Academic research is very good at recording visible outputs. We count publications, citations, grants, authorship positions and formal leadership roles. These markers matter because they shape careers, funding and reputation. But they do not always tell us where the intellectual direction of a project came from. (Bourdieu, 1988; Merton, 1968).

Before a paper is written or a grant submitted, someone has to frame the problem, ask the question, propose the hypothesis or make the conceptual connection that gives the work its direction. That early intellectual work can be difficult to see and even harder to document. In our conceptual paper, we describe this as conceptual labour and argue that it deserves more explicit recognition, particularly as generative AI changes the relationship between having an idea and turning that idea into a polished output.

The work that comes before the visible output

Research is often described as though conception, execution and presentation are simply different parts of one activity. In practice, they are not identical. One person may formulate the central question, another may design the method, others may collect or analyse data, and others may draft and revise the eventual publication.

All of these contributions matter. The problem arises when the formal record captures the later, more visible work much more clearly than the early conceptual contribution. A distinctive framing might emerge in a supervision meeting, a project discussion, an internal review or an informal exchange. By the time it becomes a grant narrative or journal article, the route by which the idea developed may be difficult to reconstruct. (Allen et al., 2019; Hosseini et al., 2024).

We use the phrase currency of conception to describe the strategic value of this early intellectual work. A strong conceptual move can shape several studies, support a programme of research and influence what later becomes fundable or publishable. Yet the person who supplies that direction may not always be the person who receives the strongest formal recognition.

The Credit Value Gap

To make this mismatch easier to discuss, we introduce the term Credit Value Gap. It describes the recurring distance between those who originate conceptual work and those who later receive formal credit and reward for it.

This can happen in several settings. In grant development, a junior researcher or doctoral student may contribute the central framing but be ineligible to lead the bid. Once the proposal is formalised and funded, the visible record may associate the intellectual direction mainly with the principal investigator. In authorship, a contributor may be credited for coding, data collection or drafting even though they also generated the central conceptual move. In supervisory relationships, the boundary between guidance, co-conception and origination can be difficult to document.

None of this means that research has a single owner or that collaborative thinking should be reduced to a ledger of ideas. Genuine co-conception is common. Concepts evolve through discussion and revision. The point is different. When academic systems rely mainly on downstream indicators of contribution, early conceptual labour can disappear from the record more easily than later forms of work.

Why generative AI changes the problem

Generative AI did not create unfairness in academic credit. The underlying problem is much older. What AI changes is the speed and cost of execution.

Drafting, summarising, coding, structuring and preparing presentations can now be accelerated. That does not remove the need for expertise or human judgement, but it can shorten the route from an initial idea to a polished and visible output. When execution becomes easier to scale, the initial framing of the problem may become even more strategically valuable.

This creates a tension. Conceptual origination becomes more important at the same time as it remains difficult to prove. A concept heard in a meeting or encountered in collaborative work can be developed into a proposal, framework, report or article much more quickly than before. Those with greater institutional authority, stronger networks or more secure positions may also be better placed to convert ideas into recognised outputs.

Our argument is therefore not that AI causes idea appropriation. It is that AI can intensify an existing weakness by lowering the cost of turning ideas into outputs while leaving the origin of those ideas comparatively hard to trace.

Why this is also an equity issue

Ambiguity around conceptual credit does not affect everyone equally. Academic disputes take place within hierarchies. Early-career researchers, doctoral students, staff on insecure contracts and others with less institutional power may find it harder to challenge how a project’s intellectual history is narrated.

This matters because formal recognition is cumulative. Being seen as the person who originated a successful idea can lead to further invitations, leadership opportunities, funding and promotion. If conceptual contributions are repeatedly detached from the people who made them, the effect is not limited to one paper or one project. It can influence a longer academic trajectory.

Generative AI may also reduce barriers for some scholars by supporting drafting, language, organisation or workload. That benefit should be recognised. But improved access to execution does not solve the question of who receives credit for conceptual direction. Both issues can exist at the same time.

What universities could do differently

The answer is not to create a bureaucratic ownership system for every thought expressed in a meeting. Conceptual work is too iterative and collaborative for that. However, institutions can make the early stages of contribution less invisible.

Light-touch documentation can help. Brief concept notes after project-design meetings, supervision records that distinguish different intellectual contributions, and contribution discussions revisited during paper or grant development can create a shared record without turning collaboration into constant accounting.

Recognition systems also need attention. Promotion narratives, contributorship statements, grant-development processes and research leadership claims could become more explicit about conceptual origination rather than inferring originality mainly from visible leadership or final authorship position. Research integrity processes could also develop better language for disputes about the source of a concept instead of treating them automatically as ordinary authorship disagreements. (Allen et al., 2019; Moher et al., 2020).

There is a necessary caution here. Any new documentation process can itself reproduce power if the people with the most authority control the record. The aim should not be perfect ownership. It should be to reduce the ease with which early intellectual contributions disappear.

Credit should follow more than the final output

Generative AI is forcing universities to reconsider many assumptions about authorship, originality and academic work. One of the most important questions may be what happens before the output exists.

If writing, synthesis and other execution tasks can be accelerated, then the ability to frame a worthwhile problem, make a distinctive connection and set the intellectual direction of a project becomes even more important. Academic recognition systems need to become better at seeing that labour.

The question is not only who wrote the paper, led the grant or presented the final framework. It is also who made the work thinkable in the first place. If universities want fairer systems of research recognition in an AI-assisted environment, conceptual labour needs to become part of the conversation about credit, integrity and academic value.

Sources and further reading

The sources below are those used in, or directly relevant to, the original article and the claims retained in this blog adaptation.

Allen, L., O’Connell, A. and Kiermer, V. (2019). How can we ensure visibility and diversity in research contributions? How the Contributor Role Taxonomy (CRediT) is helping the shift from authorship to contributorship. Learned Publishing, 32(1), 71–74. https://doi.org/10.1002/leap.1210

Bourdieu, P. (1988). Homo Academicus. Stanford University Press.

Bozkurt, A. (2024). GenAI et al. Cocreation, authorship, ownership, academic ethics and integrity in a time of generative AI. Open Praxis, 16(1), 1–10. https://doi.org/10.55982/openpraxis.16.1.654

Cheng, A., Calhoun, A. and Reedy, G. (2025). Artificial intelligence-assisted academic writing. Recommendations for ethical use. Advances in Simulation, 10, 22. https://doi.org/10.1186/s41077-025-00350-6

Haven, T. L., Tijdink, J. K., Pasman, H. R., Widdershoven, G. A., ter Riet, G. and Bouter, L. M. (2019). Researchers’ perceptions of research misbehaviours. A mixed methods study among academic researchers in Amsterdam. Research Integrity and Peer Review, 4, 25. https://doi.org/10.1186/s41073-019-0081-7

Hosseini, M., Gordijn, B., Wafford, Q. E. and Holmes, K. L. (2024). A systematic scoping review of the ethics of contributor role ontologies and taxonomies. Accountability in Research, 31(6), 678–705. https://doi.org/10.1080/08989621.2022.2161049

Merton, R. K. (1968). The Matthew effect in science. Science, 159(3810), 56–63. https://doi.org/10.1126/science.159.3810.56

Moher, D., Bouter, L., Kleinert, S., Glasziou, P., Sham, M. H., Barbour, V. et al. (2020). The Hong Kong Principles for assessing researchers. Fostering research integrity. PLOS Biology, 18(7), e3000737. https://doi.org/10.1371/journal.pbio.3000737