Not yet human-reviewed
This source record was written automatically by the research pipeline and has not yet been reviewed by the maintainer or integrated into a topic page. Treat it as a provisional summary and verify claims against the original source.
The suppressing effect of AI dependency in the workplace: Can frequent use of AI enhance practitioners’ job self-efficacy?
Citation. Ye, S., Jiang, S., & Li, S. (2026). The suppressing effect of AI dependency in the workplace: Can frequent use of AI enhance practitioners’ job self-efficacy? Applied Psychology, 75(4). https://doi.org/10.1111/apps.70121
Summary. Drawing on media dependency theory, Ye, Jiang, and Li distinguish AI use’s task-accomplishment benefit for job self-efficacy from dependency’s potential cost to mastery-based confidence. A multi-source, time-lagged field study and a scenario experiment support a competing-process account: workplace AI use has a positive direct association with job self-efficacy, while AI dependency has a negative indirect effect. Proactive feedback to AI attenuates the dependency-to-self-efficacy effect, making the reported suppressing effect nonsignificant at higher feedback levels.
Key claims
- Workplace AI use can directly support job self-efficacy by facilitating task accomplishment while frequent use can also foster AI dependency that reduces competence-relevant cognitive work and weakens mastery-based confidence.
- Across the field study and scenario experiment, AI dependency carried a negative indirect effect on job self-efficacy that suppressed AI use’s positive direct association with self-efficacy.
- Proactive feedback to AI attenuated the negative relationship between AI dependency and job self-efficacy; at higher feedback levels, the reported suppressing effect was nonsignificant.
Evidence & limitations
- The article reports a multi-source, time-lagged field study and a scenario experiment, which provide complementary but non-equivalent evidence for its competing-process account.
- The available Jina artifact contains article text and the publisher abstract but does not expose enough methods detail to record sample sizes, measures, or effect estimates; those details should be checked during human review.
- This is a 2026 peer-reviewed journal article; the mechanisms and reported associations should not be read as evidence that every form of AI use produces dependency or lower self-efficacy.