In the rapidly evolving realm of artificial intelligence and digital health, the scientific community faces a daunting paradox. Breakthrough AI-driven discoveries are advancing at an unprecedented pace, yet the traditional scientific publishing system—largely unchanged since the 17th century—remains an archaic bottleneck. This incongruity threatens not only the acceleration of medical innovation but the very foundation of trust and reproducibility in science. Dr. Boon-How Chew, a noted JMIR correspondent, articulates these challenges incisively in his latest editorial, highlighting the urgent need for a fundamental overhaul in how scientific knowledge is recorded and disseminated.
At the heart of this crisis lies a glaring mismatch: AI-enabled research methodologies generate massive, complex datasets and dynamic models that demand transparency and continuous validation. However, the predominant scholarly communication model still revolves around static, paper-based articles with opaque narrative summaries. These static outputs strip away the richness of the underlying data and analytical processes, rendering verification and reproducibility virtually impossible, especially for intricate AI systems. As Dr. Chew poignantly notes, “The black box of a clinical AI model cannot be built on the black box of a nonreproducible study.”
Economic factors compound these structural flaws. Prestigious academic institutions allocate tens of millions annually to maintain subscriptions for top-tier journals, while researchers face prohibitive article processing charges—sometimes exceeding $11,000 per paper—to share their work openly. This economic model not only restricts access but entrenches inequities in knowledge dissemination, disproportionately affecting researchers and institutions with limited resources. The financial barriers stifle collaborative progress and exacerbate disparities in global scientific participation.
Furthermore, the crisis of reproducibility threatens the reliability of AI-powered discoveries. Studies suggest that between 50% and 90% of published research findings across disciplines cannot be reliably reproduced—a catastrophe for any field that relies on scientific rigor. In AI-driven medicine, where clinical decisions hinge on validated evidence, the stakes are particularly high. The present publishing model fails to mitigate this challenge, as it focuses on final narrative claims detached from verifiable datasets or code repositories.
While a myriad of AI-based writing assistants and research tools—such as Paperpal, Elicit, and ResearchRabbit—have emerged to streamline manuscript preparation, these innovations merely accelerate the generation of traditional outputs without addressing core systemic issues. They improve efficiency but do not transform static articles into interactive, transparent research outputs. Consequently, the cycle of non-interactive publications perpetuates, with consequent limitations on peer validation and dynamic scientific discourse.
Dr. Chew advocates for a revolutionary reconceptualization of scientific publishing, describing it as the need for a “new operating system for science.” This envisioned digital system comprises enriched, dynamic research objects that embed data, methodologies, analytical logs, and peer review commentary within an interconnected ecosystem. Such a model would ensure that every claim made in a publication is firmly anchored in accessible and reusable evidence, fostering reproducibility and trustworthiness by design.
Technological advances today provide the tools necessary for this transformation. Cloud computing, blockchain for immutable data logging, advanced data visualization, and interoperable metadata standards pave the way for fully integrated digital publications. Researchers could share not just final conclusions, but live datasets, executable code, and versioned analysis histories. This transparently collaborative environment would enable other scientists to reproduce findings instantaneously and build upon them without redundancy or guesswork.
However, Dr. Chew emphasizes that technology alone is insufficient; collective will and structural incentives must align to catalyze this paradigm shift. Academia, publishers, funding bodies, and policymakers must collaboratively embrace open standards and incentivize the publication of comprehensive, dynamic research objects. Reforming peer review processes to accommodate iterative validation and continuous updates will be equally essential. This collective action is critical to avoid the ossification of knowledge in predigital formats that hinder scientific progress.
The consequences of inaction are dire. Persisting with the traditional publishing model in the age of AI risks relegating groundbreaking discoveries to languish in dusty archives, disconnected from the interactive, reproducible frameworks that modern science demands. This would stifle innovation, slow clinical translation, and ultimately impede improvements in patient outcomes. Digital health, reliant on rapid feedback loops and integrative analytics, cannot thrive under static publication regimes.
The vision, as laid out by Dr. Chew, is inspiring: a scientifically rigorous, transparent, and dynamic publishing ecosystem that matches the pace and complexity of AI-enabled discovery. It proposes a future where every scientific assertion can be interrogated at granular levels, reevaluated, and expanded upon—ushering in an era of accelerated innovation and democratized access to knowledge. Such a system would restore confidence in scientific outputs and catalyze a new golden age of digital medicine.
As the global scientific community stands at this crossroads, the challenge is clear: abandon the ossified infrastructures of predigital science and embrace a publishing revolution. This transformation is not merely a technological upgrade but a cultural shift towards openness, reproducibility, and equity. It promises to unlock the full potential of AI in revolutionizing healthcare and beyond.
Academics, publishers, and funders must recognize that the future of science hinges on this profound modernization. The tools are near readiness, the scientific need is acute, and the economic imperative is clear. What remains is collective courage and commitment to build and implement this visionary new operating system for science that matches the extraordinary capabilities of AI-driven discovery.
Subject of Research: People
Article Title: Our AI-Powered Discoveries Are Trapped in a Predigital System
News Publication Date: 31-Mar-2026
Web References:
https://www.jmir.org/2026/1/e96018
References:
Chew B. Our AI-Powered Discoveries Are Trapped in a Predigital System. J Med Internet Res 2026;28:e96018. DOI: 10.2196/96018
Image Credits: Boon-How Chew, MD, MMed, PhD.
Keywords
Scientific community, Academic publishing, Publishing industry, Digital publishing, Open access, Artificial intelligence
