Twelve developers inside a coordinated AI devtool rollout name a management expectation gap the acceptance models miss

arXiv·low signalXBlueskyLinkedInCopy link

arXiv 2609.20977 (17 Sep) reports a case study of a coordinated AI devtool rollout at a large Swedish telecommunications company, with semi-structured interviews across three sites analyzed through process coding and thematic analysis, interpreted post hoc with TAM2. Findings on use cases, productivity, frustrations and tool limitations corroborate existing survey work, but the interviews surface two things surveys do not: a management-developer expectation gap mapping onto TAM2's subjective norm and voluntariness constructs, and the heavy weight participants place on perceived risk, a factor TAM2 and comparable acceptance models do not represent at all. The tools land as helpful but fallible assistants whose value depends on organizational expectations, system scale and developer skill.