AI Tools to Combat Cognitive Biases: How Investors Can Use Them

Many regulators view AI as a significant source of potential harm to investors, particularly in cases involving robo-advisors and trading recommendations. Investors themselves are also often deeply skeptical of such advisors. However, Narmine Nahidi, an associate professor of finance at the University of Exeter Business School (UK), argues in her new book, *AI, FinTech, and the Future of Robo-Advisory*, that AI tools can be effectively used to correct cognitive biases among investors.

In her book, published in May of this year, Narmine Nahidi points out that cognitive biases systematically influence investment decisions and reduce their effectiveness. These biases are linked not only to human thinking but also to the functioning of the brain. Specifically, this refers to the frontal lobes and areas associated with the processing of information and emotions.

That is why it is difficult to eliminate cognitive biases simply by providing information about risks. According to Nahidi, strategic intervention—changing the architecture of decision-making through the use of AI tools—will help.

Nahidi describes a “three-tiered” approach to such changes. The first is financial education and self-education for investors, aimed at developing “literacy in cognitive biases.” The better a person understands typical cognitive traps and is able to analyze their own decisions, the easier it is for them to resist cognitive biases.

The second layer is process organization. The author suggests structuring your investing or trading in a way that makes it easier to make well-considered decisions. For example, you can introduce a mandatory pause before making major trades or use simple and intuitive methods for assessing risks.

The third layer is the AI itself, which provides real-time personalized adjustments and recommendations precisely in situations where the first and second layers fall short—for example, due to stress, cognitive overload, or overwhelming emotions.

Such AI services already exist, but they analyze not so much the market as the investor themselves: their transactions, reactions to volatility, portfolio structure, and trading frequency. Nahidi herself cites platforms such as Betterment and Wealthfront, as well as hybrid services from major players like Vanguard Digital Advisor and Nutmeg, which combine algorithmic portfolio management with elements of human advice.

This is how AI identifies consistent patterns in human investment behavior—such as panic selling after an asset’s decline, following the herd, or ignoring historical data. Based on these patterns, AI predicts moments when an investor is highly likely to make a decision influenced by cognitive biases and intervenes. For example, it may suggest postponing a trade, remind the investor of long-term goals and the importance of diversification, and present alternative scenarios and portfolio performance trends.

Do they work? According to Nahidi, yes. She writes that they reduce impulsive trading, improve discipline in following long-term strategies, and increase portfolio diversification.

However, Nahidi adds, a sense of proportion must be maintained in all cases. The more effectively AI corrects cognitive biases, the greater the likelihood that it will shift from “protecting people to covertly manipulating their decisions.”

Of particular concern are cases where investment platforms incorporate gaming mechanics and digital engagement algorithms that encourage frequent and risky trading, Nahidi writes. Rather than reducing behavioral biases, such solutions reinforce loss aversion, herd behavior, and impulsive behavior, turning investors’ psychological vulnerabilities into a source of profit for the platform.

To protect themselves against such risks, Nahidi believes investors must take matters into their own hands. First, you need to understand why you’re being advised to take a specific action, what data is being used to support it, and what logic underlies the recommendations. Let’s recall the first layer—self-education!

Second, it’s important to ensure that AI is based on your informed consent: behavioral analysis and intervention algorithms require notification, with a clear description of their purposes and consequences.

This article was AI-translated and verified by a human editor