Back
Before AI, These Three Ideas Changed Finance
29 August 2026 by Ed Freitas
Artificial intelligence is changing finance quickly, but it did not arrive in an intellectual vacuum.
Long before algorithms could assess loans, execute trades, detect fraud, or generate financial analysis, researchers were already trying to answer three difficult questions:
How should we balance risk and return?
How should financial uncertainty be priced?
Why do intelligent people repeatedly make irrational financial decisions?
In my introductory post, The Impact of AI in Financial Decisions, I looked at how AI is becoming an automated decision-making engine inside one of the world's most regulated industries.
But to understand where AI may take finance next, it helps to revisit some of the ideas that brought finance to where it is today.
Markowitz and the Mathematics of Risk
In 1952, Harry Markowitz published Portfolio Selection, the paper that helped launch modern quantitative finance.
Its central idea now sounds almost obvious: investors should not evaluate an asset only by its individual return. They should consider how it interacts with everything else in the portfolio.
That changed the conversation from choosing good investments to constructing an effective combination of investments.
Today, AI can analyze thousands of assets, correlations, scenarios, and market signals at a speed Markowitz could never have imagined. But the underlying question remains the same: how much risk are we accepting for the return we expect?
AI makes the calculation faster. It does not decide whether the assumptions behind it are sensible.
Black, Scholes, and the Price of Uncertainty
In 1973, Fischer Black and Myron Scholes published The Pricing of Options and Corporate Liabilities.
The Black-Scholes model gave finance a systematic way to estimate the value of options. More importantly, it demonstrated that uncertainty itself could be modeled and priced.
Modern AI takes this much further. Models can process market movements, news, sentiment, volatility, and alternative data almost instantly.
That capability is powerful, but it introduces a familiar danger: a sophisticated output can look more certain than it really is.
A model may produce a precise number while still relying on incomplete data, unstable relationships, or assumptions that fail during unusual market conditions.
Precision and certainty are not the same thing.
Kahneman and the Irrational Decision-Maker
Daniel Kahneman's Thinking, Fast and Slow added another essential piece to the puzzle.
Financial decisions are not made by perfectly rational people. They are influenced by overconfidence, loss aversion, anchoring, familiarity, fear, and countless other biases.
AI is often presented as a way to remove those weaknesses from decision-making. In some situations, it can.
But AI systems learn from human-generated data, operate according to human-defined objectives, and present results to human decision-makers. Bias does not simply disappear because a model is involved. It may instead become less visible and easier to scale.
Old Questions, More Powerful Tools
Markowitz helped us think about risk. Black and Scholes helped us think about uncertainty. Kahneman helped us understand the person making the decision.
AI now sits at the intersection of all three.
It can optimize portfolios, price complex instruments, detect behavioral patterns, and support decisions at enormous scale. But it also inherits the assumptions, limitations, and blind spots behind those decisions.
That is why the future of AI in finance cannot be understood by studying AI alone.
Sometimes, the best way to understand what comes next is to return to the ideas that started the conversation.