With the Rise in Popularity of AI Platforms, Marc Cooper Discusses Technology in Finance and its Consequences
Rise Of The AI Investment Bankers
by CEO Marc S. Cooper
The widespread media fascination with ChatGPT—the chatbot platform developed by OpenAI that delivers seemingly intelligent responses to queries—has drawn public attention to the possibilities ahead for artificial intelligence.
In the world of finance, that AI future has already arrived.
Big banks and other financial institutions now use machine learning software to help assess credit risks, onboard clients and detect fraud. Robo-advisory online platforms offer investors low-cost, automated investment services with little human interaction.
As far back as the 1970s, bankers began adopting algorithmic trading strategies, powered by high-speed computers and complex mathematical formulas, to lower transaction costs and execute trades in a matter of milliseconds.
Until recently, the parameters of algorithmic trading were primarily set by humans. No more. Increasingly, the machines themselves are influencing investment decisions. BlackRock, the world’s biggest fund company, has reduced its reliance on human stock pickers by relying more on self-learning AI computer code in its exchange-traded funds business. JPMorgan uses machine learning applications in some of its buy-side trading strategies to better cope with market volatility. The bank is also funding one of Wall Street’s most ambitious AI research programs.
In my lifetime, big leaps in technology, from automated teller machines and spreadsheet software programs to chatbots and mobile banking, have allowed us to transact, save and invest far more conveniently and affordably.
I recently amused my nephew, a software engineer, by reminding him that in the early 1980s when I started out, bankers ran their financial models on VAX FORTRAN, a computer language designed by IBM in the 1950s. Then came the widespread use of computerized spreadsheet programs such as VisiCalc and Lotus 1-2-3 that were far easier for banking analysts to use. Such software products, in my opinion, played an outsized role in triggering the leveraged buyout boom of inefficient conglomerates during the 1980s.
What technology has done for finance, and for business more generally, has been amazingly beneficial in my view. That doesn’t mean there aren’t systemic risks worth thinking about, as the financial world continues to embrace an emerging technology such as AI. Are we really ready for a world in which machines increasingly play a bigger role in the pricing and trading of stocks, bonds and currencies, the very lifeblood of global capitalism?
AI machine learning models are only as good as the data sets—compiled by humans, mind you—they are trained on. Flawed or biased data may lead to bad investment decisions or discriminatory lending practices.
One feature of human intelligence is the ability to simulate in our minds possible scenarios in the future. Yet that’s generally bound by our experience, what we think we know about the world. What happens when a machine’s ability to anticipate the future is taken to another level, outstripping our human ability to keep up?
In the context of finance, I think we need to consider whether AI-led markets might bring significant risks, perhaps triggering future black swan events that threaten the global financial system. The 2008 financial crisis is a reminder of what happens when investor exuberance and financial innovation get ahead of our ability to manage risk. Few worried that derivatives such as synthetic credit debt obligations would result in massive leverage on bank balance sheets tied to shaky subprime mortgages. Fewer still anticipated elevated default levels sinking the American housing market and destabilizing markets worldwide.
Now imagine a financial universe in which AI-powered trading programs instantaneously move trillions of dollars around the globe using trading strategies bankers and regulators struggle to keep up with, or even entirely understand.
Rather than the financial markets becoming more efficient, might they actually become more volatile? It’s not hard to imagine a scenario in which all the AI programs see the same opportunity, or risk, and pile in simultaneously, leading to market-destabilizing price gyrations. During the 2010 flash crash, high-frequency algorithms helped trigger a severe market decline in a matter of minutes that vaporized roughly $1 trillion in market value.
What if there were a blind spot in the historical data, or a flawed assumption, built into AI investment programs that spawned a surprise crisis? If so, what then? Would regulators and bankers have an override switch and be able to intervene quickly enough to clean up the mess?
There’s an entire genre of AI science fiction that contemplates the “singularity,” that moment in time when the intelligent machines we created turn out to control us. It’s great entertainment, yet I’m not so sure I share that doomsday outlook. There are aspects of human intelligence that will be very hard to replicate in AI programs. On balance, I think AI has enormous upside potential to make our lives more productive and prosperous.
However, I do worry about the very human tendency to relentlessly push out our technology and capabilities without perfect insight into the possible outcomes. That kind of risk-taking has allowed us to do remarkable things as a species. Yet when it comes to AI and finance, we’ll need plenty of human wisdom in how we manage the AI investment bankers of tomorrow.
This article originally appeared on Forbes.com.