Sleep disruption increases preference for artificial intelligence

When a person’s biological sleep cycle falls out of sync with their daily schedule, the resulting fatigue can trigger feelings of social anxiety. To avoid the pressure of human judgment in this anxious state, people often prefer to be evaluated by artificial intelligence instead. These findings were recently published in the journal Computers in Human Behavior.

The human body operates on an internal biological clock that regulates sleep and wakefulness. Often, the demands of work, school, or socializing clash with this natural rhythm. Social jetlag happens when a person’s biological sleep cycle falls out of sync with these external time constraints.

A common example is staying up late on Friday and Saturday, sleeping in on Sunday, and then forcing oneself awake early on Monday morning. The resulting fatigue feels similar to traveling across time zones. This temporal misalignment does more than just cause daytime sleepiness. It can disrupt emotional regulation and make people hyper-sensitive to how others perceive them.

Researchers Xiaoyu Zhou, Yuxuan Chu, Liwei Zhang, and Jianping Liang of Sun Yat-Sen University wanted to know how this specific type of fatigue influences human behavior in the digital age. They hypothesized that social jetlag induces social anxiety, which is an intense worry about being negatively judged by others.

Artificial intelligence is increasingly stepping into evaluative roles across many industries. According to industry reports cited by the researchers, the vast majority of companies have integrated algorithmic tools into their hiring processes, and several dating applications claim to use AI to evaluate user profiles.

Free daily newsletter

Because these algorithms are widely viewed as consistent, objective machines, being evaluated by an AI might feel less threatening than facing the subjective judgment of a human being. The research team suspected that people suffering from social jetlag might actively seek out these automated evaluations to cope with their anxiety.

To test this idea, the researchers conducted four interrelated experiments. In the first study, they recruited 250 participants and assessed their natural chronotype, determining whether they were biologically wired to be morning or evening people. The participants were then randomly assigned to complete a second survey at either 8:00 a.m. or 8:00 p.m. the following day.

If a natural evening person was forced to complete the morning survey, the mismatch between their biological clock and the required time served as an experimental induction of social jetlag. In the second survey, participants read a scenario about registering for a dating application. They were told the platform would evaluate their attractiveness to recommend potential partners.

The participants then had to choose whether they wanted this evaluation to be conducted by a human staff member or an AI system. The researchers found that participants experiencing social jetlag were substantially more likely to choose the AI evaluation. By contrast, those who completed the survey at a time that aligned with their natural clock showed no strong preference between the human and the AI.

The second study explored the underlying psychological mechanism by measuring social anxiety. The research team recruited 226 participants and calculated their real-world social jetlag by comparing the middle point of their sleep cycles on workdays versus free days. The participants then imagined passing a resume screening for a job and had to choose between being interviewed by a human recruiter or an AI recruitment system.

After making their choice, the participants filled out a questionnaire designed to measure their current levels of social anxiety. The results mirrored the first experiment: participants with greater social jetlag were more likely to choose the AI interviewer. The data also revealed that social anxiety acted as a statistical bridge between the two variables. Greater social jetlag was associated with more anxiety about being judged, which in turn was associated with a preference for the AI system.

For the third study, the team wanted to ensure that people were actually fleeing human judgment, rather than just acting out of laziness or a general distaste for social interaction. They recruited 240 participants on a Monday morning, a time when weekend-induced social jetlag is especially common. The setup was similar to the dating app scenario used in the first experiment.

This time, however, the researchers provided no descriptive information about how the AI or human evaluators operated, forcing a purely intuitive choice. Alongside social anxiety, the researchers measured the participants’ desire to avoid interpersonal contact and their motivation to complete tasks efficiently.

Once again, social jetlag predicted a strong preference for the AI evaluator, with 80 percent of the participants choosing the algorithm. The statistical analysis showed that social anxiety was the sole psychological driver of this choice. Neither a general desire to avoid people nor a need to be efficient explained the preference for AI.

The final study examined whether this preference depends on how people fundamentally view artificial intelligence. The researchers recruited 218 participants and presented a scenario involving an AI-based virtual app that evaluates a user’s clothing choices. The participants were split into two groups and given different descriptions of the technology.

One group was told the AI evaluated outfits objectively based on standardized data, while the other group was told the AI had learned subjective aesthetic biases and evaluated outfits emotionally. When the AI was framed as an objective and unbiased tool, participants with greater social jetlag reported a stronger intention to adopt its evaluations.

However, when the AI was described as subjective and capable of emotional bias, the effect vanished. The socially jetlagged participants no longer showed a greater intention to adopt the AI’s evaluations, suggesting that they were specifically seeking an escape from subjective judgment.

While the experiments demonstrate a consistent pattern, the researchers note a few limitations to their work. The studies relied on self-reported measures of sleep schedules and hypothetical scenarios. These methods may not perfectly capture how people behave in high-stakes, real-world situations.

Real-life evaluations often involve power dynamics that a survey cannot fully replicate. A real job interview, for instance, carries financial consequences that might alter how an anxious person weighs the benefits of an algorithmic assessment against a human one. Future studies could track actual user behavior on digital platforms to see if social jetlag predictably shifts real-world choices.

Researchers might also explore how other factors, such as a person’s baseline technical literacy or their general distrust of opaque algorithms, interact with their social anxiety. The current findings focus specifically on short-term sleep disruptions. This leaves room to investigate how chronic social jetlag, such as the fatigue experienced by permanent night-shift workers, influences decision-making over time.

The study, “Too tired to be judged by humans: Social jetlag increases preference for AI through social anxiety,” was authored by Xiaoyu Zhou, Yuxuan Chu, Liwei Zhang, and Jianping Liang.