60/100

Moderate evidence

Cite
More
EmailLinkedInX

Pilot study of 18 students suggests writing without AI came with higher stress

Study title: Conceptualizing "Noaiphobia": Psychophysiological stress during unaided versus AI-assisted writing in an exploratory pilot study

Evidence type
Controlled or validated comparison
Sample
n = 18
Sources
1 sourcePeer-reviewed
Published
3 min read

Study title: Conceptualizing "Noaiphobia": Psychophysiological stress during unaided versus AI-assisted writing in an exploratory pilot study

Researchers at a Tunisian sports institute tested 18 sports science students who said they use generative AI daily. Each wrote a 10-line paragraph in French on bioenergetic systems, first without AI and then with AI on their phones. Average heart rate was 93.10 bpm unaided versus 77.15 bpm with AI, a gap of 15.95 bpm.

Read the study abstract

The rapid integration of generative AI into academic workflows has raised concerns about psychological dependence and about what happens when access is lost. This paper proposes "Noaiphobia" as an exploratory construct: the acute psychophysiological stress response associated with the unavailability of AI support during task execution. We tentatively frame it as arising from three processes: withdrawal-like reactions to habitual reliance, an increase in intrinsic cognitive load once an external scaffold is remov…

The problem

In a pilot study of 18 Tunisian sports science students who use AI daily, mean heart rate was 93.10 bpm writing without AI versus 77.15 bpm with it, and self-rated stress was 73.06 mm versus 28.06 mm on a 100-mm scale.

In brief

Self-rated stress on a 100-mm scale averaged 73.06 mm unaided and 28.06 mm with AI. The authors use the results to propose 'Noaiphobia', a term for the stress response when AI support is unavailable during a task, and call the work conceptual groundwork.

The design cannot show that withdrawal from AI caused the difference. Unaided writing was always done first and is more effortful, so task difficulty, order and unfamiliarity could produce the same gap. The authors say a counterbalanced, effort-matched study with a measure of AI reliance is needed.

Main points

  1. The 18 participants were sports science students averaging 21.70 years old; eight were male and ten female. Eligibility required self-reported daily generative AI use and at least 75% in French.
  2. Each student wrote for 5 minutes without AI, rested 10 minutes, then wrote for 5 minutes with their own AI tools such as ChatGPT, Gemini or DeepSeek.
  3. Mean heart rate rose by 15.95 bpm (95% CI 11.56 to 20.34) without AI. Resting heart rate, measured a week earlier, averaged 62.98 bpm.
  4. Perceived stressfulness differed by 45.00 mm (95% CI 41.53 to 48.47) on a 100-mm scale, and Wilcoxon tests confirmed both comparisons (p < 0.001).
  5. Changes in heart rate and in perceived stress were positively but not significantly correlated (r = 0.44, p = 0.071).
  6. Heart rate was recorded with a Beurer PO 80 finger pulse oximeter, with 0.0% artifact in all recordings and no data excluded.
  7. Descriptively, women showed a larger heart-rate gap than men (19.29 vs 11.77 bpm), while stress ratings were nearly identical (45.10 vs 44.88 mm); these were not tested.

The full story

Participants were tested individually in a quiet classroom between 9:00 and 11:00 a.m. Phones were switched off and left on the desk in the unaided trial. The topic was chosen from a separate survey of 245 sports science students, in which 78.6% said they felt confident writing about bioenergetic systems.

The authors built the 'Noaiphobia' idea from behavioural addiction, cognitive load and stress-appraisal theories, and named it by analogy with nomophobia, the distress of being without a phone. They state it is a transient, descriptive state, not a clinical diagnosis, and that nothing in the data shows habitual AI use is an addiction.

Because the unaided condition always came first, AI availability is fully confounded with order. By the second trial, students knew the task and setting better and had already worked out content. The authors say this means the observed gap is better read as an upper bound, though they could not quantify the order effect.

Other gaps noted include no measure of AI dependency, no manipulation check that AI was actually used, no check that heart rate had recovered between conditions, no scoring of the written texts, and an unrecorded first language for participants. The authors propose counterbalanced designs, validated dependency scales and wider samples.

Sources

1 on file
  1. Addictive Behaviors Reportsdoi:10.1016/j.abrep.2026.100765
    Peer-reviewed2026-10-25

Talk it over

Three questions to start on, two more if you have time
Unaided writing always came first for the 18 sports science students, and the second task was done with their own AI tools. How would you weigh the 93.10 bpm against 77.15 bpm gap given that order?

Start with a question

Pilot study of 18 students suggests writing without AI came with higher stresshttps://truvace.com/item/conceptualizing-noaiphobia-psychophysiological-stress-during-unaided-versus-ai-a

  1. Unaided writing always came first for the 18 sports science students, and the second task was done with their own AI tools. How would you weigh the 93.10 bpm against 77.15 bpm gap given that order?

    Answer in the thread
  2. The authors propose 'Noaiphobia' for stress when AI support is unavailable. What result from a counterbalanced, effort-matched study would make you accept the term, and what result would make you drop it?

    Answer in the thread
  3. If you were one of the students asked to write a 10-line paragraph on bioenergetic systems with your phone switched off on the desk, what would you want the researchers to know about how you felt?

    Answer in the thread
If you have more time
  • Look up how nomophobia was first measured. What would a similar scale for AI reliance need to ask, given that this study took no measure of AI dependency?

    Answer in the thread
  • Women showed a larger heart-rate gap than men (19.29 vs 11.77 bpm) while stress ratings were nearly identical. What larger sample would you want before reading anything into that?

    Answer in the thread

What readers say