In AI-assisted content production, teams may share a common tone and set of writing guidelines. These reference points create continuity across texts, but say little about how teams actually approach a subject. Their expertise is also reflected in the questions they choose to explore, the distinctions they consider important and the level of precision they expect from an explanation. For that expertise to remain present in the content, these choices need to be conveyed alongside the writing instructions, becoming part of the shared editorial reference points.
Treatment choices within shared editorial reference points
These choices clarify how a subject should be developed. The angle defines which aspect of the subject the content should emphasise. Professional terminology refers to the terms used to name the elements that the team distinguishes in its work. The level of precision, finally, determines which information needs to be made explicit for those distinctions to be understood. These instructions therefore concern the substance of the text. Two pieces of content can consequently use the same tone while providing different explanations.
In our article A Shared Editorial Framework, Adapted to Each Format, we show that shared reference points can remain consistent from one piece of content to another without preventing a subject from being organised and developed according to its purpose. Here, the focus is more specific: identifying the reference points that allow teams to preserve their own treatment choices. Which aspects of the subject do they want to emphasise, which terms do they use to describe them and how far do they want the explanation to go?
Consider an example to make these choices more tangible. A company decides to write a text explaining the response time for a technical support request, using the following angle: the distinction between stating a response time and establishing when that period begins. In other words, rather than simply explaining the stated duration, it also wants to clarify when that duration starts to run. “Indicative response time” is used to describe the stated duration, while “complete request” refers to the point from which that period can be assessed. The company also expects the text to explain what happens when information is missing, without presenting the indicative response time as a guaranteed response deadline.
Expressed as writing instructions, these points might read as follows: “Explain when the indicative response time begins to apply. Use the terms ‘complete request’ and ‘missing information’ in their respective meanings. State that, for an incomplete request, the starting point cannot yet be established.”
An instruction such as “write clearly and professionally” can accompany these points, but it cannot replace them. It explains neither why the starting point requires clarification nor which terms are needed to distinguish between the different situations. Expertise becomes transferable when substantive choices are explicit enough to guide the writing while still allowing each piece of content to retain its own wording.
How teams’ treatment choices appear in AI-assisted content
The difference between stylistically consistent writing and treatment that accurately reflects teams’ choices becomes very clear in the text itself. The two examples below concern the same response time. Both use a restrained register and address the same subject. Only the treatment instructions provided for the second make the distinction intended by the teams immediately apparent.
Development limited to stylistic consistency
A response time is provided for requests received. This information indicates when a response may be expected. If further details are required, they are requested so that processing can continue. The stated timeframe therefore helps explain the general response process while allowing for information specific to each request.
The passage is easy to read, but it leaves a central question unanswered: from what point does the response time begin? The reference to further details being required also does not explain how they relate to that starting point. A reader can identify the subject without seeing the distinction the teams intended to convey.
Development that reflects the treatment choices
The stated response time is indicative. It begins when a complete request is received, meaning one that includes the information required for it to be reviewed. When information is missing, further details are requested; the starting point for the response time cannot yet be established. This distinction clarifies how the stated timeframe should be understood: it separates the date on which a request arrives from the point at which all the required information has been provided.
In this second version, the chosen angle structures the explanation. The expressions “indicative response time” and “complete request” refer to two specific elements rather than simply giving the passage a more technical appearance. The text develops the relationship between them far enough to explain why, in this constructed example, the arrival of a request alone is not sufficient to establish when the response time begins.
This level of precision does not result from longer sentences or a more expert-sounding tone. It depends on the instructions provided about what the content needs to distinguish. If AI-assisted writing produces fluent wording but omits the starting point, stylistic editing alone will not restore the intended treatment. Review must also assess whether the intended distinctions are present and whether their meaning has been preserved. The CNIL also notes that AI-generated outputs may require human review.
Shared editorial reference points therefore provide a basis for assessing the substance of a proposed text: does it address the question selected by the teams, use the terms that express their distinctions and provide the explanation needed to connect them? In the example above, the answer is visible in the paragraphs themselves. It depends neither on the identity of the writer nor on uniform wording across several texts.
Conclusion
Shared editorial reference points preserve teams’ expertise when they specify choices of angle, professional terminology and level of precision. In AI-assisted content, that expertise remains visible when the explanation reflects the intended distinctions. Consistent writing can maintain continuity of tone; on its own, it does not convey what the teams have chosen to explain about the subject.
