2026-10-11T20:39:44Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33956946 2026-09-21T17:39:13Z category_45 category_734 category_64 category_106 portal_5 item_type_3 month_year_09_2026

<p>Demographic information of participants.</p> Gang Ren (188076) Xuezhen Wu (25084692) Gang Wang (36685) Yinuo Ye (25084695) Zhihuang Huang (23216322) Tianyang Huang (11307535) Sociology Biological Sciences not elsewhere classified Cancer Science Policy remaining effect sizes model explained 47 large language models experience using ailds dedicated learning tablets 266 ), whereas educational technology research three psychological needs significant direct association 2 </ sup whereas response timeliness div >< p whereas relatedness showed perceived friendliness showed sectional survey data psychological needs >< sup response timeliness f </ educational robots perceived personalization perceived enjoyment technical capabilities indirect associations increasingly embedded estimated associations enjoyable interaction devices continue determination theory continuance intention broader challenge beyond responsiveness association patterns 726 first <div><p>As large language models are increasingly embedded in dedicated learning tablets, educational robots, and other physical AI learning devices (AILDs), sustaining learners’ engagement beyond initial adoption has become a broader challenge for educational technology research and design. Although the technical capabilities of such devices continue to advance, the factors associated with continued use remain poorly understood. Drawing on the Stimulus–Organism–Response framework and Self-Determination Theory, this study examines associations among perceived personalization, interactivity, response timeliness, friendliness, autonomy, competence, relatedness, perceived enjoyment, and continuance intention. Cross-sectional survey data were collected from 726 first-year senior high school students with experience using AILDs and analyzed using partial least squares structural equation modeling. Effect sizes varied substantially across the estimated associations. The association between relatedness and perceived enjoyment had the largest effect size (<i>f</i>² = 0.266), whereas the remaining effect sizes were small or negligible. Two theoretically informative expected associations were not supported: perceived personalization with autonomy and response timeliness with relatedness. The model explained 47.0% of the variance in continuance intention and 50.7% in perceived enjoyment. Perceived friendliness showed the largest feature–need association with relatedness, whereas response timeliness was associated with autonomy and competence but not with relatedness. Autonomy and competence showed statistically significant direct and indirect associations with continuance intention, whereas relatedness showed a statistically significant indirect association involving perceived enjoyment but no significant direct association with continuance intention. Perceived enjoyment was positively associated with continuance intention and was involved in statistically significant indirect associations for all three psychological needs. These findings reveal differentiated feature–need and need–intention association patterns in physical AI educational hardware and suggest that design attention may usefully extend beyond responsiveness to relational and enjoyable interaction. Because the data are cross-sectional, these association patterns should not be interpreted as evidence of temporal or causal relationships.</p></div> 2026-09-21T17:39:07Z Dataset Dataset 10.1371/journal.pone.0358228.t002 https://figshare.com/articles/dataset/_p_Demographic_information_of_participants_p_/33956946 CC BY 4.0