Answers are grounded in this paper and cite their sources.
What happens when participants in sustained human–AI research begin to investigate not only their results but the methods and records that produce them, and how can we tell who initiated, authorized, interpreted, or implemented a change? We reconstruct a bounded participant case from contemporaneous engineering artifacts, recovered March–April 2026 messages, 24 preselected source notes, and 118 timestamped messages from June 7–9, 2026. The record shows a human operator renegotiating an AI advisor's working discretion; AI advisors proposing relations, changing local criteria, and formulating a new higher-order research question; a human-initiated challenge to the method directed at a second AI role; and participants reinterpreting their own procedures. We analyze participant-performed infrastructural inversion, making the apparatus an object of inquiry, separately from operational method revision, documentably changing practice. An April method inquiry involved both; a June discovery registry changed admission procedures for mainly practical reasons; a June reinterpretation of earlier engineering work as research infrastructure displayed inversion without a demonstrated operational change. A hypothesis advanced by an AI reviewer during manuscript development, that higher-level transitions followed human authorization, was not cleanly falsified on its literal terms; however, the stronger reading that research ideas originated with the human is inconsistent with the recorded distribution of proposals, interpretations, and decisions, which retrospective accounts also compressed. We contribute a provenance-sensitive account of differentiated initiative and of participants' scrutiny of their own research apparatus, without estimating prevalence or inferring a general mechanism of autonomous human–AI system evolution.
No takes yet. Share an insight, caveat, or question.
Jia Shen (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: