Music and AI-generated suggestions: a signal for negative emotions, but not anxiety

Three interventions after a demanding task

The researchers randomly assigned 113 Chinese undergraduate students to three conditions. The first combined music from Chinese films with verbal suggestions generated by a large language model, using Eastern imagery. The second provided a standardized music therapy recording with conventional verbal guidance. The third used piano music without verbal guidance.

After a cognitively demanding 40-minute task, participants were assessed before and after the intervention. Measures covered state anxiety, positive and negative emotions, and heart rate. The 40 minutes refer to the preceding task, not the listening duration, which the abstract does not specify.

Results differed depending on the measure

After adjustment for baseline values, the group exposed to suggestions using Eastern imagery had lower negative emotion scores than either of the other two groups. The abstract reports a statistically significant difference but does not provide the absolute differences in scores. It therefore remains difficult to assess the practical significance of this difference.

No between-group difference was detected for positive emotions or state anxiety. This finding does not demonstrate that the interventions are equivalent, but it does not support the superiority of the AI-based approach on these two measures.

Adjusted heart rate, meanwhile, was lower in the group receiving the standardized music therapy recording. This physiological result therefore favors a different intervention from the result for negative emotions. A lower heart rate alone is not enough to establish a clinical benefit.

The specific effect of AI remains undetermined

The main limitation lies in the comparison itself: the conditions differed not only in their verbal suggestions, but also in the music used and their duration. The trial therefore compares bundles of elements rather than the same music with or without culturally adapted suggestions.

Cultural adaptation is an avenue to test here, not a demonstrated benefit. These exploratory findings in Chinese students do not allow conclusions about other populations, a diagnosed anxiety disorder, or lasting benefits. The authors call for larger controlled trials.

What we can take away

This study offers a lead on negative emotions after mental exertion. It demonstrates neither a specific effect of AI nor a greater reduction in anxiety, and does not justify a practical protocol.