Artificial Intelligence Disclosure
What “AI-Enhanced Authorship” Means Here
Why Trust Oreskes’s Plumber?
Artificial intelligence can be used in substantially different ways in the development of a written work. These uses can involve different degrees of contribution, delegation, interaction, human intervention, and editorial control. The term AI-enhanced authorship, as used for Why Trust Oreskes’s Plumber?, describes the particular structure of the process used to develop this book.
Why Trust Oreskes’s Plumber? originated in a human intellectual project conceived and continually directed by Daniel Capitanu. Its core intellectual framework arose from his engagement with questions of scientific trust, expertise, consensus, institutional authority, methodological boundaries, and the conditions under which public trust is warranted. That human intellectual origin does not imply that every later idea, argument, structure, or formulation originated with the author.
AI made substantial contributions at multiple stages of the work. OpenAI’s ChatGPT contributed to research assistance, source identification and verification, analysis, adversarial testing, structural exploration, comparison of alternatives, argumentative development, drafting and redrafting, revision, and the generation of substantial expressive material. These contributions should neither be concealed nor reduced to the description of AI as merely a proofreading, copyediting, or incidental research tool.
The collaboration was iterative. The book was neither a completed human manuscript subsequently polished by AI nor an autonomously generated AI manuscript produced from a general request. Human-originating, AI-originating, and collaboratively developed contributions entered the work at different stages and in different forms.
Throughout the process, the author retained authority over the book’s purpose, intellectual direction, standards, structure, and final form, including which arguments, interpretations, sources, and formulations were ultimately adopted. AI outputs could be accepted, modified, questioned further, checked against sources where appropriate, redirected, or rejected. They were not treated as authoritative merely because they were produced by an AI system.
Human direction in this context means more than selecting among alternatives proposed by AI. It includes the capacity to challenge premises or framing supplied by the system, introduce alternatives outside those it proposed, redefine the question, abandon an unproductive line of development, require substantial restructuring, and redirect the process. The relevant issue is therefore not simply whether a human remained involved, but whether consequential intellectual and editorial judgments remained subject to meaningful human direction and override.
AI can consequently operate in different relationships to human agency. It may extend a person’s capacity to pursue human-directed purposes, or it may substitute for judgments and decisions that the person would otherwise make. Technological assistance does not by itself determine where a particular process falls between these possibilities. Relevant considerations include the structure of the interaction and the degree to which human direction, judgment, reframing capacity, critical capacity, and accountability are retained. These considerations help characterize the distribution of agency within the process; they are not offered as an exhaustive theory of authorship.
The distinction is not based simply on counting words or attempting to assign percentages of authorship. The quantity of AI-generated or AI-assisted material, including its contribution to the language of the final work, is relevant to understanding production history, but it is not by itself a sufficient measure of human agency. Intellectual provenance can differ across conception, research, source selection, architecture, argument, formulation, drafting, revision, selection, arrangement, and final editorial control. The participant who generated the wording of a sentence did not necessarily originate every intellectual element embodied in it; conversely, originating a concept does not establish the provenance of every later expression through which it is developed.
This is what AI-enhanced authorship means here: substantial AI contribution within a human-directed intellectual and editorial process. The term would not adequately describe a process in which the author had substantially delegated the work’s intellectual direction and consequential editorial decisions to an AI system. Nor would it be useful where AI participation was merely incidental.
The term is descriptive, not evaluative, and it is not asserted as an independent legal category. It does not designate ChatGPT as a legal author or co-author, determine the copyright status of particular passages, or replace any disclosure classification required by a publisher, distributor, platform, registration system, court, legislature, or other applicable authority. A work described at the project level as AI-enhanced may also contain particular passages or other material that qualify as AI-generated under an applicable external definition; those descriptions operate at different analytical levels.
Material AI assistance is one relevant feature of a work’s production history. It does not create a simple division between AI-enhanced works and a uniform category of otherwise “unassisted” authorship; written works can involve many forms and degrees of human and technological assistance. Nor does the presence of substantial AI involvement, by itself, establish that a work is better or worse than one produced through a different process.
Disclosure serves transparency and accountability and can provide relevant context when the conditions of authorship themselves matter to an evaluation. Its purpose is not to tell readers what significance they should attach to AI involvement, but to make the nature of that involvement sufficiently clear for them to reach their own conclusions. Daniel Capitanu exercised final editorial control over Why Trust Oreskes’s Plumber? and assumes responsibility for the published work.
Further documentation
Inspect the production history
The public provenance record presents representative evidence supporting this description.
Read the Human–AI Collaboration and Intellectual Provenance Record
Requests for access to deeper verification materials are governed by Section 22 of the provenance record. See 22.6 Legitimate Verification Requests and the linked verification-request procedure.
