This second post by Neil Selwyn and his collaborators shares some of the preliminary results from their research survey, which was discussed in an RW post earlier this year ‘Academics’ uses of GenAI‘. 

---

Image from Lance Chang | unsplash.com

“I think because it’s not openly discussed and yet it’s so easy to use, I feel guilty. I question if I’m cheating somehow? Or going to get caught?” (#116)

We ran a short anonymous survey between December 2025 and March 2026 focusing on ‘grey uses’ of GenAI – aspects of AI use that academics feel their colleagues and peers might well judge inappropriate. The survey attracted over 200 responses (a lot more than we were expecting) and we are now working through the data. As a start, here are some initial observations from our first passes of the data:

1. Different stances toward GenAI use

Respondents report a wide range of uses that might well be perceived as inappropriate, from minor cases of tweaking language through to drafting student feedback, writing substantive ‘bits’ of academic articles, drafting research proposals and producing complete ethics applications. Emerging across these different examples are a few distinct types of respondents:

• A surprising number of people took time to respond that their uses of GenAI were perfectly acceptable and should not be judged by others as problematic. These academics justified their GenAI uses as strategic and ultimately smart ways of working – “it makes many work processes a lot easier and faster” (respondent #133)
• On the other hand, a sizable proportion of respondents described a sense of inadequacy and “shame that you have to do this” (#6). These respondents framed GenAI as something that they did not want to be using in their work: “it feels like cheating, like maybe I haven’t done the work properly” (#37).
• In the middle, was a large group describing themselves as feeling pragmatically compelled to use GenAI. This was something that they acknowledged was not ideal but that they nevertheless felt sometimes pushed into doing, largely by the pressures of academic work. As one respondent put it: “There are plenty of slippery slopes in the use of GenAI, I feel. It starts with using it for small innocent tasks, but quickly (and as workload/stress increases), the incentives/ temptation for using GenAI grow” (#81)

2. Different drivers of academic GenAI use

As is often the case with surveys of academic work, lack of time comes up as a predominant factor driving respondents toward GenAI, often compounded by the stresses of precarious employment. Tasks such as marking student work and writing personalised feedback are seen to have become more onerous and intensive, with ‘AI proof’ assignment tasks requiring more feedback within ever “restrictive marking times (which are never enough)” (#128).

On the flipside is what some respondents saw as ‘toxic’ and ‘dehumanising’ expectations to be ‘productive’ and publish at scale. Respondents describe the ways in which GenAI is entwined with an ever-competitive scramble for jobs: “the academic job Market sucks and I simply don’t know how else I’m supposed to keep up … It’s awful.” (#80)

3. Positive and pragmatic justifications toward GenAI

Against this background, our survey throws us various justifications for using GenAI to assist with different writing and thinking processes.

For example, some respondents describe GenAI as an expedient way to off-load what they see as ‘empty’ writing tasks – what one respondent justified as “[the] outsourcing of tasks that were already nonsense to begin with. They were already the production of slop, so there’s little harm to my mind in outsourcing that slop production to an AI” (#73).

Our survey highlights various examples of this type of outsourcing. For example, it was reasoned that GenAI-generated feedback was appropriate when students were using GenAI to produce their assignments. Similarly, GenAI was acceptable when research tasks start to feel ‘boring’, ‘annoying’ and ‘professionally uninteresting’ (#8). Similarly, in terms of writing documents that are unlikely to be read: “I almost don’t care anymore. Why give up my weekend to create a resource that almost no-one will read?” (#49).

Another recurring theme is using GenAI for writing support, especially when not writing in one’s first language or when one feels prone to making errors or perennially struggling to find an appropriate voice: “By now, I should know the ropes right? However, writing never comes easy, my tone is often off (too critical, too informal, etc.) so GenAI is a great help” (#137).

We also found respondents who describe themselves using chatbots because they cannot find colleagues to talk through research and writing with: “I do not have colleagues willing to act as sparring partners, as they are busy with their own work” (#131).

In all these cases, GenAI tends to be presented as a justifiable source of assistance that compensates for lack of support elsewhere and helping academics work more effectively and efficiently. The key here is a confidence that these academics remain responsible for the final products:

“It’s still all my work. I give the ideas, edit it, work with the AI” (#135)

“It’s my work, I’ve just outsourced copy editing to GenAI rather than a person” (#157)

4. Existential fears around GenAI

 Other respondents, however, frame their engagements with GenAI in more cautious and sometimes more fearful terms. One source of conflict is contravening what students are being told about inappropriate uses of GenAI. Some academics describe “feeling really uncomfortable and hypocritical” (#96) about their own GenAI use: “[I] fear that my academic papers will, at some point, be marked as plagiarism” (#131).

These respondents tend to be reluctant to talk to colleagues about GenAI use for fear of being ‘judged’ (#52), and “lazy, stupid, surplus to needs, and unethical” (#182). Respondents talk about GenAI fuelling feelings of inadequacy, lacking expertise and being “lazy or intellectually inept” (#50). As one survey response put it:

“That I am an imposter, not able to create my own text and not as talented because I can’t seem to create anything without having it through AI. I don’t know how people did this before AI!” (#84)

Other responses highlight fears around “los[ing] my ‘academic imagination’” (#111), and “giving away my critical thinking abilities” (#122). Ultimately, however, respondents expressed frustrations at being stuck in cycle of reluctant GenAI use:

“I use it for the things I wish each semester that I had time to do a better job on and that I always tell myself I’ll do a better job of next time” (#46)

The need for further conversations

At the moment, a lack of openness around GenAI seems to be leading to wildly divergent professional practices around GenAI use and non-use. These emerging practices and norms are certainly not shared and likely to undermine trust, whether that’s trust in our colleague’s work and/or trust in our own academic abilities.

Crucially, our small survey suggests a split between those academics who feel that GenAI provides a justified boost to their working lives, and a lot of other academics who are clearly less comfortable with their use of GenAI. For this latter group, at least, GenAI is not something that is helping them work smarter or to feel better about themselves. Equally as important, although not the focus of our survey, are the voices of many others who are abstaining from making any use of GenAI at all.

These differences need to be better acknowledged and given serious consideration and reflection within academic departments, subject associations, and other scholarly communities. In a practical sense, this includes conversations around what uses of GenAI academics collectively might agree on as acceptable, if not advisable. On a more political note, we also need to talk about how GenAI feeds into wider structural problems within higher education … and how turning to GenAI is likely to compound (rather than alleviate) these problems.

Above all, is the concern looming throughout our survey responses that GenAI is clearly bound up in a range of work-related emotions. These include having an impact on some academics’ sense of professional self-worth; for others, extenuating the toxic nature of their working conditions or simply making GenAI users resentful towards others who they feel might judge them in a poor light.

However, we also see opportunities for more affirmatory conversations such as using the current moment as a prompt to build academic solidarity. Having these conversations needs to be a key part of mentoring and supporting early career academics. More senior colleagues need to talk openly about “how people did this before AI” and listen to junior colleagues’ experiences of how GenAI impacts on their workloads and sense of how they are coping with the strains of academia.

Academic work has never been easy and GenAI clearly taps into some of the base problems and faultlines in our workplaces and profession. It is crucial that we collectively develop a better sense of what place we want this technology to have in our work.

The first stage of addressing a problem is talking about it so, we look forward to hearing these issues talked about a lot more from now on!

——————-

This post was written by Neil Selwyn (Monash University), in collaboration with Marita Ljungqvist & Anders Sonesson (Lund University); Magda Pischetola & Lucas Cone (University of Copenhagen).