본문 바로가기
  • Home

The Effect of Restrictive vs. Unrestricted Acceptance in Conversational AI on User Psychology and Behavioral Intention

  • Industry Promotion Research
  • Abbr : IPR
  • 2026, 11(3), pp.165~178
  • DOI : 10.21186/IPR.2026.11.3.165
  • Publisher : Industrial Promotion Institute
  • Research Area : Interdisciplinary Studies > Interdisciplinary Research
  • Received : May 27, 2026
  • Accepted : June 26, 2026
  • Published : July 31, 2026

Choi Woong 1 Kim Seung In 1

1홍익대학교

Accredited

ABSTRACT

This study experimentally investigates how the acceptance style of conversational AI (restrictive vs. unrestricted) affects users' psychological responses and behavioral intentions. A between-subjects A/B experiment was conducted in April 2026 with 200 Korean adults (Group A: restrictive AI n=100; Group B: unrestricted AI n=100). Group B reported significantly higher post-interaction negative emotions (t=-3.478, p=.001, Cohen's d≈0.49), consistent after covariate control (β=0.499, p=.004). Bootstrap mediation analysis (5,000 resamples) confirmed partial mediation by perceived empathy (indirect effect=0.134, 95% CI [0.020, 0.268], BCa [0.033, 0.289]). No significant differences in relationship substitution or continued use intention were found. These findings support the relational uncanny valley concept and offer implications for adaptive friction and transparency-based AI service design.

Citation status

* References for papers published after 2025 are currently being built.