Robots displaced a generation of blue-collar factory workers. Next up, robots will disrupt white-collar knowledge workers — especially in finance.
The growth in computing power and the unleashing of countless sensors and sources of streaming data (phones, watches, email) have created immense opportunities for analyzing your personal and public lives. Ninety percent of all the data in existence today was captured in the past two years. This data and computational power — the Internet of Things colliding with Moore’s Law — are poised to unleash a new generation of robots powered by artificial intelligence that will outperform humans in ways we can scarcely understand.
AI is about autonomous learning and mimics the way a human brain learns: looking for patterns within large amounts of data, such as speech, images, text, or anything else. And, just as neuroscientists do not yet know exactly how the brain works, data scientists also often struggle to explain how AI works. For example, the CEO of a well-known hedge fund recently indicated that his fund was now using a variety of high-performing AI strategies. He also admitted this left him rather uneasy, because his team could not actually explain how the AIs made money.
If a hedge fund can’t explain how a bot — a piece of software — is delivering returns, why does that bot need to sit in a hedge fund? Shouldn’t we expect technology firms to deliver these bots without the high fees? In other words, who will be the biggest losers and winners as AI takes hold? I have some ideas:
Asset Managers: I recently asked a room full of asset managers to identify the single biggest structural innovation they were focused on. Sixty-seven percent said technology. Why? Well, consider this: Google recently announced it has machine-learning software that builds ... machine-learning software. Think about that, and you’ll see why asset managers are worried. Now, I don’t think all managers will lose out to robots, but most will have to shift their focus.
Once all factual knowledge becomes available through a voice command via your watch and displayed on a contact lens, one’s value-add in finance will transition from “what” is known to “who” is known. Today, both the “what” and the “who” are important. Tomorrow, however, humans’ comparative advantage will clearly be human relationships. Asset managers that proactively build networks of trust and mutual understanding with clients may end up replacing managers that have exclusively sought content knowledge. Trust requires human handshakes and laughter, and that I believe is the future of asset management (and a compelling one).
Asset Owners: I recently surveyed 300 large allocators on where they believed the next big technological challenges and breakthroughs would come from for their community. The responses were overwhelmingly clear. Among six options, 76% chose a single one: start-ups. I initially thought that was a good sign because, like the asset managers above, these Giants clearly realized that technologies would change their operating model. So I then asked how many had a proactive policy to track these start-ups that would disrupt their business. The answer was less encouraging: zero.
Every Giant needs a more proactive technology strategy and policy in the face of the AI revolution. Technology simply moves too fast to be complacent. Even as an academic in Stanford’s engineering school, I am surprised by the frenetic pace of change. For example, I drove home on October 14th, 2015, and parked my normal car outside my house. When I got in my car the next morning, a mobile software update had transformed it into a self-driving vehicle.
We may not have driverless portfolios yet, but we will. We don’t have matchmaking tools for investments akin to dating apps, but we will. It is not even difficult to imagine how a driverless investment portfolio might come to exist. With the right data infrastructure, an AI will assess liability profiles in the context of resource budgets and objectives and then drive accordingly.
The future abstractions and models are far beyond our comprehension. In 20 years, everything quantifiable in finance and investment will be the robots’ domain. But at the same time, trust will become the ultimate currency. The rise of AI robots in our industry will — ironically and necessarily — re-humanize what we do.