Authors: Deidre Rose, Eric Henry, Mary-Lee Mullholland, Joseph Wilson

On May 22nd, as part of the Canadian Anthropology Society’s (CASCA) 2026 conference, the CASCA Labour Committee organized a roundtable entitled “AI, Academic Precarity, and Anthropological Teaching and Research in Canada.” Since 2017, the CASCA labour committee has been tracking the extent of precarious employment and working conditions of/for anthropologists in Canada. In this spirit, since artificial intelligence (AI) is reshaping scholarly labour in various ways, we were interested in how these new realities are affecting academic labour and, perhaps, making employment increasingly precarious for everyone. The goal of the roundtable was to open discussion on the impacts of AI on the teaching and research practices of anthropologists in Canada. We asked, “How is AI reshaping anthropological teaching and scholarly labour amid academic precarity in Canada? Do contract faculty, graduate students, and independent scholars turn to AI tools—such as large language models, transcription software, and analytics platforms?” After the speakers, we invited participants to engage in a collective discussion about AI. Rather than assessing AI as either a solution or a threat, we encouraged dialogic reflection on how anthropologists might collectively negotiate accountable, reflexive, and just uses of AI under conditions of precarity, or on whether this is impossible. 

Approximately 25 people attended the roundtable in addition to the organizers and scheduled speakers. The speakers included Deidre Rose, who introduced the session, Eric Henry, who hosted and gave a talk focused on language and academic writing; Joseph Wilson, the author of  Humans of AI: Understanding the People Behind the Machines, who spoke about his ethnographic research, and Mary-Lee Mulholland, whose talk “Falling Through the Cracks: The Intersection of Post-Secondary Institutional Policies and Generative AI” addressed policy challenges. More on these topics will follow. Attendees included doctoral candidates employed as graduate teaching assistants, sessional faculty, and full-time faculty, as well as a few Independent Researchers. During the discussion, concerns often reflected the speaker’s employment status and position within the university’s power structure, particularly regarding academic freedom, job security, and engagement with AI in academic labour. The roundtable thus reflected the mandate of the labour committee.

Deidre Rose spoke about the motivation for the roundtable, including her own experiences with the challenges of teaching in the absence of clear and consistent AI policies, as well as impressions gleaned from attending various AI training sessions and two AI- and Teaching-themed conferences. The impacts for all faculty included increased workload related to investigating suspicious papers, for example, tracking down hallucinated references and filling out the forms or meeting with students to address academic integrity concerns, and higher enrollment caps based on the idea that faculty could now use AI grading tools like Feedback Fruits and could therefore handle additional students without needing additional teaching assistant support. Other concerns stemmed from issues raised (often in the chat) at several of these AI-training sessions, where Rose got the (not surprising) impression that precariously employed academics had more concerns over job security and were perhaps less likely to embrace technologies than our tenured colleagues. Moreover, AI is scalable and promises measurable outputs, both favoured goals of neoliberal institutions (Rose, 2020). It is no surprise, then, that at some universities, administrations are encouraging people in academic and administrative positions to “leverage AI” for greater “efficiencies,” resulting in a corresponding offloading of tasks, and an overall increase in class size and workload, often in the absence of concrete policy regarding its use, a topic addressed by Mary-Lee Mullholland (see below).

The second speaker, Eric Henry, asked “What is writing?” in terms of both faculty, students and the general public. Eric started with the point that “slop” is not an exclusive feature of AI-writing, but has always been part of our engagements with texts: there is plenty of human-generated slop out there too (see Henry 2023). There was a collective fantasy, especially early after the introduction of AI tools, that we could detect, police, and manage AI usage in the classroom - we could use AI detectors that root out AI-generated text, craft clear institutional policies, and therefore treat AI like any other form of academic integrity violation. It has become clear that these are not realistic responses.

AI has infiltrated nearly every form of technological interaction and software platform we now use; we cannot put it back in the proverbial box. In the university classroom, this means that even when students do not “rely” on it to generate their work, AI shapes their reading, writing, search queries, and research practices. Faculty also use it for tasks such as syllabus design, identifying pedagogical resources and even - as Mary-Lee pointed out (see below) - student grading and feedback. 

Nevertheless, the intelligence manifested by AI tools is, fundamentally, “artificial” and is particularly bad at the high-level analytical thinking that we take as our aim in teaching. It cannot replicate the human skills that are key to a university Arts education: interpersonal communication, analysis, complex reasoning and so forth. One potential way forward is to lean into these skills and make their value explicit. Students who only learn to use AI to do their work for them will never be hired by anyone else to do those tasks; the employer will just farm them out to AI themselves. By shifting our attention away from grading things like grammar towards more complex analysis (applying concepts, comparing perspectives, synthesizing different approaches) we can re-focus our attention on the “human” side of intelligence. 

The question of Pandora’s Box, or “Black Box” (cf Gillani et al, 2023). was addressed by the next speaker, Joseph Wilson, who spoke about his ethnographic research on training AI language models and his work at a startup that manufactures computer chips designed for AI workloadsJoseph’s book, The Humans of AI, is a public anthropology project that explores the invisible, or “ghost,” workers involved in AI, as well as the engineers and developers behind today’s AI applications. Joseph mentioned the Luddite Lab as an example of organized resistance against AI being forced into all aspects of our lives.   He addressed how AI is reshaping teaching practices, which segues into the lack of consistent and coherent policy at the departmental, college, and institutional levels. He argued that one of the best ways of encouraging students to think critically about generative AI is to introduce them to the “human stack,” the network of people working behind the scenes alongside the “tech stack.” Mapping the human network responsible for developing AI not only allows students to question the Silicon Valley narrative that AI functions autonomously, but also introduces them to methodologies of the social sciences, including ethnography, participant-observation, discourse analysis, media studies, and archival research. Finally, Joseph noted that students are a vector of change; they are in a position to voice their concerns about AI and the erosion of their academic experience to administrators. As many schools, facing budget cuts and the rhetoric of the inevitability of AI, enter into partnerships with AI companies, students start to question the value of what they are receiving in exchange for their tuition. This was the focus of the next speaker’s talk, which was motivated in part by student complaints about faculty use of AI in fulfilling some of their teaching obligations.

Mary-Lee Mullholland presented a short talk,  “Falling Through the Cracks: The Intersection of Post-Secondary InstitutionaI Policies and Generative AI,” where she noted that currently, Mount Royal University, like other universities, is struggling to find its feet in the shifting landscape of AI in the context of teaching, learning, research, and administration. In particular, she focused on the context of teaching and learning. As Mary-Lee has done previously (Mullholland, 2020), she spoke about universities' policies as tools of governance and surveillance that, at best, create meaningless work (Graeber’s bullshit jobs), and, at worst, become tools of surveillance (in the Foucauldian sense). 

Speaking from the perspective of a Department Chair, Mary-Lee noted that a lack of university policy has created challenges when students report that they suspect an instructor may be using AI in course preparation or grading. Whereas student use of AI almost always falls under policies around academic integrity, AI use by instructors does not. The question is: what, if any, policies should or could be used to establish the limits and opportunities for instructors' use of AI. At issue is that AI is not a simple or discreet object that can be easily described or governed; it is fluid and pervasive. As a result, AI exists in and between policies, including those on academic integrity, copyright, intellectual property, privacy, code of conduct, and, more broadly, various articles within collective agreements.

Regarding the latter, Mary-Lee emphasized that the definition of teaching activities in Collective Agreements will be an important site of governance.  Specifically, how and when teaching activities, such as developing course content, designing assignments, and grading, can be subcontracted to third parties, and by whom, must be addressed with care.  If, for example, we permit subcontracting our teaching duties to AI, how might that impact intellectual property rights, the employment of precarious faculty, and the importance of academic rigour? These issues are not far off in the distance, but already at our collective doorsteps.  We have all heard stories from our students and in the media of students who feel violated and “ripped off” when they suspect (or know) that faculty are using AI to complete some or all of their teaching obligations. Where is the accountability and integrity if we subcontract academic labour out to AI? At stake are the fundamental practices of our daily work - often protected or policed under collective agreements and other university policies. Mary-Lee closed her presentation by asking, “Is it best to use existing policies to protect our labour or is a standalone AI policy necessary?” 

The themes that came up during the roundtable discussion, including those addressed by the organizers and scheduled speakers, included workload, academic integrity, workload, job security, writing and thinking, and ethics. There was agreement that there is a need for coherent policy guidelines and strong language in our Collective Agreements – for tenured, precarious, and non-academic workers whose jobs are, at least potentially, open to challenges related to AI technologies.

Another participant raised concerns about AI’s association with fascism and the concomitant polarization around how we see and engage with these new (and not-so-new) technologies. Several questions related to skills arose: How do we ensure students graduate with a good understanding of these technologies? Should we view AI as a tool or as a potential threat (a replacement for our thinking, writing, or employment)? One participant suggested that anthropologists tend to think of AI in terms of replacement and advised against thinking of ourselves in terms of skills that technology might automate. Environmental concerns, the neoliberal university and its overall economic structure, concerns for student and faculty integrity, and the current political climate were all raised. In a follow-up email, Robin Whitaker, president of the Canadian Association of University Teachers, sent a list of resources, including a link to a policy statement on Generative Artificial Intelligence.

Overall, there was a general sense that there is a need for clearer policy guidelines, stronger language in our Collective Agreements, and a more nuanced understanding of what the label “AI” encompasses, what it can and cannot do and how it operates, and what the role of the university is in the current political and economic climate as neoliberalism has perhaps given rise to technofacism (Gonzalez, 2026) what is the role of our discipline in response, particularly in a climate where many universities are being seduced by AI marketing promises?

Works Cited:

CAUT. Policy Statements. Generative Artificial Intelligence (AI). https://www.caut.ca/policy-statements/generative-artificial-intelligence-ai/. 

Gillani, N., Eynon, R., Chiabaut, C., & Finkel, K. 2023. Unpacking the “Black Box” of AI in Education. Educational Technology & Society 26(1): 99–111. 

González, R. J. (2026). American technofascism. Human Organization 85(1): 3–7. https://doi.org/10.1080/00187259.2025.2585410 

Henry, E. 2023. Hey ChatGPT! Write Me an Article about Your Effects on Academic Writing. Anthropology Now 15(1): 79-83.https://doi.org/10.1080/19428200.2023.2230097 

Luddite Lab  https://labor.dair-institute.org/. Accessed May 12, 2026.

Mulholland, M (2020). Honor and Shame: Plagiarism and Governing Student Morality. Journal of College and Character 21:2, 104–115.

Rose, D (2020). A Snapshot of Precarious Academic Work in Canada. New Proposals: Journal of Marxism and Interdisciplinary Inquiry 11(1): 7-17 https://ojs.library.ubc.ca/index.php/newproposals/article/view/192381

Smith, J. (2026). Humans of AI: Understanding the People Behind the Machines. Toronto: University of Toronto Press.