The paradox of productive irritation: mapping the stress–coping loop that sustains generative-AI engagement

Abstract

Generative AI promises academic efficiency yet often delivers flawed answers, leaving users “irritated but engaged.” To explain this paradox, we merge the Transactional Model of Stress and Coping with Cognitive Dissonance Theory and survey Indonesian and Taiwanese academics who had used ChatGPT for at least a month (N = 388). Partial Least Squares analysis shows that response failures and low AI literacy sharply raise frustration; frustration, in turn, both directly sustains continuance intention and indirectly does so through heightened resistance to change. The indirect route dominates in Indonesia, where higher switching costs foster inertia, whereas Taiwanese users convert frustration into exploratory recommitment. These findings re‑cast resistance as an adaptive buffer rather than a mere barrier and reveal culture‑specific coping paths that keep imperfect AI in daily workflows. The study advances stress‑and‑dissonance theory integration and guides institutions toward balanced strategies combining accuracy auditing, literacy scaffolding, and context‑sensitive expectation management.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.