Many individuals are worried about an AI-dominated future due to AI’s stratospheric growth in public consciousness. Will AI change industries? Does it democratize or solidify them? Will it improve or worsen outcomes? Finance has been changed in the last decade by the same forces driving AI: the distribution of powerful computation and the plethora of data. The financial experience is exciting and depressing for an AI-dominated future. It implies that AI will alter some (but not all) industries, benefit larger actors.
Finance is an obvious laboratory for AI research since information processing is important to financial markets. Financial institutions of all types invest extensively in technology and data before other businesses to compete effectively. The expertise of finance may not fully explain the scope of newer huge language models that have wowed the world in the previous six months. However, the evolving competitive dynamics in finance over the previous decade suggest what may happen across many industries when AI gets cheaper and more widespread. Finance will always be the economy’s canary in the coal mine, regardless matter how future AI versions turn out.
First, AI looks to affect industry dynamics swiftly. Asset management is an example. Two major upheavals have occurred in the previous 15 years due to technology and data domination. First, passive fund managers (who invest in indexes without research) have increased and active fund managers (stock pickers) have declined in the mutual fund business. Data and technology made passive investment more competitive and made it harder for active managers to gain informational advantages, accelerating this change. In the previous eight years, the ratio of passive to active assets has increased from 0.6 to 1.2, a major market share shift. Passive fund managers have shown they can replicate many active fund management methods at a tenth of the cost, destroying active fund managers’ ability to charge hefty fees (up to one percentage point of assets under management).
Second, quantitative investing has replaced fundamentals-driven long-short strategies in hedge funds. The ability to swiftly analyze vast amounts of data and construct short-term plans looks to be outperforming slower and deeper analysis that led to long and short investment decisions. These finance tendencies show that an AI-dominated future can quickly create outsized winners and losers.
However, financial experience reveals that not everything changes as soon as predicted. Financial trading, with its mix of macroeconomic, sentiment, and company-specific data, has changed swiftly, but wealth management and lending have altered less.
Robo-advisors’ capacity to overtake the enormous financial advisory complex may be stalling. The customer side of finance prefers humans. AI has not altered lending as expected, and AI-powered lenders have had many issues. The additional data on individuals and company credit may not be as relevant as in financial markets.
AI’s ability to disrupt industry dynamics appears to be tied to the information problems it solves. Financial markets are complex information problems that demand plenty of data and computational power. Similar fields like medication design may be disrupted by AI. However, several industries, such as manufacturing and services, may be more relevant to wealth management or financing than AI. The finance business suggests that AI can preserve human-facing services with limited data and fast change. AI can still improve decision making, but it will likely be incremental (as in wealth management and lending) rather than transformational (as in money management).
Finance can also reveal whether AI will democratize or consolidate. It appears the explanation is clearer here. In financial markets, scalability and speed seem to be key to AI success. When technology and data dominate, winners keep winning and investing in them is the essential point. A smaller quant fund has trouble getting data feeds and computer capacity compared to larger firms. Passive investing fees continue to fall as larger businesses share scale with investors, locking out upstarts. Scale is crucial in AI-transformative industries, and promises for a big unleashing of smaller players that challenge existing giants seem overblown.
Does the finance industry’s experience indicate that AI is good for humans? The world of banking is more gloomy here. Active managers who charged high fees for little results were replaced, which is good news. Financial markets are not processing information well and may be becoming worse. The development of passive investors and quant funds may disregard the hard job of evaluating slow-moving, confusing, firm-specific information. As data and computing dominate, enterprises may overuse fast-changing hard data like stock price movements and credit card spending data. Softer data (e.g., corporate prospects, management quality, pricing strategy long-term effects) can be downplayed, even though it matters to markets.
This last lesson may generalize well, I fear. AI’s ability to examine hard data unstructuredly without human intervention will change the world, just as financial markets have. That change may be limited to data-rich, fast-changing environments. Winners will be the largest corporations that can invest in processing power and data to differentiate strategies. Even though it matters most, the ability to consider softer data may lose value in the short term.
Can financial markets capitalize on AI without ignoring these core issues? The current equilibrium is a financial market dominated by huge companies providing commodity services cheaply while ignoring softer information. The difficulty for finance—and probably all of us—is to recognize that managers and leaders’ toughest questions are not exclusively data-driven. What will help my business prosper in 10 years? How can I best use funds to develop and improve customer service? Hard data will inform these decisions, but it may not be conclusive. These choices demand ingenuity and conviction. As AI makes hard data cheaper and more efficient, these judgments will become more important. To recognize the significance of these human questions does not limit AI’s ability to aid us; it just reasserts that AI is a technology and that managers and investors benefit most from these fundamentally human undertakings.
