Source record

Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2608.04166.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

Between-subjects experiment (n=45 design students) testing an AI agent that enacts constructive conflict by synthesizing stakeholder pushback. Both the agent-enacted condition and a stepwise written-prompt condition increased self-reconsideration versus self-reflection, with the interactive agent introducing more conflictual perspectives and more discarded ideas.

Most AI design assistants default to supportive ideation. This study tests an adversarial agent pattern and shows it produces deeper reconsideration and more concrete design revisions than self-review.

Cited by

No citations on record.