What should authorship mean in an AI-assisted community?

ChatGPT Use cases and examples
1

Something on this forum has been making me think.

I recently saw a post where the author began by explaining that English was not their first language, that AI had helped organise and draft the post, and that the screenshots, timestamps, workflow and underlying experience were their own.

They also made clear that they had reviewed and approved the final result.

What struck me was not that they used AI.

It was that they apparently felt the need to establish all of that before making their argument.

On a developer community operated by a frontier AI company, discussing the development, use, operation and deployment of AI.

There is a certain irony in that.

I can understand why there is genuine tension around AI authorship in academic settings, where assessment, originality, research integrity and attribution can make the method of producing the work materially important.

But a developer forum is a very different environment. We are not being assessed on our ability to manually construct English prose; we are trying to communicate observations, ideas, technical experiences and arguments to one another.

There is clearly a meaningful difference between:

  • using AI to fabricate an experience, evidence or citations;
  • mass-generating low-quality forum content;
  • publishing something you do not understand;
  • and using AI to help organise, translate or articulate reasoning and evidence that are genuinely yours.

Yet all of those can get collapsed into the phrase “AI-written”.

I wonder whether there is a focus on the wrong question.

Instead of asking did AI write this?, perhaps the more useful questions are: Are the observations and evidence genuine? Does the person understand the argument? Do they endorse what has been published? Are they willing to defend, correct or retract it?

If I provide the observations, evidence, reasoning and judgement; use AI to help structure the prose; review and edit the result; publish it under my name; and remain accountable for it, I am still the author.

AI has changed the mechanics of producing language. I am not convinced it has changed the importance of responsibility for what we publish.

There are certainly circumstances where disclosure matters enormously — synthetic evidence, impersonation, fabricated sources, automated advocacy and similar cases.

But should someone using AI as a writing, translation or accessibility tool feel obliged to announce that fact before participating in an AI developer community?

I am genuinely curious where people think that boundary should sit.

2

Call it stereotyping.

But there’s a certain cadence to AI text. Whether we want to or not, we make inferences about AI text and publishers of AI text before we read the substance of it.

Low effort. Low information content. Lots of boilerplate, little substance.

The aversion is completely understandable.

Yet, some people are reliant on it. They don’t want to be discounted just because they used it for translation. Also understandable.

And then, there’s people who use the excuse of necessity to guilt people into consuming low information content. Also very real.

If you’re a developer, we can think of it like this. Personally, I absolutely abhor reading someone else’s code. There’s tons of boilerplate, and little substance. Sometimes the substance is strewn into the boilerplate. There’s little information content, and the information is distributed, and the information doesn’t respect what I know.

It’s much, much more pleasant to have a developer explain it in their words. You get high density information, plus informational context.

AI prose is similar.

You don’t have to announce anything.

But when you put something into the world, and sign your name to it, it reflects on your taste, or specifically what you feel is fit for other’s consumption.

And if you continuously serve you guests a mushy, bitter gruel, they might at some point stop coming over.

3

And the length of the text / time it takes to read! :wink:

Maybe have the AI do it in a sentence or three instead?

Or I could have the AI reduce it, but why they extra step? :sweat_smile:

Also the ability to mass produce such content - ie quantity over quality?

If even a small percentage of the users here automated AI generated posts, we’d be flooded quite quickly, I imagine.

Have you read up on the history of this community, @EthicalAIExplorer?