Return to the Memorandum: Why Scientific Responsibility Must Be Revisited Again and Again
Some ideas must be returned to—not because they were forgotten, but because the world has caught up with them.
Our Memorandum on Personalized Responsibility, Unified Oversight, and Human–Artificial Participation in the Global Scientific Literature Space began with a simple observation: scientific knowledge no longer lives in separate worlds. It moves continuously across journals, books, preprint servers, repositories, platforms, databases, scholarly networks, the open web, and, increasingly, artificial-intelligence systems that search, summarize, rank, translate, recommend, generate, and evaluate information.
That transformation has not slowed. It has accelerated.
Science No Longer Has One Door
For much of modern scientific history, a journal served as the principal gate through which an idea entered the recognized literature. Publication was not the only form of scientific communication, but it was the primary signal that a contribution had been selected, assessed, and placed into a durable record.
That architecture has changed.
Today, an idea may first appear in a preprint, a book, a repository, a dataset, a code archive, a professional network, an online seminar, an institutional platform, or an open public discussion. A paper may be formally peer reviewed yet remain poorly supported, inaccessible, irreproducible, or later shown to be unreliable. Conversely, a useful contribution may emerge outside a prestigious journal and still deserve serious attention.
The old question—“Was this published in the right journal?”—is therefore no longer enough.
A finding may be amplified, distorted, misunderstood, or silently buried by algorithmic systems long before a human reader examines its methods. Search engines, recommendation systems, citation indexes, social platforms, and AI assistants increasingly influence what becomes visible, what is cited, what is trusted, and what is forgotten.
Science now lives in one connected, real–virtual literature space. Its formats are diverse, but its ethical problems are shared.
Trust Must Travel With the Work
If scientific knowledge travels across many venues, trust must be able to travel with it.
Scientific trust cannot rest mainly on the prestige of a container. It must rest on something more durable:
- Identifiable and accountable contributors
- Transparent methods, evidence, data, and conflicts of interest
- Clear distinction between evidence, interpretation, and hypothesis
- Visible correction, criticism, and post-publication discussion
- Fair institutions capable of responding to serious misconduct
- Protection for vulnerable researchers, reviewers, and whistleblowers
This does not mean that journals no longer matter. They do. Careful peer review, editorial judgment, and established publication standards remain valuable. But journals are now part of a wider literature space. They cannot, by themselves, carry the entire burden of scientific legitimacy.
The same standards of integrity should apply wherever a claim is made, circulated, evaluated, or used: in a journal article, a book chapter, a repository record, a platform post, a clinical guideline, a dataset, a public summary, or an AI-generated synthesis.
Artificial Intelligence Is Already Involved
The same is true of artificial intelligence.
AI is no longer only a technical instrument used at the edge of scientific work. It increasingly participates in the practical life of knowledge. It helps people find literature, draft text, translate ideas, analyze data, identify patterns, screen manuscripts, detect anomalies, recommend citations, summarize evidence, and shape what becomes visible.
Its influence may be indirect, but it is real.
The question, therefore, is not whether artificial intelligence will participate in science. It already does. The question is whether that participation will remain hidden and ungoverned, or become transparent, auditable, and accountable.
AI cannot be an author in the human sense. It cannot accept legal or moral responsibility, disclose conflicts of interest, answer allegations of error or misconduct, or give meaningful consent. Human beings and accountable institutions must remain responsible for scientific claims and for the systems they choose to deploy. Current publication-ethics guidance likewise treats human authors as fully responsible for the final work and rejects listing AI tools as authors. [web:33][web:41][web:46]
But refusing AI authorship does not mean ignoring AI participation. When AI has played a meaningful role in writing, analysis, peer-review support, literature synthesis, ranking, recommendation, image generation, or integrity screening, that role should be disclosed in proportion to its importance. In consequential settings, the system should be documented, auditable, monitored for error and bias, and open to correction and appeal.
The future of science cannot be built on invisible systems that shape knowledge without anyone being able to ask how they work, whose interests they serve, or who is responsible when they fail.
Responsibility Without Censorship
The need for responsibility does not justify a single authority that decides what may be thought, investigated, or published.
Science requires disagreement. It advances through criticism, unconventional hypotheses, failed ideas, replication, correction, and sometimes the patient defense of a view that initially appears unpopular. Disagreement must not be confused with misconduct.
At the same time, freedom of inquiry cannot become a shelter for fabricated data, false identities, manipulated peer review, hidden conflicts of interest, citation cartels, metric gaming, deceptive AI use, or dangerous misrepresentation of evidence.
A responsible literature space must hold both principles at once:
Openness without responsibility can spread confusion. Responsibility without pluralism can become control. Science needs both freedom and accountability.
This is why integrity should be organized through federated, transparent, and appealable structures rather than through a single centralized gatekeeper. Different disciplines, regions, journals, repositories, universities, professional associations, and public-interest organizations may develop their own mechanisms. But they should work from shared principles: evidence, proportionality, due process, correction, protection from retaliation, and meaningful opportunity for appeal.
The Questions Still Before Us
We return to the memorandum not to declare it complete, but to make it more open.
The unresolved questions are substantial:
- How should researchers disclose meaningful AI involvement without creating meaningless bureaucratic checklists?
- How can peer review become more transparent while still protecting reviewers from harassment, coercion, or professional retaliation?
- What should a scientific “passport” contain—and who should govern it so that it does not become a system of permanent stigma, automated exclusion, or centralized control?
- How can corrections, retractions, major critiques, and verified updates travel across journals, repositories, platforms, search systems, citation indexes, and AI-generated summaries?
- How can researchers from under-resourced institutions and countries participate fairly in global systems of verification, visibility, and scientific influence?
- What safeguards can ensure that integrity systems investigate genuine misconduct without suppressing dissent, creativity, intellectual risk-taking, or minority scientific views?
These are not merely technical details. They are questions about the future moral architecture of science.
The UNESCO Recommendation on Open Science has already established a broad international direction: scientific knowledge should be more open, equitable, inclusive, transparent, and grounded in quality and integrity. Adopted by 194 countries in 2021, it offers an important global reference point—but its principles now need practical forms of implementation in an AI-mediated literature environment. [web:35][web:37][web:38]
An Invitation to Build
The memorandum remains an invitation.
It invites scientists, reviewers, editors, publishers, librarians, research institutions, professional associations, public-interest organizations, platform designers, AI developers, and readers of science to help build a literature space worthy of the knowledge—and the power—it now carries.
The task is not to protect old hierarchies merely because they are familiar. Nor is it to celebrate technological openness while ignoring manipulation, inequality, and the disappearance of responsibility.
The task is to build shared standards, distributed safeguards, transparent procedures, and independent paths of appeal across the many places where science now lives.
Some issues must be picked up again and again.
Scientific responsibility is one of them.
Read the Memorandum
The full foundational document is available here:
Mykola Iabluchanskyi, Vladimir Shlyakhover, Alexander Martynenko, Yuriy Dimashko, Gianfranco Raimondi, and Andriy Yabluchanskiy
Mykola Iabluchanskyi together with Andriy Yabluchanskiy
Comments
Post a Comment