Peer review capacity is often discussed as a numbers problem: too many manuscripts, not enough reviewers, and too much pressure on the people willing to review. But what if that framing is too narrow?
For Peer Review Week 2026, we asked our Chefs to look beyond the familiar call for more reviewers and consider two questions:
- What is one uncomfortable truth about peer review that the scholarly publishing community doesn’t talk about enough?
- Beyond recruiting more reviewers, what conversation about peer review capacity are we overlooking?
Rick Anderson
For the question of what uncomfortable truth we don’t talk about enough, I would say: The fundamental uncomfortable truth is that there’s no way for a traditional peer review model to keep up with the exploding volume of scholarly content hitting the system. What makes this a wicked problem is that a major factor in the current size of that explosion — AI-generated content — creates a new need for even greater pre-publication scrutiny of scholarly outputs. Post-publication review is, I’m sorry, a non-serious solution, but the uncomfortable truth is that serious scalable solutions are kind of hard to imagine. I can’t imagine such a solution that doesn’t involve AI, and I don’t know how reasonable it is to expect that AI will ever be able, reliably, to police itself. Hopefully someone out there with a more creative mind than mine will come up with something soon.
And beyond simply recruiting more reviewers, I’m not sure this question is being overlooked, exactly, but I think it’s the one that deserves much more of our serious attention at the moment, given the onslaught of nonsense, misinformation, and disinformation overwhelming the scholarly ecosystem: Can we harness AI to solve, effectively and reliably, the radical scale problem we now face in trying to validate scholarly outputs prior to publication? And if so, how?
Ashutosh Ghildiyal
When I think about the uncomfortable side of peer review today, AI and the question of reviewer judgement immediately come to mind: Peer Review Integrity challenges are arising due to AI. What journals ultimately need from reviewers is judgment and the willingness to take responsibility for it.
Today, reviewers have access to AI systems that can handle much of the validation work involved in reviewing and can even suggest critiques. For reviewers facing fatigue, who are time-strapped and unrecognized, the temptation to rely on these tools for judgment and cognitive work is high.
The problem is that fluency is not the same as judgment. AI can support the mechanical and cognitive aspects of a review, but it should not replace the human judgment that makes peer review truly valuable.
Reviewing is cognitive labor. If we continue expecting largely invisible cognitive labor from unpaid reviewers while AI makes it easy to generate polished review reports for free, we should not be surprised if reviewers increasingly delegate that cognitive labor to these systems. So, recognition is important. Supporting reviewers with tools that can perform mechanical checks is important. Making the peer review process cognitively fulfilling and interesting is important, so designing the process with technology that supports and enhances cognitive flow, while making the process both cognitively and materially rewarding, is important.
Beyond the question of finding more reviewers, I think we need to ask whether every kind of scholarship needs to be reviewed in the same way: Scholarly work serves different purposes. Some research aims to establish or test claims, while other work explores ideas, develops conceptual frameworks, generates hypotheses, or connects knowledge across disciplines. Applying validation-oriented evaluation criteria to exploratory research can direct scarce reviewer expertise toward questions that are not central to the work.
The longer-term opportunity is to design a peer review ecosystem that makes better use of reviewer expertise by aligning evaluation criteria with the purpose of the scholarship.
Hong Zhou
One uncomfortable truth is that peer review focuses heavily on what is being submitted today, while we rarely ask how reliable the existing scholarly record really is.
Publishers invest significant effort in screening new manuscripts, but much less attention is given to the millions of papers already published. How many contain serious errors, manipulated data or images, unsupported conclusions, inaccurate references, or findings that would not withstand closer examination? We do not yet have a clear or transparent picture of the scale of the problem.
In the past, systematically reassessing the legacy literature would have been almost impossible. AI and large-scale analytical tools are beginning to make this feasible by identifying suspicious patterns, anomalies and potential integrity or quality issues across very large collections.
The difficult question is whether publishers and the wider research community are willing to look. Doing so could lead to more corrections, expressions of concern and retractions, as well as reputational and financial consequences. But avoiding the issue does not protect trust. A transparent, proportionate and evidence-based review of the existing literature would give us a more honest understanding of the scholarly record and ultimately make it stronger.
For the second question, recruiting more qualified reviewers is important, but we should also ask why the system needs so many more reviewers in the first place. Submission volumes are rising, research is becoming more complex, and reviewers are often spending time on papers that are clearly unsuitable, technically weak, or potentially fraudulent.
Peer review capacity should therefore be addressed much earlier in the workflow. Checks should begin during authoring, preprint preparation, and submission, not only after a manuscript reaches reviewers. Identifying problems with methodology, reporting, references, statistics, data, images, or integrity at an earlier stage could reduce avoidable reviewer workload and improve manuscripts before formal review begins.
We also need to recognize that recruiting more reviewers does not automatically lead to better reviews. Human reviewers can miss errors, introduce bias, or apply standards inconsistently. The overlooked conversation is how human and AI review can complement each other.
AI could support technical, quality, integrity, and methodology checks, while human reviewers focus on interpretation, originality, and scholarly judgement. In time, AI agents may also support genuine reproducibility by running code, analyzing data and comparing results with the claims made in the paper.
Todd Carpenter
The uncomfortable truth of peer review is that it is fundamentally a hard job and something that is surprisingly undervalued by the community that relies so deeply upon it. The motivations for those with the expertise to undertake the work are shifting, as the number of faculty positions decreases and the requirements to maintain those jobs increase. All the while, the demands for the work are growing, the expectations about turnaround time are decreasing, and the depth of knowledge required is expanding.
The percentage of faculty with tenure or in tenure-track positions declined from 53.1% in 1987 to 31.8% in 2023, according to a 2024-2025 economic report by the American Association of University Professors. Just under half (48.6%) of those employed in a teaching position at a US institution worked in the role part-time, and of those, less than one-third are even eligible for tenure. Beyond this, of all of the things considered in a tenure application portfolio for that small percentage who are graced with the opportunity for tenure, work on peer review and other services to the field are the least important criteria in the P&T review process.
With the expectations of the rigor of the volunteer peer review work increasing, the pool of people willing to do the work for promotion-based recognition shrinking, and the volume of the work in terms of the amount of content needing review is growing rapidly, it is frankly surprising the volunteer system of peer review isn’t more troubled.
In terms of the other question, gift exchange economies, such as the current model for peer review, are often beset by a range of problems. These markets regularly face the free rider problem, in which a small group provides most of the labor or goods, while the majority consumes without contributing back. These markets rely heavily on trust and social accountability, making them difficult to scale beyond small, tightly knit communities. Gift economies also create an underlying presumption of social debts, in which participants can feel anxious, manipulated, or heavily pressured to reciprocate. Finally, gift networks can become insular, where outsiders who do not fit the community culture are excluded. Each of these describes in some way the challenges currently faced by our model for peer review.
This is not a challenge that we can simply “technology” our way out of. This is a social problem related to how the work is compensated. Reflecting on the fact that nearly a majority of teaching faculty are part-time, they could potentially become a pool of eager, professional peer reviewers if such work were to be compensated. To be clear, setting up this market should certainly not be driven by the courts or regulatory maneuvers, as is being attempted by some recent court cases. People should have every right to engage in a gift exchange, or to withhold participation if they so desire.
Given the paltry benefits, there is no reason to believe peer reviewers are compelled to do the work. Movement toward a peer review model based on monetary exchange should be based on the premise that it would simply work better. Using another form of free labor — in the form of technological capital — to avoid paying for important services will not address this core problem. If there is one thing we cannot and should not outsource to technology is human judgement. Nothing is more fundamentally human than using our knowledge and experience to reflect on new information and assess its quality. This is what peer review is at its core. The problem, in effect, is we are getting what we pay for.
“Let us remember that the automatic machine… is the precise economic equivalent of slave labor. Any labor which competes with slave labor must accept the economic consequences of slave labor.”
Haseeb Irfanullah
The one uncomfortable truth, whether we admit it or not, is that the current human-dependent peer-review system is failing us. The reasons I say this are straightforward: the prevailing system is overrated, full of flaws, exploitative, and maintains perpetual injustice; the human face in it is gone; it is emotionally draining for authors and reviewers, and no longer represents “good karma.” And the system is so unfair that 100% GenAI-reviewed preprints seem to be a better alternative for scholarly communication.
Looking beyond recruitment, you may ask: If the system is so broken, why do we celebrate Peer Review Week (PRW) every year? I also have similar questions: What is the value of discussing peer review every September? What do we gain from it? If the purpose is to explore new avenues to improve the peer-review system, do we really believe we can transform an exploitative system without addressing the deep-rooted injustice in it?
Over the last couple of years, the themes of PRW have been finalized through worldwide voting; indeed, a highly democratic process. It is an irony, though, since scholarly publishing’s directions of development, including the peer-review process, are dictated by large commercial publishers, whose profit margins match those of the tech giants. That’s why we see free-human-labor-dependent peer review increasingly being glorified, while a sector-wide shift to an efficient AI-dependent review system seems to be never reachable. This transformation will not be achieved, not because required innovations are absent, but rather because the publishing industry systematically won’t allow the ‘human element’ to be removed. Just imagine, the raw inputs (i.e., the manuscripts) to the publishing industry are written by humans (at least for now), and the final products (i.e., the journal articles) are consumed by humans (mostly, for now). If human reviewers are removed from this production line, what intellectual value addition would publishers claim? How will the industry declare itself to be any different from manufacturing, say, plastic bottles?
Any attempt now to improve peer-review capacity is nothing but a band-aid on the peer-review system. We need to strategically control the wild expansion of scholarly publishing, which is increasingly commodifying knowledge, affecting the mental health and career of researchers, and maintaining an unethical and unjust knowledge ecosystem. Unless we start talking about planned degrowth of academic publishing, discussing the future of the (peer-)review system will be meaningless.
Alice Meadows
My answer to both questions is essentially the same. I don’t think we talk enough about the ethos of peer review and why it’s a critical part of being a scholar/scientist — and I think our conversations about capacity likewise overlook the impact and importance of peer review as a form of community service. Over the years, survey after survey has found that a large majority of researchers at every stage of their career see peer review as a vital part of their contribution to their community and discipline. As a result, attempts to incentivize them financially haven’t (yet, anyway) been successful, but peer review training opportunities — such as HHMI’s Transparent and Accountable Peer Review program (TAP) for graduate students and postdocs, IOP’s comprehensive program(including their free Peer Review Excellence training and co-reviewing option), and many more — continue to be highly used and valued. With AI-generated content and reviews snapping at our heels, I’d like us to reiterate why communities — people! — are so vital to scholarship and science, and how peer review (by humans) is critical in order for those communities to thrive and grow.
Stephanie Lovegrove Hansen
The uncomfortable truth, to my mind, is that our current methods of evaluating research no longer align with the way it’s being consumed. Previous methods of evaluating quality have been overrun by the AI-empowered onslaught of submissions, such that it’s all editors can do is put things in piles of “slop” and “not slop.” As we attempt (and fail) to balance quantity and quality, the resulting corpus of content is likely diluted from the quality needed to hold up the bar in an age of AI-mediated consumption. Editorial judgment (not just workflows and throughput) needs to be supported and bolstered.
For the second question, I think conversations with current researchers reflect pre-AI methods and levels of rigor — checking versions of record, validating citations, and generally applying critical thinking to AI outputs. What happens when generations born into and trusting these tools become the baseline and enter the reviewer pools? Is that a slippery slope to just having AI reviewers? And whose job is it to hold and teach that standard moving forward?
Roohi Ghosh
The uncomfortable truth is that peer reviewers are doing a lot more today than they were expected to do before. From identifying AI hallucinations to detecting papermill activity or the inappropriate use of AI, the reviewer’s role has evolved. But their incentives have not. Also, AI is getting smarter, and sometimes being able to detect these AI-introduced errors is becoming difficult, and reviewers are increasingly becoming accountable for signing off without completely knowing how to spot these errors. The truth is that while the core nature or purpose of peer review remains unchanged, the role that the peer reviewer plays has changed, and so has the quantum of work. Too much research, and still the dependence on a handful of reviewers is putting a strain on the system. It is critical to understand what peer review quality means. What should peer reviewers be accountable for? And what is outside the scope of their responsibilities? Until this is clear, trying to solve for reviewer capacity will be difficult.
For me, the conversation beyond reviewer numbers comes down to protecting the part of peer review that requires human judgement: How do we protect the human part of peer review, i.e., the part that only humans are uniquely qualified to assess? This is the conversation that I hope we have this peer review week.
—–
Taken together, these perspectives suggest that peer review capacity is about much more than the number of reviewers available. The questions they raise, from how we value reviewer contributions to how we use technology and define the role of human judgement, may be just as important to the future of peer review as the question of capacity itself.
What do you think we are overlooking?
AI Use Disclosure: AI tools were permitted for drafting and editing contributions to this post. Where used, contributors remain responsible for the content, views, and final wording of their responses.
