A recent study has shown that people who have experienced discrimination are more likely to rely on AI-based recommendations instead of human advice. The fear of being judged or embarrassed can push individuals toward AI, which they perceive as more neutral and unbiased. However, if AI systems themselves contain biases, they could unintentionally reinforce the very discrimination that people are trying to avoid. This phenomenon is part of a broader psychological concept called "algorithm aversion," which refers to people's tendency to prefer human judgment over algorithmic decisions—even when the algorithms are more accurate. However, this preference isn't always consistent. When individuals face discrimination in public or social settings, such as during a service interaction, they may shift their trust from humans to AI. This shift is especially noticeable among those who have experienced unfair treatment based on irrelevant factors like race, gender, or ethnicity. The emotional trigger behind this behavior is embarrassment. When people feel judged or treated unfairly due to characteristics they can't control, they often seek out systems they believe are impartial. Discrimination in areas like access to credit or insurance is not uncommon, and such experiences can lead to stress, a sense of unfairness, and changes in behavior. These psychological effects can influence how people choose to interact with services or make decisions. In one study, participants were asked to evaluate loan applications in a controlled setting. Some were asked about their self-perceived skin color, while others weren't. All applications were ultimately rejected, but those who answered the skin color question felt they had been discriminated against. In a follow-up phase, participants were given the choice between a human evaluator or an AI evaluator. A significantly higher number of those who felt they had been discriminated against chose the AI, which was described as "emotionless." This shift from distrust in algorithms to a preference for them—sometimes called "algorithm appreciation"—is driven by the desire to avoid embarrassment and social judgment. People who feel embarrassed after a discriminatory experience are more likely to choose AI for its perceived neutrality. However, the perception of AI as neutral is not always accurate. Algorithms can inherit biases from the data they're trained on, and these biases can be difficult to detect. The study highlights that the use of AI is not only driven by efficiency or performance but also by psychological needs shaped by past experiences. For businesses, this means that offering AI-based options in sensitive areas like banking or healthcare could help some customers feel more at ease. However, the results also serve as a warning: if AI systems are not designed with fairness in mind, they could unintentionally replicate or even worsen the discrimination they were meant to avoid. True safety in AI requires more than just perception—it must be real and verifiable.