Recent decades have witnessed a huge explosion in the number of research papers and scientific journals, due to digital transformation and the diversity of disciplines, but this unprecedented growth has not always been matched by qualitative control, which has opened the door wide to the commercial exploitation of the academic field.
The frantic race for publication has led to the emergence of opportunistic entities known as Paper Mills, entities that sell academics looking to increase their research budget the opportunity to publish quickly in exchange for a superficial review, emptying the process of its content. At the same time, commercial publishing houses make huge profits from Article processing fees, which creates an economic incentive to accept more papers regardless of their quality.
Some companies have exploited these loopholes to fund low-quality research or write research papers on behalf of others (Ghostwriting) in order to distort scientific evidence in favor of their products, in which case some company employees may write research papers attributed to ostensibly independent academics, in order to influence scientific evidence, guide public policies, and shape public opinion to serve the interests of their products.
The danger of corporate influence is evident in the issue of investigating the safety of the herbicide (glyphosate), as legal documents revealed that the comprehensive scientific review on the safety of the product, which was cited globally, was in fact written by employees of Monsanto, the producer of the pesticide, and published in a magazine with well-known links to the tobacco industry, and even after exposing the truth, the effect of the misleading paper lasted for years.
In response to these challenges, several grassroots and institutional initiatives have emerged that serve as an additional line of Defense to strengthen the integrity of science, most notably: Retraction Watch, which tracks withdrawn papers and reveals cases of academic misconduct, the Data Collada initiative, which develops methods to identify data manipulation in research, as well as investigative journalism, which reveals the hidden influence of companies on scientific research. As effective as these efforts are, they remain limited in Impact and extremely expensive, highlighting the inadequacy of relying on peer review as the sole guardian of scientific reliability.
But at the same time, a new hope looms on the horizon, as it is expected that artificial intelligence will be able to strongly enhance these efforts and help identify falsified research efficiently and quickly, which may contribute in the future to purifying the scientific record and supporting confidence in it.
Second, will artificial intelligence be the solution
Until recently, the technical tools that help in academic auditing were focused on detecting traditional plagiarism, but the scene is changing rapidly thanks to advances in artificial intelligence, advanced tools have appeared such as: (ImageTwin), (Proofig), which use advanced machine learning algorithms to scan millions of shapes and graphs for signs of manipulation or duplication, or use artificial intelligence tools to write research.
At the same time, natural language processing tools can identify distorted phrases, which are illogical or artificial language constructs that often refer to research produced by what are known as (research paper factories), and in addition, bibliometric dashboards, such as those provided by the academic search engine Semantic Scholar, provide insights into how research papers are cited, whether to support their results or to refute them, providing indications of actual value for research in a scientific context.
But the future holds something greater, as artificial intelligence is expected to soon reveal more subtle flaws in research, especially with the development of artificial intelligence models characterized by advanced reasoning capabilities in mathematics and logic. For example, the Black Spatula project aims to test the ability of the latest artificial intelligence models to verify published mathematical proofs.
These models have already demonstrated the ability to validate complex mathematical proofs and reveal contradictions that human reviewers have been unable to see.
These capabilities show that such systems, if given full access to scientific research databases and sufficient computing power, can soon conduct a comprehensive audit of the entire world's academic record. Such an audit is likely to reveal some cases of outright fraud, but what is more important, it will reveal a much larger mass of routine, unimpressive research and common mistakes, which is completely at odds with the heroic image of scientific discoveries promoted by the media.
Third, the effects and outcomes stemming from the thorough review of the academic record.
While the exact extent of fraud in scientific research is still unknown, we know that a large percentage of published research is not influential, and scientists themselves are well aware of this fact, as many of these published research papers are never cited or rarely referred to.
The solution lies not in denying flaws, but in leading the process of their disclosure, so the scientific community should take the initiative to conduct this audit itself, and prepare for its results transparently, and this requires a radical reformulation of the overall picture of the role of the scientist, adopting a more realistic and truthful picture of the scientist as a contributor to a gradually developing collective understanding. The myth of the heroic parade of individual miraculous discoveries should also be abandoned.
Because most scientific research today is not an intellectual revolution, but incremental cumulative efforts aimed at enhancing knowledge, practiced in the context of education, academic guidance, and community participation, recognition of this fact will not diminish the value of science, but will make its image more resilient in the face of AI-driven scrutiny.
Science has never derived its strength from infallibility, but its credibility lies in the constant willingness to correct and reform, and now, publicly demonstrating this readiness is an urgent necessity, before the credibility of science completely collapses.
This discovery, that is, much scientific research has little impact-may be as shocking to the general public as the discovery of fraud, because it contradicts the heroic and dramatic image of scientific discoveries, promoted by academic institutions and specialized media.
But the biggest danger is that the results of this comprehensive audit will be subject to exploitation in disinformation campaigns, anti-science groups may use these results as conclusive evidence of the Narrative (Science is flawed), which seriously undermines public confidence, and what complicates the matter is that artificial intelligence may be perceived as neutral and competent, which gives its results additional credibility in the eyes of the public.
And here lies the paradox: artificial intelligence may reveal the flaws of the scientific system, but at the same time, it may be used to undermine confidence in it, as these discoveries, despite their accuracy, may be exploited in disinformation campaigns, especially with the use of artificial intelligence itself to produce false content or direct misleading narratives.
Fourth, transparency is the first line of Defense for scientific trust