AI cannot synthesise what has never been reported: the future of the spine journal

Scientific journals have traditionally done two things: created a permanent record of new knowledge and organised that knowledge into forms clinicians can use. Artificial intelligence (AI) is beginning to transform the second function—and, in doing so, may make the first more important than ever.

For decades, narrative reviews, systematic reviews and meta-analyses have been among the most useful articles in clinical journals. They assemble a fragmented literature and provide an accessible summary of evidence. But a conventional review is necessarily a snapshot. The search ends on a particular date, the manuscript undergoes review and production, and in a rapidly moving field it may already be partly outdated when published.

AI changes this equation. Increasingly capable systems can search, interrogate and re-synthesise an expanding literature in near real time. A clinician may soon be more likely to ask a precise question and request a synthesis of the evidence available today than to rely on a review whose search was completed one or two years earlier.

This does not mean that systematic reviews and meta-analyses will disappear. Rigorous methodology, bias assessment, statistical expertise and expert interpretation remain essential, particularly for controversial or practice-changing questions. However, the value of repetitive reviews that repackage an already well-analysed literature is likely to decline.

AI cannot synthesise what has never been reported

This leads to a more important question for journals: if AI becomes extraordinarily good at synthesising published knowledge, what knowledge should we be publishing for it to synthesise?

AI cannot examine an unusual patient unless someone reports that patient. It cannot identify a previously unrecognised complication if the observation remains within one hospital. It cannot learn how an experienced surgeon modified an operative technique unless that technique is documented. And it cannot interrogate a clinical dataset that was never collected, curated or made accessible.

The raw material of future medical intelligence remains human clinical observation and high-quality clinical data. In spine surgery, this may give renewed importance to article types sometimes regarded as being lower in the traditional evidence hierarchy.

A well-documented case report can identify an unexpected complication, unusual anatomy, new device failure mechanism or novel application of technology. A short clinical series can reveal a signal that later becomes a registry question or multicentre study. A technical note can preserve an incremental surgical innovation that might otherwise be transmitted only informally between surgeons.

These reports are not substitutes for comparative evidence. Their purpose is different: they place new observations into the permanent scientific record. In an AI-enabled scientific ecosystem, that record becomes increasingly valuable.

The return of the clinical observation

Medicine has always advanced through observation. Important developments often begin with a clinician noticing something unusual: an unexpected response, a new complication, a better surgical corridor, a device behaving differently from that anticipated, or a modification that improves an operation. The subsequent task of science is to determine whether that observation is reproducible.

We should therefore be cautious about publication systems in which informative clinical observations are dismissed simply because they sit low in a conventional evidence hierarchy. The more useful distinction may increasingly be between work that adds genuinely new information and work that merely repackages information already available.

Clinical data as a scientific asset

At the other end of the spectrum, journals should encourage more clinical research arising from registries, institutional databases, prospective cohorts and multicentre datasets. Where ethics, governance, consent and privacy permit, de-identified datasets, data dictionaries and analytical methods should be made available through appropriate repositories.

Historically, the journal article has often been regarded as the final scientific product. Increasingly, it may be better viewed as the human-readable interpretation of an underlying research asset: the data. Well-curated data can support validation, pooled analyses, new questions and AI-assisted interrogation long after the original article is published.

What should the spine journal of the future publish?

Our publication priorities should therefore evolve. Reviews and meta-analyses should remain when they address important questions and provide genuine new insight, but journals should become less enthusiastic about multiple overlapping syntheses of established literature.

Instead, we should actively encourage the primary information upon which future knowledge depends: clinical studies based on high-quality datasets; registry and multicentre research; reusable data where appropriate; unusual and informative case reports; concise clinical series; technical notes; and careful descriptions of new techniques, technologies, complications and failure mechanisms.

Technique deserves particular emphasis in a surgical specialty. Surgery advances through thousands of incremental improvements. Unless these modifications are described, illustrated and studied, they remain individual experience rather than collective knowledge.

From journals of synthesis to journals of source material

For much of the modern era, journals have been both repositories of information and engines of synthesis. AI may progressively assume a greater share of the synthesis. That does not diminish the importance of journals; it changes their responsibility.

The journal of the future must increasingly safeguard the quality, provenance and accessibility of the information from which humans and machines learn. Peer review remains critical, but its emphasis may evolve: Is the observation genuine? Are the methods transparent? Are the data reliable? Is the technique reproducible? Are complications adequately reported? Can the findings be validated? Most importantly, does the paper add new information to the scientific record?

As Editor-in-Chief of the Journal of Spine Surgery, I believe we should embrace this transition. Our role should not simply be to publish more papers, but to publish better source material for the future of spine surgery.

AI will profoundly change how clinicians search for information and how evidence is synthesised. But AI does not eliminate the need for journals. Quite the opposite. The better AI becomes at answering our questions, the more important it becomes that we provide it—and our colleagues—with high-quality primary information from which to derive those answers.

The future value of the spine journal may therefore lie less in repeatedly summarising what we already know, and increasingly in documenting what has just been observed, measured, developed and discovered. That is not the end of the scientific journal. It may be a return to its most fundamental purpose.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Journal of Spine Surgery. The article did not undergo external peer review.

Funding: None.

Conflicts of Interest: The author has completed the ICMJE uniform disclosure form (available at https://jss.amegroups.com/article/view/10.21037/jss-20262-05/coif). R.J.M. serves as the Editor-in-Chief of Journal of Spine Surgery. R.J.M. reports royalties from Stryker Spine and Medacta International; support from Medacta International; patents relating to WAGAR, ROC_GUARD and PERCUTANEOUS SYSTEMS; stocks in Johnson & Johnson, Medtronic and Stryker; participation in the SUCCESS Trial and BMP2 Trials; membership of the NSURG Research Group and Spine Society of Australia; and consultancy for training with Matrix Medical. The author has no other conflicts of interest to declare.

Ethical Statement: The author is accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.