Too tired to be judged by humans?

Across four experiments, greater social jet lag was linked to stronger preference for AI evaluators, largely through social anxiety — and the effect depended on whether AI was framed as neutral.

PSYTECH / LENS
TechnologyAI systems
Human systemSocial cognition × Emotion regulation
EvidenceSource-bounded
The question

What happens to trust when the system becomes part of the social judgement?

Browse this human systemsocial cognitionemotion regulationsleep arousal

What happened

A study reported in Computers in Human Behavior tested whether social jet lag — the mismatch between a person's biological sleep rhythm and social schedule — changes who they would rather be evaluated by. Across four experiments involving dating profiles, recruitment and virtual outfit reviews, participants experiencing more social jet lag showed a stronger preference for AI rather than human evaluators.

What the evidence says

The reported pattern was not simply a preference for convenience. Social anxiety and fear of negative human judgement were central to the effect. In one experiment, 73% of participants experiencing social jet lag chose AI evaluation, compared with 58.8% of those without it. The preference also depended on how the AI was described: presenting it as objective and neutral strengthened the effect, while describing it as subjective and potentially biased weakened it.

PSYTECH / FIG. 01Mechanism diagram

Who feels safer to judge you?

How could sleep timing alter evaluator preference?

Reported pathway · four interrelated studies

    1. Source reports
      Social jet lag

      Mismatch between biological timing and social schedule.

    Reported anxiety pathway

    1. Source reports
      Social anxiety

      Anticipated pressure of human judgement.

    Reported mediation

    1. Source reports
      AI evaluator preference

      A less threatening evaluation option in the study settings.

Evidence boundary

This is the authors’ reported pathway, not a universal causal chain. Preferences in study scenarios do not establish behaviour in consequential hiring or dating.

Sources & figure notes
  1. Computers in Human Behavior · study abstractPrimary research

    Abstract: four interrelated studies and the reported social-anxiety pathway. Full methods and endpoint-level results were not inspected for this figure.

    Used for: Social jet lag; Social anxiety; AI evaluator preference

  2. Tech Xplore · research summarySecondary account

    Source of the published brief and the perceived-objectivity condition. No percentages from this summary are reproduced in the figure.

    Used for: Perceived AI objectivity

Original PSYTECH figure · AI-assisted editorial construction · no human review recorded

Evidence boundary

PSYTECH is reporting a secondary research summary rather than independently reproducing the study. The experiments illuminate a mechanism under controlled conditions; they do not establish that sleep disruption will cause the same choices in every real hiring, dating or assessment setting.

Why it matters psychologically

Fatigue may change more than performance. It may change who people are willing to be judged by. Here, social anxiety appears to make algorithmic evaluation feel safer when the system is perceived as less personally judgemental or biased than another human. That makes perceived machine objectivity part of the psychological mechanism, not merely a product feature.

What remains uncertain

The important next question is whether the effect survives outside imagined or online evaluation scenarios, especially where AI assessment has real consequences. Replication across populations, direct behavioural settings and different forms of sleep disruption would clarify how durable the preference is.

Primary source

- Out-of-sync sleep schedules may steer people toward AI for dating and job assessments — Tech Xplore / Phys.org. URL checked 2026-10-06.

Evidence & provenance1 source · production record

Source record

Sources stay inspectable so the boundary between reporting and interpretation remains visible.

  1. 01

Production

Deterministic editorial system

Source processing and editorial assembly are rule-based.

Generative contribution

AI-supported editorial production

Generative AI contributed to headline, standfirst, editorial framing and copy editing. Source claims and evidence boundaries remain tied to the recorded source.

Recorded model: GPT-5.6 Sol

Human review

Not recorded for this release

Editorial boundary. PSYTECH News is an editorial publication, not a medical service. Nothing here is diagnosis, therapy or clinical advice.