Generative AI and writing

How does generative writing intersect with generative AI?


Eleanor C Sayre


Generative writing helps you generate ideas. Generative AI takes prompts from you, then generates text. How might using generative AI intersect with generative writing?

What purposes does generative writing serve?

Generative writing helps you generate new ideas in two major ways. In the act of writing, you record your thinking at this point in time, document what you’ve done, and your current reaction to it. In the act of reflecting, generative writing helps you think again about what you thought before, notice how your thinking has changed, and expand on (or amend) your prior work.

Generative writing helps you publish more papers in two major ways. In the act of writing, you practice articulating your ideas into words and develop your voice as a writer. In the accumulation of writing, you build a library of documentation about your project, from your ideas about theories to the details of what you did and how.

As we consider how to use generative AI with generative writing, we need to think about these two purposes (generating ideas and publishing papers) as well as these three activities (writing, reflecting, and accumulating).

AI and deep expertise

In the topics that you have deep expertise, you’ll notice all the ways that AI-generated text isn’t quite right. However, the farther you stretch from your deep expertise, the more alluring and persuasive AI-generated text becomes. This is a problem because generative writing helps you grow your expertise, so you’re operating at the edges of your ability to detect AI’s limitations.

Is the text the point?

Well, it depends.

The point? When and why.
Yes. Sometimes, you need the words on a page for documenting your ideas or actions, or so that you can reflect on them later. Your prose should be clear enough to read later.
No. The act of writing can cause your ideas need to coalesce, so you’re writing to figure out what you care about and how. Your prose could be terrible, and that’s ok.
Table 1: Is the text the point? Yes and no.

Generative AI is great at producing a lot of words quickly, and the prose is reasonably good. It’s terrible at documenting what you’ve done (because it doesn’t know), but it can be a good thought partner to help you reflect on your ideas or clarify what really matters to you.

If you think having readable text is the point of writing, it’s really tempting to use AI to get words on a page. For some people, there’s a huge stumbling block at the beginning of writing. The blank page stares at them and they struggle to get started. Asking AI to fill the page so that they can edit the words can jumpstart their writing and help them articulate what matters.

Other people really struggle in the editing process. Perhaps they don’t write enough and AI can expand their sentences into paragraphs. Perhaps they struggle with flow and AI can smooth out the language. Perhaps they write too much and AI can help condense or reorganize.

More specialized topics means less helpful AI.

As your topics become more specialized or technical, AI becomes progressively worse at helping. It’s better at writing a few paragraphs of marketing copy than it is at writing a scientific abstract. It struggles with literature reviews that delve beyond a few seminal papers. AI will confidently report what you should do, but don’t mistake confidence for truth.

Where do ideas come from?

If the text is not the point, then we need to think about ideas. Where do ideas come from?

Because research is a communicative and collaborative act, a natural part of doing research is talking with other humans about what you’re doing, why it matters, and how to proceed. Generative writing helps you have this dialog with yourself; generative AI can act as a thought partner to help you clarify and expand on your ideas.

Generative AI is not a collaborator.

AI doesn’t have the deep experience in your context that your local collaborators do, nor the field-specific knowledge that your peers and mentors have. You can use it as a thought partner, but not your only partner. You still need other humans.

AI is poor at prompts like these:

  • I am stuck on
  • What matters to me is
  • I am really interested in

These prompts are about you: what do you care about? Your generative writing on these prompts is about identifying and articulating your interests and stumbling blocks. The text is not the point; the thinking is the point.

AI can be good at prompts like these:

  • What is a general idea for
  • What are common pitfalls for
  • How could I rephrase this

These prompts can help you explore new ideas, orient you to new projects, and clarify or reframe your writing.

AI may be forbidden for data analysis.

If your data include confidential information, then uploading it to a generative AI may break confidentiality. If your data are covered by an IRB which doesn’t mention generative AI analysis, you probably can’t use generative AI. If your data are about students or student records, your university legal team can advise about how federal and state law (as well as university policy) apply.

When in doubt, just don’t.

In the US, research on humans needs to be reviewed for ethical considerations by each institution’s Institutional Review Board (“IRB”). It’s pretty common to use “IRB” and “ethics board” interchangably.

As a thought partner, generative AI can help you iterate through your ideas, refine how you want to talk about them, and suggest to you general directions for future work. Having a dialog with generative AI can be productive. However, because generative AI can produce so much text so quickly, it’s easy to mistake “having a lot of text” for “doing a lot of thinking”. Be mindful about what you’re asking it to do and why.

Back to top


This article was first written on October 21, 2023, and last modified on November 21, 2023.


For attribution, please cite this work as:
Sayre, Eleanor C. 2023. “Generative AI and Writing.” In Research: A Practical Handbook.