Why the author matters

Increasingly, when we read student work, peer reviews, draft policies, or even colleagues’ emails, an unspoken question intrudes: was this written by a person, by AI, or by some combination of the two? This moment is rarely dramatic. It is quiet, often fleeting, and unevenly distributed across institutions and disciplines. Yet it subtly reshapes how we read, how we interact, and how we relate to one another at work (Davis, 2023). 

This is not simply a technical concern about authorship. It alters the orientation we bring to academic practice itself. Where interpretation once took precedence, verification now competes for attention and in an attention deficit economy, the risk of loss of cognitive engagement increases. Where judgement was exercised with a degree of assumed trust, it is increasingly accompanied by doubt and vigilance. These shifts are emerging gradually, but their cumulative effects are significant. 

Invisible labour as identity work in AGPT 

Generative AI is often discussed in terms of the AI dividend or the efficiency gains, but in what we now describe as the after‑GPT (AGPT) era, generative AI has produced a new form of academic labour that is largely invisible. This labour is not primarily about learning new tools or redesigning tasks. Rather, it consists of sustained cognitive and emotional work: monitoring uncertainty, exercising heightened vigilance, and repeatedly re‑asserting professional judgement in contexts where its legitimacy is no longer self‑evident (Bedard et al., 2026). 

We conceptualise this labour from observation. Academics must now decide beyond the how to read texts, to how much to trust them (Joubert & Strever, 2025), and how to justify decisions that once required little explanation. This adds to the mental load and is labour because it consumes attention, time and emotional energy (Irving-Walton, 2026). It is invisible because it is rarely named, measured, or formally recognised. 

Crucially, this labour is identity‑laden. Academic expertise has long been bound up with the capacity to exercise professional judgement — to evaluate quality, coherence and originality through disciplinary ways of seeing. In AGPT, that judgement is still required, but its credibility is less legible. When text can be generated with ease by non‑human systems, the link between written output and human expertise becomes opaque. As a result, academics find themselves repeatedly doing the work of making their judgement credible, both to others and to themselves. 

Why invisible labour matters 

This new invisible labour matters because it reshapes academic relationships, not just workloads. Persistent doubt and vigilance alter how trust is extended and received. Reading becomes more forensic; collegial interactions more tentative. Over time, this corrodes the relational fabric that underpins teaching, research and leadership. 

It also generates identity tensions. For many academics, authority has not rested on formal power but on recognised expert power (French & Raven, 1959) and sound judgement. When the signals that once supported that authority are disrupted, uncertainty is experienced as more than a technical problem but as a challenge to professional self‑understanding. This is further intensified by the lack of agreed and shared norms about what constitutes acceptable AI use, leaving individuals to navigate disagreement without shared reference points. 

At an organisational level, invisible labour creates friction. High‑level strategies may articulate systemic responses to AI, and specific practices may be redesigned locally, but the translation between the two now requires sustained interpretive work under conditions of reduced trust. This burden is felt acutely where judgement must be exercised and defended - particularly in leadership and mid‑management roles - but it is not confined to them. 

As we move from a before‑GPT (BGPT) to an AGPT university, the challenge is not only how to integrate new technologies, but how to recognise and address the unseen identity work they challenge. Without acknowledging this invisible labour, efforts to respond systemically risk misdiagnosing the problem they are trying to solve.  

The forms of invisible labour described above are not confined to any single domain of academic work. The following examples are illustrative rather than exhaustive, intended to show how identity tensions and relational uncertainty translate into everyday academic labour. 

Leadership and mid‑management 

Decisions about assessment design, acceptable AI use, or staff and student conduct must often be justified to multiple audiences who hold divergent beliefs about legitimacy and risk (Ransome, 2026). This places leaders in a position of sustained vigilance: interpreting ambiguous signals from policy, translating high‑level strategy into local practice, and absorbing uncertainty on behalf of others. The resulting invisible labour is not simply administrative, but emotional and cognitive, as credibility must be continually re‑established in conditions of reduced trust. In leadership and mid‑management roles, tensions on our identity emerge from the need to exercise and defend professional judgement in the absence of shared norms. 

Research 

In research contexts, trust has traditionally rested on assumptions about authorship, expertise and scholarly integrity. In AGPT, these assumptions are unsettled. Academics increasingly read texts — drafts, reviews, and even published work - with heightened suspicion, prompting questions not only about quality but about process. This places tensions on the scholarly identity by shifting attention from intellectual engagement to verification. The invisible labour here lies in the ongoing effort to reconcile disciplinary standards with uncertainty about how knowledge is produced, while maintaining confidence in one’s own judgement and in the scholarly community more broadly. 

Teaching and assessment 

In teaching, this identity tension is closely tied to assessment and feedback, where professional judgement has long functioned as a proxy for both expertise and care. When the relationship between student thinking and the artefact becomes opaque (Rizvi, 2025), academics appear to be working harder to decide how to interpret submissions and how to justify evaluative decisions. This produces sustained doubt and vigilance, as well as emotional labour in managing fairness, consistency and relational trust with students. Traditional routine judgement has become a state of continuous mental negotiation, much of which is unseen and unsupported. 

Thoughts to ponder 

The challenge posed by generative AI in higher education is often framed in terms of skills, assessment design, or technological capability. What this framing misses is the growing volume of invisible labour now required to sustain academic work in AGPT. This labour is cognitive and emotional, rooted in doubt, vigilance and the repeated justification of professional judgement. It is also deeply tied to academic identity (Watermeyer et al., 2024). 

As the legibility of expertise diminishes and norms remain unsettled, trust can no longer be assumed to travel easily through established processes. Yet trust has not become less important. On the contrary, it matters more than ever, precisely because the practices that once stabilised it have been disrupted. Invisible labour emerges where trust falters, filling the gap left by weakened signals of credibility and authority. 

If this labour remains unacknowledged, institutional responses to AI risk focusing on visible artefacts - policies, tools, redesigned assessments - while overlooking the relational and identity work required to make them function. Recognising invisible labour does not resolve uncertainty, but it does make it governable. It allows us to see that the task ahead is not only technological or procedural, but fundamentally relational: how are academic judgement, expertise and trust are sustained when their traditional foundations are no longer held? 

References 

Bedard, J., Kropp, M., Hsu, M., O.T. Karaman, O.T., Hawes, J. & Rosen Kellerman, G. (2026). When Using AI Leads to “Brain Fry”. Harvard Business Review March 6. https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry 

Davis, M. (2023). What do artificial intelligence systems mean for academic practice? SRHE News Blog https://srheblog.com/2023/07/28/what-do-artificial-intelligence-systems-mean-for-academic-practice/  

French, J. R. P., Raven, B.H.(1959). The Bases of Social Power. In Cartwright, D (ed.). Studies in Social Power. Ann Arbor, MI: Research Center for Group Dynamics, Institute for Social Research. pp. 150–167. 

Joubert, D., Strever, A. (2025). Towards pedagogies of distrust: Higher education learning in the age of generative Artificial Intelligence. Critical Studies in Teaching and Learning, 13(SI2), 59-77. https://doi.org/10.14426/cristal.v13iSI2.2506  

Irving-Walton, J. (2026). Judgement under pressure: generative AI and the emotional labour of learning SRHE News Blog https://srheblog.com/2026/03/24/judgement-under-pressure-generative-ai-and-the-emotional-labour-of-learning/ 

Ransome, E. (2026). What Generative AI reveals about staff capacity and institutional risk in higher education. HEPI Blog https://www.hepi.ac.uk/2026/04/01/what-generative-ai-reveals-about-staff-capability-and-institutional-risk-in-higher-education/  

Rizvi, L. (2025). The AI storm is here: Learning to swim in the AI flood, CABS Insights – Opinion https://charteredabs.org/insights/opinion/the-ai-storm-is-here-learning-to-swim-in-the-ai-flood  

Watermeyer, R., Phipps, L., Lanclos, D. et al. (2024). Generative AI and the Automating of Academia. Postdigital Science in Education 6, pp 446–466 https://doi.org/10.1007/s42438-023-00440-6