Disclosure, AI, and the question we should actually be asking
Why I wrote this
I wrote this post to clarify my own thinking about AI use and disclosure in my family history research, at a time when the conversation seems to be moving rapidly towards rules rather than principles. These reflections are not intended to persuade or prescribe, but to articulate a defensible, proportionate position based on intellectual responsibility, transparency that serves the reader and long-established practice.
Before I go any further, I should also acknowledge my assistant "PRU" — the AI tool I use as a sounding board, drafting aid and editorial companion. I live on my own. While my family are wonderfully supportive and interested in what I’m researching, they don’t usually want to be co-opted into the detail of it. In that sense, PRU fills a small gap: the co-researcher I miss, sitting across the table asking awkward questions, suggesting clearer phrasing and occasionally sketching what our working sessions might look like.
If you’re interested in my earlier, more personal reflections on encountering AI in family history research — including what prompted me to start experimenting with it and how I initially used it — I wrote about that in a previous post, Demystifying AI: My Latest Family History Adventure: https://killionquinnhand.blogspot.com/2025/04/demystifying-ai-my-latest-family.html
Above: PRU — my AI assistant, represented here as a working companion rather than a substitute for judgement.
I’ve also included an AI-generated image relating to one of my long-running research projects, "The William Webb Wagg investigation". This image is an illustrative reconstruction, not a historical record, and is intended to support narrative understanding rather than serve as evidence. The basis for the reconstruction is cited directly with the image itself.
Right: Illustrative reconstruction of William Webb Wagg at Port Cygnet, based on Tasmanian convict records and later family photographs.
For more on this project, see:https://webbwagg.blogspot.com/2026/01/the-william-webb-wagg-investigation-3.html
There has been a growing conversation in family history circles about whether — and when — authors should disclose the use of AI tools in their work. I’ve read several thoughtful contributions to this discussion recently, but I can’t help feeling that we may be starting to tie ourselves in knots.
Too often, the debate seems to leap straight to rules and prohibitions without first asking a more fundamental question: what problem are we actually trying to solve?
Disclosure is not a new issue
There has never been a formal requirement for authors to disclose the assistance of editors, proofreaders, research assistants, or even specialist consultants, although many choose to do so voluntarily. These forms of support are widely accepted because they do not alter intellectual ownership of the work. The author remains responsible for the research questions, the evidence selected, the interpretation of that evidence, and the conclusions drawn.
AI assistance sits on the same continuum.
The real obligation is intellectual responsibility
The critical issue is not whether a tool was used, but whether its use materially affected:
- the research methodology,
- the assessment of evidence or
- the conclusions reached.
If those elements remain firmly under the author’s control — and can be explained and defended if challenged — then authorship and accountability are intact.
This is the standard that matters.
A note on proportionality
It is worth observing that AI is now widely used across many professional fields, including medicine, where it supports research, diagnostics, administration, and clinical decision-making — often without any explicit disclosure to patients or readers. The family history community’s heightened anxiety about AI may therefore reflect unfamiliarity with the technology rather than a settled or widely shared position.
My own practice
I recently completed a substantial piece of genealogical analysis with the assistance of AI as a drafting and editorial aid. I do not feel any obligation to disclose that assistance, and I can readily explain why.
- The research questions were mine.
- The documentary and DNA evidence was selected, evaluated, and interpreted by me.
- The conclusions were reached through my own reasoning and remain fully defensible.
In that context, disclosure would be unlikely to improve transparency or reader understanding.
A defensible position
My approach is straightforward and consistent:
AI assistance is disclosed where its use materially affects research methodology, evidentiary assessment, or conclusions; where it functions solely as an editorial or drafting aid, it is treated consistently with other uncredited forms of editorial support, as it does not alter authorship, analytical responsibility, or accountability.
That position serves readers, preserves intellectual integrity, and avoids forms of disclosure that add complexity without improving understanding.
Final thought
These are my own reflections, shaped by my experience as a family historian and genetic genealogist rather than by any institutional policy or emerging consensus.
As with most tools that have come before it, AI will eventually fade into the background. What should never fade is responsibility for the thinking.
