Know what your AI tool is optimized for. Engagement or speed isn't the same as your best outcome.

People default to one failure mode, either algorithm aversion or automation bias, and rarely calibrate.

Track your tool's track record for your specific situations, not just its general reputation.

When the algorithm and your gut disagree, let that friction prompt reflection.

Imagine you're one of the estimated 55 percent of Americans who rely on artificial intelligence (AI) to help manage your finances, select an investment portfolio, or create a financial plan (TD Bank, 2026). Or you’re a human resources manager who relies on an AI algorithm to rank your candidates to streamline the search process. In both cases, you are probably saving a ton of time, but at what cost? More importantly, do you trust the machine to create that plan or make the new hire, or do you trust your gut?

As artificial intelligence continues to dominate headlines and adoption among consumers and industrial companies is at an all-time high, this is an important area of inquiry and an under-examined category of decision science. It’s not fully intuitive, and it’s not fully rational. Seymour Epstein’s cognitive-experiential self theory (CEST) holds that two systems drive how people take in and act on information: One is explicit and reasoned, operating mostly in conscious, verbal terms. This is the fully rational perspective. The other is implicit and reflexive, built up over time through experience rather than analysis. This is the intuitive side of decision-making, relying on gut or judgment. Yet in between the two sits a mediating third party: something that feels objective, but isn't always right.

AI Decision Failure Modes

With decision-making, there are often multiple modes of failure specifically relating to the incorporation of AI. One of these is algorithm aversion, in which people who have been burned once by an AI error swing reflexively to the opposite side, distrusting future recommendations. Dietvorst and colleagues (2015) found that people lose confidence in an algorithm faster than in a human after seeing the same........

© Psychology Today