Introduction

At present, UK universities are seen by academic staff as largely inhospitable and impoverished places of work, where academics are subject to a toxic corporate audit culture (Shore and Wright 2000; Smyth 2017; Watermeyer et al. 2023; Welsh 2021); low reward and recognition for tasks with limited or non-obvious positional return(s) and marginalising effects (cf. Cardozo 2017); depreciating remuneration (for all bar managerial elites) (cf. University College Union 2022); work precarisation (University College Union 2020) accentuated by digital transitioning (cf. Ivancheva and Garvey 2022); and endless marshalling by ubiquitous datafication and surveillance technologies (Gourlay 2022; Jandrić 2020). Trust in university leadership appears to have hit an all-time low (Watermeyer et al. 2022), while industrial action has become a normalised aspect of the academic year, such is its frequency and failure to find any source of resolution, never mind restorative justice.

Deterioration in academics’ working lives is argued to have peaked during the recent Covid-19 pandemic, where the physical closure of university campuses and an emergency transition to remote and digitally enabled forms of working is claimed to have further intensified an already unsustainable work burden (Watermeyer et al. 2021). The affordances of technology, especially videoconferencing platforms and the sharing capabilities embedded in LMS/VLE platforms, social media, and multivalent platforms such as MS Teams and Google, made it possible for academic work to continue during the pandemic emergency (Dalipi et al. 2022; Hughey and Kirk-Jenkins 2021; Mitchell 2023; Moorhouse and Kohnke 2021). Clearly the trauma of living through a global pandemic made teaching learning, and research experiences challenging if not impossible in many situations (Dunn et al. 2022; Okeke-Uzodike and Gamede 2021; Taylor and Frechette 2022).

It is also the case that digital tools and platforms opened up academic practice and access to people who had been chronically under-served in universities, including racially minoritized people, people with disabilities, first generation students, the economically vulnerable, and people with caring responsibilities (McLay Paterson and Eva 2022). But those experiences gained during emergency remote teaching do not appear persuasive enough to embed those inclusive digital practices across the sector. In many cases, preexisting inequities in the sector were amplified by the Covid emergency response (Arday and Jones 2022; Cho and Brassfield 2023; Czerniewicz et al. 2020; Flynn and Noonan 2020; França et al. 2023; Francois et al. 2023; Górska et al. 2021; Lekchiri et al. 2022; Hadjisolomou et al. 2022; Njoku and Evans 2022).

At the point of writing, 3 years after the initial arrival of Covid-19, compassion fatigue and the economic priorities of institutions are driving the snap-back (Bryant 2021, 2022) to less digital mediation and more physical presence in buildings on campus. Accommodations that helped people manage workload during the height of the crisis are being rescinded, and the expectations are that we as a sector go ‘back to normal’ (Carr 2021; Lewis et al. 2021; Price et al. 2022). The co-existence of these expectations with the reality that many have left academia, because they died, because they quit, or because their department was eliminated during a budget emergency that came to a head during the pandemic, has meant for an increasingly jarring subjective experience of academic work.

Correspondingly, reports of academics experiencing chronic fatigue and decline in physical and mental health (Gewin 2021; Morrish 2019) have become more commonplace as has a trend of capitulation and them abandoning university posts (Mitchell 2022). While the pandemic produced opportunities for positive critical reflexivity for some academics (cf. Jandrić et al. 2020, 2021a, b, 2022) the resilience of hegemonic institutionalism within universities has rendered the pandemic no more than an interstice to the ‘normal’ order of things. Academics have thus tended to refocus on the degradation of life in universities and a legacy of personal and professional injury.

Much as many disconsolate workers in other job sectors, many academics have sought and continue to seek, where so able, ‘better’ forms of employment outside of UK higher education and/or in other job sectors. For those without an exit strategy and forced to remain in their university posts, a trend of ‘quiet quitting’ has become prevalent (Forrester 2023). The persistent experience of trauma is layered upon pre-existing harms caused by the neoliberal logic informing much of academia, globally. The presence of audit culture in academia (Shore and Wright 2000), with its quantified notion of value and its enforced scarcity of resources, is not new, but the moment of the Covid-19 emergency and aftermath provided another moment for the regime to strengthen, in the absence of revolutionary pushback (Welsh 2021).

Much as in the USA, Australia, New Zealand and other elite higher education settings, UK academia has found itself reformed by the emergence of neoliberalism in the 1980s as the de facto policy settlement for public institutions (cf. Maisuria and Cole 2017). Universities’ penetration, or be that predation, by new public management technologies as antidotes to the assumed profligacy and inefficiency of collegial governance and ambivalence concerning the public value of higher education, is argued to have induced a step-change in academics’ focus. Their primary role as producers and facilitators of new knowledge would be challenged by a corporatist vision of the university that emphasised cost-effectiveness, economic function (and frugality) and the efficacy of their contribution as determined by the scale—ostensibly more so than quality—of their productive output.

Academia would surrender to and find itself bedazzled by the seductions of quantitative mesmerisation (Watermeyer and Olssen 2016). Its value has been reduced to an expression of numerical worth declared immodestly—lest anyone should not hear—as a triumph of rankings (Brankovic 2018; Brankovic et al. 2023; Hazelkorn 2011) and technical rationality (cf. Feenberg 2017). Performance-related anxieties of being in any way ‘less’ than one’s colleagues and therefore at risk of managerial opprobrium, would hasten academics towards a treadmill of one-upmanship. A justification of professional identity forged through quantitative superiority would prevail. Tools of metricisation would not only service productivity monitoring but, in fitting with an ‘image boosting business’ (Alvesson 2013: 95), aid and abet academics’ personal brand management.

Today, many academics draw their self-concept—and license to participate—relative primarily to their acquisition of research publications; research citations; research income; research partners; doctoral students; social media followers; public and policy stakeholders; patents and other such tangible quantitative representations of productive output and its influences. Even teaching, treated by many academics as a second order function some way behind research, is dominated not by a commitment to pedagogical innovation but what Hayes (2021b) calls a ‘McPolicy’ of measuring excellence.

Indeed, the present ubiquity of performance evaluation systems in the UK that (bean)count academics’ scholarly contributions and more recently even their public impact, speaks to the unmitigated success of academic labour’s metricisation and the collusion of institutional and scholarly leaders (the professoriate are unabsolved). This takes place within a system of ‘competitive accountability’ (Watermeyer 2019) wielded in appeasement of their political paymasters and their own seemingly rampant insecurities (Hayes 2021a). Yet more than this, academia’s productivity mania reveals the extreme vulnerability faced by those either less able, unable, or unwilling to game the abacus and the inevitability of their compliance no matter how ideologically anathema.

What, therefore, appears an insatiable desire among academics to collect, curate, and sometimes garishly parade measurable outputs as markers of esteem, is core to their survival in a hyper-competitive and ever accelerating work regime. Each new acquisition, however, offers only temporary respite from the endless torrent of performance demands made of academics and fails to mask the cracks to their resolve caused by endless subjection to peer-review and the ineluctable threat of judgements, which, more often than not, declare their work is simply not good enough. While rejection is a quotidian aspect of academic life – be it for instance in the form of an article rejected by a journal editor or a grant proposal rejected by a funding body – its devastation to morale and the resilience required of academics, if they are to survive a life of hyper-competition, is far from trivial. Academics necessarily, therefore, commit themselves to its avoidance or otherwise devise forms of compensation that alleviate the severity of its blows.

Most often this transpires with academics upping their output in an effort to offset their losses. The spectre of rejection may also stimulate risk-aversion and self-censoring among academics as individualists, trapped in the mould of homo economicus, who neglect the affordances of peer production and collective intelligence (cf. Peters and Jandrić 2018). Academia as an intellectual project is consequently diluted both by a flooding of output and a superfluity of prudence. Notwithstanding, a production line of academic output shows no sign of abating. In fact, recent technological developments are ensuring it will gain pace and in so doing further throw into doubt the contribution of academics.

In this article we focus on the effects of Large Language Models (LLMs) as a type of generative artificial intelligence (GAI). With the publication of ChatGPT in November 2022, GAIs have unsettled the UK’s higher education community and raised profound questions of what counts as authentic ‘authorship’ (Thorp 2023) with vast implications for scientific integrity and (mis)conduct. LLMs landed in a moment of acute Covid fatigue, while the pandemic was (is) still very much a global concern, and in particular in the context of neoliberal logics and audit culture shaping higher education institutional priorities. This allows us the opportunity to witness the impact of that convergence on the practices (and priorities) of academic staff. While considerable attention has been given to the (mis)use of LLMs by students and the potentially corruptive effects of natural language processing when applied as a substitute for human content creation and the generation of work subjected to summative assessment (cf. Cassidy 2023)—causing universities in some higher education settings to ban its use (Yang 2023)—knowledge of their use by academics is largely scant.

The transformational potential of LLMs for academic practice, and for academic researchers particularly, is asserted widely (Dwivedi et al. 2023), yet is also simultaneously declaimed as hype (Jandrić 2023). As an attempt to better understand the use of GAIs by UK academics as a ‘workforce in crisis’ (University College Union 2022), we designed and distributed a national online survey in June 2023. Survey findings elucidated a breadth of ways GAI is being used by academics yet homogeneity in how they rationalise its benefit as a labour accelerator that might soften the terms of their precarity.

We argue, however, that academics’ application of digital tools for the purpose of productive efficiency and gain, risks the further discombobulation of their collective identity and purpose and the further waning of their critical agency. We relatedly consider the potential threat of GAIs to the resilience of academics in a milieu of pronounced work intensification, where automated efficiencies produce opportunities for new forms of work exploitation. We also show how GAIs are being utilised opportunistically by academics, both in accommodating the performance demands of institutional managers and also subverting their authority by helping them to pursue professional interests that fall outwith their contractual remit. GAIs are therefore, discussed as a potential means of reclaiming academic autonomy through the reorganising and reclaiming of academic labour. We do not offer solutions, but rather reflect on the potential of slow scholarship (Berg and Seeber 2016; Wahab et al. 2022) and the praxis of refusal (Simpson 2007) catalysed by GAIs in re-engaging scholarly craftsmanship (Sennett 2008) and disrupting a neoliberal insistence on zero sum games and quantification.

Methods

To understand the use of GAIs by academics in higher education, an anonymous online survey was developed and distributed. The survey was launched in June 2023, 7 months after ChatGPT was originally launched, and remained open for 2 months.

The target population for the survey was staff working in UK universities both in academic and professional services roles. The survey was distributed via professional mailing lists, social media, and other online platforms. This convenience sampling method was not designed to capture a representative sample; rather, data was sought to illuminate general patterns and trends characterising the current use of GAI tools in UK university settings. Our research thus offers no more than a snapshot of current trends and perspectives on the use of GAI, yet a useful platform for ongoing and future research that explores commonalities and divergence in its application as may be found in different role-based, disciplinary, service-related, institutional, and sector contexts.

In total n = 428 participants responded to our survey, 284 of whom identified as ‘academics’. It is this testimony which we focus upon within the following discussion. Responses from staff working in professional service divisions within UK universities (and typically in non-academic roles) were removed from the sample post hoc. Their testimony is equally important, especially given the diversity of their roles and the variety of ways GAI might be applied and the impact of such, and is deserving, therefore, of separate analysis for future presentation.

The online survey was designed and distributed via Qualtrics.Footnote 1 The survey consisted of demographic and occupational questions; closed-ended questions asking about the respondents’ use of GAI tools; and open-ended questions exploring how GAI is being used by respondents and the perceived impact of its use on their work. Descriptive statistics were employed to define overall trends in the population and to frame our analysis of the qualitative data. Open-ended questions were thematically analysed (Braun and Clarke 2006). Responses were read and coded by an initial researcher before being validated by the whole research team using Dovetail.Footnote 2

Table 1 provides demographic information of the academic participant group. The vast majority of respondents were based in English and pre-1992 universitiesFootnote 3 and were employed on full-time and open-ended contracts. The ‘Associate Professor’ category was most represented among respondents. All disciplinary categories were represented within our sample. While there was no significant statistical variation in these, the three most populous disciplinary groups represented within the sample were ‘Education’ (11.9%), ‘Social Studies’ (10.9%), and Business and ‘Administrative Studies’ (8.09%).

Table 1 Respondent demographics
Full size table

Findings

Table 2 shows the proportion of respondents who reported using GAI tools; whether GAI tools are changing how they work; and whether they anticipate using GAI tools more or less in the future. Within our sample of n = 284 academics, there was a roughly even split between those using and those not using GAI tools, with a tiny majority associated with those using GAI tools for work-related purposes. Despite almost half of our respondents stating not using GAI tools, over 70% of respondents stated that GAI tools are changing how they work. A further 83% of respondents stated that they anticipated using GAI tools more in the future. Less than half a percent of respondents stated that they anticipated using GAI tools less.

Table 2 Participant use of GAI
Full size table

For each of these three questions, respondents were offered an opportunity to provide open-text insights that would help to contextualise their answers. This qualitative data provides the bulk of our analysis. Three core thematic areas emerged from our close reading and coding of open-text responses, which we now discuss.

De/Stabilising Status and Value(s)?

The cultural world of academia described almost 40 years ago so exactingly by the French sociologist, Bourdieu in Homo Academicus (1988) is little changed. It endures as a world of rigid stratification and social delineation sculptured by the extremities of capital the wealth or poverty of its acquisition. It persists as a world where power by most academics (and in a variety of explicit and more subtle though no less strenuous ways) is chased, rarely sacrificed and distributed with prodigious asymmetry. It is a world as such, which provides impunity to an elite minority while a majority subaltern struggle to articulate and defend their worth.

Status games are ever an indivisible feature of academic life and obligation for those who seek professional longevity. Where academics are inattentive to the cultivation of status, achieved most readily by hoarding, sometimes ingenuously, artefacts of scholarly distinction—their productive outputs—they are vulnerable to the inclemency of those who judge their contribution. Yet, in a cultural world where the qualification of labour as ‘excellent’ has become so ubiquitous as to be meaningless, achieving professional distinction may lead to a reenergisation of more overtly discriminatory appraisal and new criteria of what counts.

In the milieu of academics’ wider use of GAI, a reconsideration of productive excellence appears underway. Much as the world of student assessment has been upset by the application of GAI in misrepresenting learner competency (Cotton et al. 2023), its use by academics as a tool of productive efficiency, raises similar concerns of inappropriate and unethical use. GAI may also be understood to threaten hierarchies of power within academia and destabilise ‘the order of succession’. This may be achieved by mobilising the celeritas of those who want to ‘cut corners’ (for example, by importing into the university field properties or powers acquired on other terrains). Gravitas, the healthy slowness which people like to feel is in itself a guarantee of reliability (in writing a thesis for instance) and the most authentic proof of obsequium, unconditional respect for the fundamental principles of the established order’ (Bourdieu 1988: 87).

In the absence of a regulatory framework for GAI, and the present frailty in how academics distinguish the quality of their contributions within an ocean of outputs, there will emerge individuals, as within our survey, who will assert their own academic capital through the denigration of others. Specifically, our survey results point towards the valorisation of academic labour through non-use of GAI, where GAI is understood to conceal the limitations—and thus lower status—of academics as researchers whose output may be harder won or slower. We find for instance statements like:

Some colleagues who are not particularly good writers use these tools to speed up the writing process. (Professor, Post-1992 University)

GAI tools in this context are interpreted not only for disguising the deficiencies of those who find academic writing challenging but for supporting an accelerated production line of research. For some, GAI tools offer alleviation from performance pressures and enable less prolific researchers to emerge from the penumbra of high output authors. GAI may serve accordingly as both a research accelerator and status equaliser. However, when used to scaffold (s)lower output producers in closing a performance gap, and thereby helping them to gain parity of esteem in the company of ‘high-producers’, GAI may be found to erode the credibility to claims of research excellence, while also simultaneously undermine the efforts of high-yield researchers.

Unsurprisingly, therefore, we found survey respondents speak defensively to concerns of research quality sparked by academics GAI use. Some, for instance, recommended that disclaimers be added to research undertaken with the aid of GAI tools:

I would not be surprised if we have disclaimers on final products that explain how much was generated by AI and which human set it up and how and who checked it. (Senior Lecturer, Post-1992 University)

Others among our sample were found to stridently assert their scholarly credentials, claims to authorship and creative autonomy in reference to abstaining from GAI. Some even suggested that GAI tools are superfluous to academia as intellectual labour.

I am a good writer and prefer to write my own material in my own way. (Professor, Post-1992 University)

I don’t know what I’d use it for. I write my own papers/lectures etc. (Professor, Post-1992 University)

No need. I’d rather use my own brain. (Lecturer, Post-1992 University)

I can't exercise my social and cognitive presences if I’m not the one doing the writing. (Senior Lecturer, Pre-1992 University)

Resistance to the use of GAI tools by academics was also seen by one of our respondents to reveal the extent to which academic labour has become intellectually impoverished. LLMs are witnessed from this purview as a further threat to professional self-concept where academics’ working lives have become so intertwined with higher education’s bureaucratisation and its performative vacuum:

I see substantial resistance in the university sector among my colleagues towards it - which I largely fail to understand. I don't want to spend my time editing. I want to spend my time pushing forward science. I think too many academics are threatened because what they largely do is editing. (Professor, Pre-1992 University)

We also find, perhaps rather uniquely in the space of responses which are otherwise conspicuously unreflective of the political economy of UK higher education, intimations of GAI as a catalyst of rentiership and assetisation in higher education (cf. Komljenovic 2021) and the potential for appropriation by techno-solutionists and/or digital capitalists (Sadowski 2020) far removed from aspirations of the university as a site of critical pedagogy (Giroux 2011):

I’m concerned it will be seen as a way to save money and resources in a challenging HE sector and we are going to hand over the power to influence learning through all stages to ‘tech bros’ who don’t really seem that committed to human values and development, critical thinking, self-reflection, reflexive practice. (Position not reported, Post-1992 University)

Relatedly, respondents commented on the necessary investment of GAI within university curricula and as part of students’ higher education, in part to exploit its various affordances but also as a means of reasserting the intrinsic value of human capability:

Our students need to graduate as masters of the technology, regardless of discipline, understanding its positive uses, its limitations and that true human inspiration and creativity lies beyond. (Professor, Pre-1992 University)

A Clearing Space—But for More Meaningful Work?

For many of our respondents, the contribution of GAI is more to menial than cognitively complex and/or challenging tasks. We also find value judgements about academic functions that might reasonably be considered important, but that particular respondents deemed unworthy of their personal attention and offloaded to GAI. How academics use GAI tools reveals what they believe is important to their role and what is not, even in the context of tasks (conflated below as administrative) that require and inform experiential and reflexive praxis:

It takes some of the pain out of admin - little things like health and safety stuff, or ethics, or summarizing reports. (Senior Lecturer, Post-1992 University)

We find further evidence in respondent accounts of how academics interpret work burden and how these interpretations make a mockery of what might be construed as meaningful and virtuous aspects of academic labour:

Myself and two colleagues have been told to do a PGCert because the Head of School wants everyone to have a Fellowship thing. It’s easier to get the tool to generate some of the reflections, and stay in the word limit than spend hours thinking about something that is just a tick on a spread sheet for the University. (Senior Lecturer, Pre-1992 University)

Thus, while there are those in the techno-agnostic camp of scholarly purists who decry the facilitations of GAI, there are others who appear to be actively exploiting its tools not so much to scaffold the production line as to divert it by creating clearing spaces for forms of labour they consider to be more ‘meaningful’ or even more personally ‘profitable’.

Analogously, GAI is advocated as a means for rehumanising academic labour which in turn serves to distinguish functions that may be automated from functions that are ‘authentic’ and bound in personhood:

In the wider world, computerisation and automation is often used to eliminate drudge work that is not fulfilling, creative or dignified. Ultimately, that’s how we should be using GAI . . . Let the machines to the data entry and copywriting and leave academic and interpersonal work to real people. (Senior Lecturer, Pre-1992 University)

For some, the contribution of GAI tools goes further than alleviating labour burden and offers a bonfire of bureaucratic tasks, a wholesale clearing out, which in turn raises further questions as to what academics view as the important or rather the essential dimensions of their role. Responses to our survey reveal how some academics are already engaged in forms of labour-clearing, or more specifically, sub-contracting of work, which has been further enabled by GAI tools:

I sub-contract out to a man via fiver. He has access to my diary and does all of my basic admin. So I paid him £100 a week but frees up two days a week for me to do paid consultancy via my own company. Has transformed my earning potential … the problem previously was that their English was often hit or miss. So my sub-contractor uses the AI to generate responses that look perfect in English according to my instructions. Going to take on another sub-contractor but this one for research papers … There are some core things I cannot outsource but there is so much low-level rubbish in Universities I bet I can get to 75 or 80%. (Academic Management, Pre-1992 University)

In this incredibly revealing admission, we find academic labour not so much hollowed out by the weight of an administrative burden, as the vast part of it stripped out and delegated to machine operation. We are exposed to an assessment of academia—provided by no less than an academic—as a collection of tasks which as ‘low-level rubbish’ might be easily outsourced; a view which resonates with previous assessment of LLMs exposing academia in the grip of a crisis of intellectualism. Such an assessment is a damning indictment of the state of academia and the extent to which it has become dominated by administrative functions, yet also of the opportunities for the more entrepreneurially minded to exploit the degradation of their work—their very deprofessionalisation—for personal advantage.

Adding to or Easing Work Intensification?

While we find respondents who talk up the labour-saving benefits of GAI tools, which might ‘enable many members of the Academy to unlock their abilities’, there are others within our sample who question whether academics in delegating certain work burdens to LLMs will find themselves becoming even further overwhelmed with substitutive work tasks. As certain job tasks are passed on to LLMs, our respondents worry that other tasks will take their place; tasks that could be potentially even more burdensome. In such a context LLMs will not so much compensate for work intensification as manage and even further exacerbate it.

There is however a risk that they become essential to help us manage ridiculous workload pressures and we are simply expected to produce more outputs in ever shorter time. (Professor, Pre-1992 University)

In this given scenario GAI tools, far from facilitating a clearing space for academics, will culminate in their further marauding and closing in by alternative tasks. Thus, GAI tools may lead to new forms of work exploitation that further congest academic workloads and do so under the pretence of work abatement by means of technological mediation.

Moreover, where use of GAI tools as a solution for administrative burden end up obfuscating the importance of research safeguards and reflexive praxis, as may be understood by the delegation of an ethics of care to an automated function, academics’ productive output may further accelerate yet with deleterious consequences to the production of knowledge:

Where ethics in research is ‘overlooked’ at the expense of ‘chasing’ publications and with the integration of generative AI if used irresponsibly, could affect publications and journals. (Graduate Teaching Assistant, Pre-1992 University)

Relatedly, our survey respondents considered the deleterious effects of GAI where it is used to manage work intensification and equally accelerate academic productivity. Where some might consider the affordances of GAI in the shape of AI tutors that are able much more efficiently than human tutors to handle the significant workload involved in the triage and pastoral care of students, we found a concern that may be most efficient is not the most profitable for staff and students in terms for instance of welfare and relationship building). A concern of quality may be also observed within the braggadocio of statements that communicate the super-sizing effects of GAI to academics’ productivity which also underscore predilection for a culture of churn:

I think it’s a game-changer for productivity. Last year I submitted 4 grants in 12 months. This year I’ve submitted 12 grants in 6 months. (Professor, Pre-1992 University)

Discussion

One of the most striking aspects in the survey results is the paucity of any kind of critical reflection among respondents on the potential of GAI as a positive disruptor to the neoliberalisation of higher education and the impoverished work conditions it affects; the kind that are routinely attributed for academics leaving their university jobs. Far from being a weapon of insurgency as might be intended by cyber-libertarians, GAI is instead valorised as a technology which augments academia as a Taylorist production line and thereby reifies the neoliberal status quo. This is a disturbing finding for those who seek a reversal of higher education’s neoliberalisation and the colonisation of universities by corporate logic and those who attribute the erosion of intellectualism with the valorisation of accelerated working (cf. Berg and Seeber 2016).

The appropriation of GAI as an accelerator for research productivity may instead be seen to further exacerbate anxieties of publication volume faced by academics amidst a mandate to publish or perish and extend a trend of ‘hyperprolific authorship’ (Ioannidis et al. 2018). It may also further propel the inexorable turnout of research that lacks empirical or theoretical gain or for that matter attentive readership or readership that might benefit from so-called ‘new knowledge’. Far from ameliorating scholarly work through the emancipation of time for reading, discussion and reflection, the time efficiencies presented by GAI may no more provide intermission from the productive chase than redouble expectations made of academics in servicing a productive surplus. In their willingness, indeed eagerness to utilise GAI to grow their output rather than reclaim an intellectual space, our survey respondents reveal something of both the lack of their resistance to fast scholarship and their inability, if not reluctance, to reclaim the intellectual space through an investment in slow scholarship (Berg and Seeber 2016). Their co-option of GAI in such terms, must however be considered in light of the relative insecurity of their positions; our survey sample after all is dominated by early to mid-career academics.

Where GAI is being implemented for purposes of further speeding-up academia, its application is also instructive of how academic labour is increasingly moving from a collegial to individualistic and isolatory undertaking. Pressures of competitive accountability that stem from academia’s hardening as a prestige economy, pits academic researchers against each other in a joust for positional goods. It takes not only an act of will but a position of power (or, flying under the radar of prestige economies) to engage in collaborative scholarship against the pressures of rugged individualistic scholarship generated by the neoliberal academy. GAI tools can be used to circumvent the need for research collaborations, supporting academics as solo-researchers, and importantly from an outputs perspective, as single-authors. As such, claims to excellence may be exclusively owned as opposed to collectively shared giving momentum to academic celebrity and the cult of the individual.

Greater use of LLMs by academics risks the danger, however, not only of more fragmented and disconnected academic communities—the inverse of collective intelligence (cf. Lévy 1999)—but an over-reliance on tools that fail to substitute for the critical and creative work that human collaboration facilitates. Academics may become further estranged from each other where the social cohesiveness that is simultaneously nurtured by and sustains their collective action wanes and where time spent within proximity, either in-person or digitally, is adjudged to impinge on their productive capacity. Where time spent with and given to each other is perceived as wasteful (cf. Readings 1996) and disadvantageous to academic career-building, the wider application of GAI, may cause to exacerbate the toxic-masculinity that has become synonymous with hyper-productive working and may even come to further normalise a culture of overwork.

The use of GAI tools by academics may, therefore, be argued to exacerbate the kinds of frailty already endemic to academics’ sense of identity, purpose, and community and contribute to existing systemic inequalities and prejudices where its application across various academic functions differs according to the profile of its non/users. It may be the case, for instance, that those who assume the greater weight of pastoral care for students (cf. Docka-Filipek and Stone 2021)—typically women, and people of colour—are disinclined from using GAI and fail to benefit from any of its efficiency gains. At the same time, hyperprolific authors, who tend more commonly to be cisgender white men, may further their productive advantage through its use. GAI may accordingly intensify performance-based stratification and structurally racist and gendered hierarchies within universities, while also raising important ethical and even litigious questions of responsibility.

Where GAI tools are used indiscriminately and/or unreflexively as appears to be the case within our survey sample, the very nature of what counts or rather what could or should count as academic work is called into question. If, as one of our respondents claims, a vast percentage of academic work can be automated and outsourced, then what do academics understand as their substantive contribution and role? They are surely more than just bureaucrats. Or perhaps this is exactly the extent of their contribution the application of GAI reveals. Does GAI use at least partially, therefore, signal their ‘technological unemployment’ (cf. Susskind 2022).

Taken from another perspective, the use of GAI may be understood to reflect a trend of burnout and disengagement prevalent since the Covid-19 pandemic and a reassessment by many workers across all sectors of what they deem tolerable and/or valuable within their working lives (cf. Forrester 2023; Gewin 2021, 2022; Schmiedehaus et al. 2023). GAI in this case would seem to reveal that a considerable amount of what academics do holds little professional value or meaning. At a time where the academic community frequently encounters hostility from political elites and where the value proposition of higher education is especially contested, admissions from academics that much of their work may be easily automated sets a dangerous precedent. Their esteem as experts may be further threatened, and their public role diminished in a milieu where UK policymakers with revisionist agendas are already disinvesting themselves of the expertise they so heavily lent upon during the Covid-19 crisis and where anti-intellectualist and populist persuasions remain rife. The precarising effects of GAI to public estimations of academia may be considerable where further truths emerge of its (mis)use.

The techno-solutionism some associate with GAI may then far less solve than bury the problem that is central to a discourse of academics’ disconsolation. As Welsh argues in his discussion of critical anthropology of audit cultures and regimes, in this post-Covid emergency moment:

It will be difficult for authorities to pass up the opportunity presented by the increased digitization and virtualization of activities such as teaching or inter-departmental communication that has emerged during the epidemic. Added to the repertoire of control, the migration of more and more activities online can at once further the material exploitation of academic labour, while retarding the organized resistances that were possible in the more intimately collegial disciplinary institution. (Welsh 2021)

Our findings suggest that uncritical use of GAI tools, rather than helping academics resist neoliberal logics of production and scarcity, amplifies and centres such logics. Use of GAI tools may further commit academics to mass produced and high churn research, robbing them of an intellectual legacy while corroding the aesthetic value of their work. The diminution of the distinctiveness of academic labour comes for some, however, not with resistance but wilful compliance and ideas of affordances that are ultimately antagonistic to academia as an intellectual endeavour. The application of GAI as an easy fix to academics’ shortcomings also arguably removes the impetus necessary for their intellectual growth, personal discovery, and development. They may become less inquisitive, less reflexive, and more narrow and shallow scholars. Ranks of neoliberal acolytes may consequently swell as GAI seeds further apostasy of academia’s intellectual heritage.

Conclusion

Our survey findings reveal the ways that academics co-opt GAI as a labour accelerator intended to alleviate their precarity. Consequently, we learn far more about academia through the lens of GAI than we do about GAI itself. What our respondents perceive to be the possibilities for GAI is evidence of the help and support they need in their local contexts: relief from bureaucratic burdens, support in conceiving of and starting research and writing projects, time for planning and operationalising teaching plans, help in supporting students, time and energy to commit to continuing professional development, and intersecting all of these, help in surviving UK academia as a prestige economy.

Yet, what we also find in these accounts is that GAI tools no more offer a route out of academics’ precarity than consolidate their estrangement from each other, their intellectual gift, and their capacity to escape the cage of their neoliberal subjectivity. We also find respondents beguiled by the artifice of GAI as a friend-of-time and the dichotomy of its investment as an instrument of occupational resilience simultaneously affecting their onto-epistemic dissolution. The quickening of academia’s superhighway by GAI moreover reflects the further debasement of an ethics of care by academics and new depths of unconscious collusion through digital dependency that keep the prospect of ‘real’ change submerged and the political influence and economic power of digital capitalists sky-high.

While those selling GAI claim that it has the potential to solve the problem of academics having ‘no time to think’ (Menzies and Newson 2007), our survey reveals how it further contracts academics’ engagement with critical thinking and contemplative labour. While approached as a potential remedy to the precarity of academic life, it transpires that it may no more alleviate than make further conspicuous the lottery of academic success (as relates for instance to research funding) and the pressures of academic capitalism that work to unravel collegiality and scholarly kinship. In deluding academics of the merits of its productive efficiencies GAI may become a weapon of some considerable self-harm. The collateral damage to efforts intended to halt academia’s slouching towards Bethlehem may prove irreversible. The transformational potency of GAI may be thus far less than its power to buttress academia’s neoliberal settlement.

However, the potency of GAI may also be usefully channelled into a violent awakening of its use as the apotheosis of academia’s decline, from which the only chance of rescue is radical renewal. Where LLMs are made analogous with ‘stochastic parrots’ (Bender et al. 2021), the multi-applicability of their use and near future proliferation into academic work demands that academics urgently confront and honestly reappraise the purpose and value of what they do lest both are sunk (cf. Welsh 2021). This is a difficult and essential task that requires resisting the impulse of neoliberal performativity and the persuasiveness of instrumental/technical rationality, which have blunted academics’ critical reflexivity. In fact, we surmise that the uptake of GAI tools by academics—which may as our respondents suggest only escalate—provides a potentially defining moment of revelation; a summons to self-accountability and opportunity to reconstruct through ‘a celebration of awareness’ (Fromm 1971).

The extent to which academics can be accountable to themselves, or to an ethics of care that does not centre institutional priorities of productivity, depends on their access to (or distance from) power. This would not just be resistance, but refusal (Simpson 2007): a rejection of the framing of academic work as endless productivity-chasing, in favour of more meaningful and caring work. Refusal will not look the same across all of academia, for the same reasons that there is an uneven acquiescence to the current audit regime across universities: some have more means to weather the consequences of their praxis than others. But taking the decision that refusal is good, and necessary, even if not always possible, can provide the moments of pause and reflection and in corollary, opportunities for academics to exercise their agency. GAI, as a new and still relatively untested technology, is a good candidate for strategic refusal (Browne 2015; Lanclos 2019).

Instead of hiding from themselves in curations of excellence, hollow victories of prestige, or narcissist apology, GAI use could force academics to confront the extent of their drift from intellectual and pedagogical craft and their anchoring to mundane service functions. Amidst the headwinds of digital disruption GAI use could potentially energise the conversation of what academia’s greater contribution is and the nature of academic craft. We would again point to the conversation about slow scholarship as praxis, (Berg and Seeber 2016; Wahab et al. 2022) as just one opportunity for academics to assert their agency around their own academic output, and once again define value according to logics that are not based on industrial models.

The growing use of GAI tools by academics also reveals the urgency of institutional responsibility in establishing the conditions of possibility by which academics might recommit to craftmanship as ‘the desire to do a job well for its own-sake’ (Sennett 2008: 9). By extension, if the proliferation of GAI use is the acme of academics’ despecialisation and deskilling it is also an epiphanic moment that supercharges the ‘recovery of an idea’ (Graham 2006) and the potential of academics to become reengaged as authors of an intellectual life. Consequently, the automating of academia by GAI may paradoxically help to spark morale, critical and creative agency, collegiality and even ‘organizational humanism’ (cf. Otenyo 2016) by making incontrovertible and indefensible the scale of erosion of such in universities. The folly of technological solutionism (cf. Morozov 2014) that might be driving academics’ use of GAI persists. However, the provocation it stirs may yet serve to slow, and perhaps interrupt completely, the procession of a hollowed-out academic life.