A system of mutually reinforcing incentives that rewards publication volume over meaningful progress.
The research industrial complex is not the result of bad actors. Each participant — researchers, universities, funders, publishers — is responding rationally to the incentives they face. The collective result is a publishing culture that consistently prioritises metrics over knowledge, and quantity over quality.
Research industrial complex
One-sentence definition: The research industrial complex (RIC) is the interlocking system of incentives across academia — publication metrics, grant decisions, promotion criteria, and publisher revenues — that rewards output volume over meaningful contribution to knowledge or practice.
The term borrows from Eisenhower’s warning about the military-industrial complex: systems where interconnected institutional interests create self-perpetuating cycles that serve those institutions more than the public they claim to serve. In academic publishing, the dynamic manifests across every layer of the ecosystem:
- Universities evaluate faculty on publication counts and journal impact factors
- Funding bodies use those same metrics for grant decisions
- Researchers need publications for career advancement — particularly early-career academics
- Publishers benefit from a steady, growing stream of submissions
Each actor responds rationally to their incentives. The collective result is a system that drives researchers to slice work into minimum publishable units, prioritise novelty over replication, and chase citation counts among academics rather than evidence of real-world impact on practice.
How AI intensifies it
Generative AI is likely to accelerate rather than disrupt these dynamics unless deliberate countermeasures are taken. AI tools dramatically reduce the time needed to draft manuscripts, respond to reviewers, and repackage existing findings in new forms. This lowers the cost of publication without changing what is being rewarded. The risk is an AI-enabled paper mill: more papers, faster, with less connection to the kind of careful inquiry that actually advances knowledge or improves patient care.
There is also a structural concern: several major publishers have entered commercial partnerships with AI companies, allowing frontier models to train on their content. This introduces a new set of commercial incentives — generating training data — that are even further removed from the mission of advancing scientific knowledge.
The collective action problem
Reform is not simply a matter of individual will. No researcher can safely abandon traditional publication metrics while promotion committees continue using them. No journal can slow down to prioritise discourse while competing journals accelerate. Change requires coordinated action across researchers, journals, institutions, and funders — which is why previous reform efforts have struggled despite widespread recognition that the current system is dysfunctional.
The appropriate response for journals is to use AI to deepen rather than accelerate discourse: to surface connections, make peer review dialogic, and measure impact through changes in practice rather than citations among academics. See publishing-with-purpose for a developed argument along these lines.
Sources
- Rowe, M. (2025). Publishing with purpose: Using AI to enhance scientific discourse. Archives of Physiotherapy, 15(1), 90–96. https://doi.org/10.33393/aop.2025.3442
- Chu, J. S. G., & Evans, J. A. (2021). Slowed canonical progress in large fields of science. PNAS, 118(41). https://doi.org/10/gm2qmh
- Edwards, M. A., & Roy, S. (2017). Academic research in the 21st century: Maintaining scientific integrity in a climate of perverse incentives and hypercompetition. Environmental Engineering Science, 34(1), 51–61.
Notes
The analogy has limits — publishers are not intentionally sustaining health problems for profit, as defence contractors might sustain conflict. The value of the analogy is structural: it names the mechanism by which institutional self-interest can override mission without anyone intending that outcome.