Institutional Repository of Vilnius University
Dirbtinio intelekto autorystės atskleidimo poveikis pasitikėjimui karjeros patarimais: eksperimentinis tyrimas tarp Vilniaus universiteto Ekonomikos ir verslo administravimo fakulteto studentų /
Abstract
dc:descriptionGenerative AI is increasingly used to produce career advice, yet it remains unclear how students evaluate AI-generated advice and how that evaluation changes once AI authorship is disclosed. After reviewing the literature, the author found that prior studies document a disclosure effect whose magnitude depends on task type and AI familiarity. The main aim of this work is to experimentally assess whether students evaluate AI-generated and human-written career advice differently when the source is unknown, and how disclosing AI authorship changes these evaluations. Five objectives guided the work: literature review on algorithm appreciation, aversion, and disclosure; between-subjects experiment design; comparison of evaluations across blind and disclosed conditions; heterogeneity analysis; and policy recommendations. In the experiment, the author applied a preregistered three-condition between-subjects design (AI blind, human blind, AI disclosed; AsPredicted #266,264) with N = 171 VU FEBA students. OLS and ordered logistic regressions with progressive controls were used. The author found that when the source was unknown, AI-generated and human-written advice received comparable ratings (H1 not supported). Disclosure did not change overall trust ratings, but the author observed a fragile reduction in willingness to follow the advice that did not survive robustness checks; H2 received only exploratory support. These effects were not uniform. AI familiarity moderated the content effect: infrequent AI users gave the human-written text a substantial premium (d = 0.85), while frequent users evaluated both texts similarly. Gender moderated the disclosure response: women showed a consistent disclosure penalty, while men did not. The author recommends that career guidance services and policymakers account for AI experience and gender when designing disclosure policies, since uniform mandates likely produce non-uniform outcomes across subgroups. They should also measure not only stated trust in AI advice but also behavioral intention to follow it, since in this study disclosure affected the latter but not the former.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Butkus, Justas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD308317129&prefLang=en_US