This was my first at the Council of Science Editors (CSE) conference, listening, learning, and observing the realities editorial teams face as our industry navigates one of its most significant transitions. Conversations and insights from forums like these are invaluable in shaping how we continue to evolve our services and technology to better support publishers and editors.

One session in particular – “The Modern Editor’s Workbench: AI Integration, Evolving Workflows, and Preparing the Next Generation of Editors” – stood out. It surfaced not just trends, but the real tensions, trade-offs, and evolving expectations of the editor’s role today.

My thanks to the panellist for the candid and thought-provoking discussion. Here are some of my key takeaways.

The Peer Review Crunch Isn’t Coming. It’s Already Here.

Between 2018 and 2025, something quietly but profoundly destabilizing happened to scholarly publishing: the willingness of researchers to accept peer review invitations nearly halved-from ~43% to ~22%. That’s not a marginal fluctuation. That’s a structural shift, observed across tens of millions of invitations.

If you’re an editor, you don’t need the data to feel it-you’re living it. Slower turnaround times. Reviewer fatigue. Endless chasing. Increasing reliance on the same shrinking pool of dependable reviewers.

But here’s the uncomfortable truth: this isn’t just a reviewer problem. It’s a system problem.

The Myth of Availability-and the Reality of Relationships

When researchers were asked why they accept review invitations, the answers weren’t surprising-but they were revealing.

Yes, journal reputation matters.
Yes, topical fit matters.
Yes, time constraints matter.

But what stood out most? Relationships.

Researchers are far more likely to say yes when:

• They know the editor

• They trust the editorial board

• They feel part of a community

Peer review, at its core, is not a transactional system. It’s a relational one.

And yet, most workflows today are optimized for transactions-not relationships.

The Operational Reality: More Tools, More Signals, More Work

At the same time reviewer availability is declining, editorial offices are undergoing another transformation: the explosion of integrity and compliance tools.

From plagiarism detection to image forensics, fake reference identification, authorship verification, and paper mill detection-there’s no shortage of signals.

And that’s the problem.

We’ve moved from:

• Not seeing enough to

• Seeing too much

Editors are no longer just evaluating research in the manuscripts. They are:

• Interpreting flags

• Weighing weak signals

• Investigating patterns

• Making judgment calls without clear playbooks

A flag is not a decision. And multiple weak signals rarely add up neatly.

This creates a new burden:

The work hasn’t disappeared. It has shifted and increased.

Managing editors may save time through automation-but editors and editors-in-chief are absorbing the complexity.

The Hidden Cost of “Efficiency”

There’s a popular narrative that AI and automation will “save time” in publishing.

That’s only partially true.

What’s actually happening:

• Front-end efficiency increases (faster checks, automated screening)

• Back-end complexity explodes (interpretation, validation, escalation)

So while workflows appear faster, decision-making becomes heavier.

And here’s the kicker:
Even if only 3–6% of submissions raise serious integrity concerns, at scale, that’s hundreds or thousands of investigations.

Most journals are not staffed-or funded-for that reality.

The Reviewer Crisis Is Also a Filtering Crisis

One of the most overlooked insights is this:

The best way to reduce reviewer burden is not to find more reviewers.
It’s to send fewer bad papers to review.

Early-stage screening-done well-can:

• Eliminate fraudulent or low-quality submissions

• Reduce wasted reviewer effort

• Improve reviewer trust in the system

But again, this only works if:

• Signals are interpreted correctly

• Policies are standardized

• Escalation paths are clear

Otherwise, you’re just moving chaos earlier in the workflow.

The Next Generation of editors/reviewers won’t Be “Found.” They’ll Be Built.

If reviewer acceptance rates are falling, and editorial complexity is rising, the long-term solution isn’t just better tools.

It’s better pipelines.

Today, most editors emerge accidentally:

• A good reviewer gets noticed

• A colleague gets invited

• A former student steps in

It’s organic. Informal. Relationship-driven.

But that model doesn’t scale.

What’s emerging instead is a more intentional approach:

• Structured editorial training programs

• Early-career reviewer pipelines

• Student-led journals with real editorial responsibility

• Embedded learning within workflows

The key insight:

People don’t learn publishing by observing it. They learn by doing it.

When early-career researchers actively participate in:

• Reviewing

• Decision-making

• Editorial discussions

They don’t just understand the system-they become invested in it.

And investment drives participation.

Technology + Editorial Expertise: Not Optional Anymore

We are entering a phase where editorial success depends on the fusion of two capabilities:

Technological fluency

Editors must:

• Understand AI-assisted tools

• Interpret signals (not just outputs)

• Navigate integrated workflows

Editorial judgment

Still irreplaceable:

• Contextual decision-making

• Ethical reasoning

• Pattern recognition across weak signals

The future editor is not just a subject expert.
They are a signal interpreter, workflow navigator, and community builder.

So Where Do We Go From Here?

There is no magic tool. No single fix.

But there is a clear direction:

Shift left-carefully

Bring checks earlier in the workflow-but only with:

• Clear policies

• Defined escalation paths

• Training on interpretation

Standardize where possible

Across journals:

• Integrity policies

• Decision frameworks

• Response protocols

Consistency reduces cognitive load.

Invest in editorial support-not just tools

If you surface more issues, you must:

• Staff for investigation

• Provide expert support

• Enable faster, confident decisions

Build - not borrow - the next generation

Create:

• Structured training programs

• Editorial fellowships

• Student participation models

Don’t wait for editors to “appear.”

Rebuild the reviewer relationship layer

Because ultimately:

Reviewers don’t accept invitations from systems.
They accept invitations from people.

Final Thought

We often talk about scaling peer review.

But perhaps that’s the wrong goal.

Maybe the real challenge is this:

How do we build a system that deserves to be supported-by editors, by reviewers, and by the next generation coming into it?

Because until we solve that, no amount of automation will fix a participation problem rooted in trust, workload, and human connection.