Research

AI-generated text

Social jetlag increases preference for AI evaluators by amplifying social anxiety, study finds

A study published in Computers in Human Behavior reports that mismatch between biological sleep timing and social schedules — social jetlag — makes people more likely to prefer evaluations from artificial intelligence rather than from other humans.

Social jetlag increases preference for AI evaluators by amplifying social anxiety, study finds

Many people must sleep or wake at times that differ from their natural circadian rhythms because of work, school and other social obligations. Researchers call this mismatch social jetlag: the internal clock and the externally imposed schedule are out of sync.

A study published in Computers in Human Behavior examined whether disrupted sleep timing affects whom people prefer to receive evaluations from — another human or an artificial intelligence (AI). The authors ran four online experiments to test the effect across different decision contexts.

Experimental design

  • In one experiment participants imagined creating a dating profile and then chose whether a human or an AI should evaluate their attractiveness.
  • In another study participants applying for a job decided between a human or an AI interviewer.
  • The third experiment again concerned dating-profile evaluation, but here neither option was presented more favorably; instead the researchers measured the actual social jetlag of participants as the difference between their workday and weekend sleep schedules rather than inducing sleep disruption experimentally.
  • The fourth experiment investigated how beliefs about the AI’s functioning altered preferences: one group was told the algorithm makes neutral decisions based on data and rules, while the other was told the system learns personal tastes and therefore could adopt subjective aesthetic biases.

Main results and figures

Across the first three experiments the pattern was consistent: the larger a participant’s social jetlag, the more likely they were to choose the AI over a human evaluator. The first experiment showed a pronounced difference: 73% of participants with social jetlag chose the AI, compared with 58.8% in the other group.

Statistical analyses indicated that the effect of social jetlag operated largely through increased social anxiety. Disrupted sleep timing raised concerns about being negatively judged by others; in such situations AI appeared more attractive because it felt less like a source of personal rejection or condemnation.

The choice was not merely about saving time or avoiding human contact: fear of human evaluation and potential bias played an important role in participants’ decisions.

When the AI advantage disappears

The fourth experiment showed that the perceived impartiality of the AI matters. When the system was described as an objective, data- and rules-based decision-maker, social jetlag more strongly increased preference for the algorithm. However, when participants believed the AI learned personal preferences and could therefore be biased, the advantage that fatigued participants gave to AI largely vanished.

This suggests that the trust of sleep-deprived people in AI partly stems from reduced fear of personal judgment by an algorithm — an advantage that evaporates if the algorithm is viewed as potentially subjective.

Biochemical background of social jetlag

The study also summarized the neurobiological mechanisms underlying social jetlag. Circadian disruption can involve several processes: altered timing of melatonin secretion, imbalance of neurotransmitters such as orexin (hypocretin), serotonin and dopamine, and misalignment of adenine nucleotides (adenosine) accumulation and clearance relative to the biological clock. Circadian disruption acts as a stressor that can activate the hypothalamic–pituitary–adrenal (HPA) axis, producing abnormally elevated evening cortisol, which impairs sleep onset and fragments sleep architecture. These physiological changes contribute to daytime fatigue and increased social anxiety.

Practical implications

The findings are relevant because AI-based evaluative systems are increasingly used in high-stakes areas: résumé screening, candidate ranking, interview processes and dating algorithms on the internet. If sleep disruption and social jetlag increase people’s willingness to turn to algorithms, this could affect who they allow to assess and influence personal and career-related decisions.

Conclusions

The Computers in Human Behavior study shows that social jetlag influences not only alertness but also social decision-making: disrupted sleep timing can increase people’s preference for AI evaluators, especially when the AI is perceived as impartial. That perceived impartiality, however, is fragile — it diminishes if the AI is believed to learn subjective biases.