Sleeping with the Enemy: AI and the "Authenticity Penalty"

As a writer, I have lots of opinions about AI and its use in writing. And those feelings have ranged from righteous rage to deep skepticism and back again. It made me wonder how readers—and other writers—feel about it.

So, I went looking for a little data—not vibes, not talking-head OpEds, and not your cousin Betty’s tea-leaf-reading-auntie. Research. Facts. Conclusions drawn from the scientific method. I found two. Here’s the scoop:

Readers are feeling their feels

A recent University of Michigan study (16 experiments involving an impressive 27,000 participants) found that readers rate writing 6.2% lower when they believe AI helped generate the text. Readers said AI-involved writing was less authentic and less worthy of appreciation.

The kicker? This bias applied even to human-written text when study participants were told it was written by AI. However, readers with higher AI literacy show less bias and sometimes even appreciation. So that begs an interesting question, is the bias truly about quality or is it psychological?

A different study from Tokyo University suggests the latter. Their research showed “AI involvement reduces perceived trustworthiness, caring, competence, and likeability of the ‘author.’” Why Because they felt AI-assisted writing lacks effort and human sincerity.

Readers care about the perceived humanity in the work more than the words themselves.

What about AI for research and world building?

Readers, however, don’t seem to mind when AI is used for research, fact-checking, or background authenticity. And most of the time? They don’t realize it was used to begin with.

The research indicates the “authenticity penalty” applies to whatever output is generated by the writer, and not to any of the tools used in other parts of the process. They don’t see using AI to learn about a city, culture, historical details, or how a process works as any different from using Google for research.

It boils down to this: Readers don’t mind if you use AI to learn about Jakarta. They do mind if you let AI write about Jakarta.

What does that mean for us chickens (I mean, writers)?

• A Catch-22: Ethically, transparency is good but can be risky commercially if readers feel the writing is “less real.”

• Nothing lasts forever: AI literacy reduces risk, so logic says that the bias penalty may lessen as AI becomes normalized.

And we’re in the middle of it.

People use AI all the time (how many of you used the AI summary the last time you did an Internet search for something?). They use it for research, grammar checks, and brainstorming… but they resist it when it is part of fiction, personal essays, poetry—essentially, anything emotional.

Regardless of where you stand on the AI spectrum, you can’t avoid the fact that we, as writers and readers, are in the middle of a remarkable cultural moment.

Whether we like it or not, we are redefining what creativity looks like and that is fascinating and terrifying all at the same time.

How do you feel about AI in writing and writing research?