Can AI Rewrite History? The Curious Retraction of a Dead Physicist
AI systems and automated publishing workflows can influence how historical records are classified, interpreted, and ultimately understood. A recent case involving one of the greatest physicists of the twentieth century shows why context still matters.
The Curious Retraction of Max Planck
Scientific publishing depends on a stable historical record. Researchers expect journals to correct mistakes, retract fraudulent work, and preserve discoveries for future generations. Digital archives have made that record more accessible than ever before, allowing millions of papers to be searched, compared, and analyzed almost instantly.
Against that backdrop, readers of Science recently encountered an astonishing headline. Two papers by Max Planck, the founder of quantum theory and recipient of the 1918 Nobel Prize in Physics, had been listed as retracted decades after his death. Such news naturally raises questions. Had historians uncovered misconduct? Had one of the architects of modern physics committed scientific fraud?
Max Planck (1858–1947), photographed circa 1930. Planck, recipient of the 1918 Nobel Prize in Physics and founder of quantum theory, transformed modern physics with his discovery that energy is emitted in discrete quanta. Anonymous photograph credited to Transocean Berlin. Source: Smithsonian Institution Libraries, Scientific Identity collection. Public domain (anonymous work; copyright expired). Image via Wikimedia Commons.
Historians reached a very different conclusion. They found no evidence of fabrication, falsification, plagiarism, or flawed science. Instead, the papers had become casualties of changing publishing practices.
Planck wrote the philosophical essays during the 1940s, an era when scholars routinely republished lectures, essays, and previously printed material in multiple venues. Academic publishing operated under different expectations, and few readers regarded such republication as improper.
Publishing standards gradually changed. Journals introduced stricter copyright agreements, formal originality requirements, and explicit policies against duplicate publication. Digital archives accelerated those changes by making it possible to compare millions of documents automatically. Rules that once depended on editorial judgment increasingly became part of automated workflows.
According to the Science investigation, Springer digitized historical archives many years later. At some point, Planck's papers appear to have been flagged for duplicate publication or copyright concerns. Remarkably, even the journal's current editor reportedly expressed surprise when informed that the papers had been formally retracted. Historians suggested that an automated process may have produced a technically consistent result without considering the scholarly conventions that existed when the papers were originally published.
Planck's scientific legacy remains unchanged. His contributions to quantum theory did not change because two philosophical essays acquired a new label decades later. Yet the episode reveals something more significant than the status of two historical publications. Modern information systems had applied contemporary rules to documents created under a very different intellectual and publishing environment.
AI Cannot Change the Past but It Can Shape History
The Planck episode illustrates a challenge that extends well beyond scientific journals. Artificial intelligence, search engines, and digital archives excel at finding patterns and applying rules consistently across enormous collections of information. Those capabilities have transformed research by revealing relationships that previous generations might never have found.
Consistency, however, does not always produce historical accuracy.
Every historical record reflects the rules, customs, technologies, and assumptions of its own era. Remove that context, and even perfectly accurate data can produce misleading conclusions. A duplicate publication in 2026 may represent misconduct. A duplicate publication in 1945 may simply reflect normal scholarly practice.
Universities encounter similar challenges every time they digitize decades of student records and transcripts. Degree requirements change. Grading scales evolve. Academic policies are revised. An AI system comparing records across generations may identify apparent inconsistencies without recognizing that the underlying rules were different when those records were created. The same challenge appears in government archives, legal records, and museum collections, where historical context determines how information should be interpreted.
Artificial intelligence magnifies both the opportunities and the risks. Large language models can summarize archives, identify inconsistencies, and connect sources at extraordinary speed. Yet those models recognize statistical relationships rather than historical intent. An algorithm can determine that two documents appear similar. Determining whether they should be judged by the same standard often requires knowledge that exists outside the documents themselves.
Data governance therefore requires more than sophisticated algorithms. Provenance, institutional memory, and historical context remain essential components of trustworthy information. Organizations often treat context as metadata that accompanies the data. In practice, context is part of the data itself. Without it, information loses much of its meaning.
Many discussions about artificial intelligence focus on automation, efficiency, and scale. Equal attention should be given to restraint. Some decisions deserve human review because they involve more than patterns. They require understanding why records were created, what standards governed them, and how those standards have changed over time.
Max Planck's retracted papers did not expose hidden flaws in one of history's greatest physicists. Instead, they exposed a limitation in the way modern information systems sometimes interpret the past. Computers can preserve vast quantities of information with remarkable precision. Preserving history demands something more. It requires remembering that every record belongs to a particular moment in time, and that yesterday's actions cannot always be judged solely by today's rules.
Artificial intelligence will play an increasingly important role in organizing humanity's knowledge. Success will depend not only on building smarter algorithms, but also on preserving the context that gives information its meaning. AI cannot rewrite the past, but it can influence how future generations understand it. Protecting history therefore requires more than preserving records. It requires preserving the context that makes those records true.
Further Reading
- Why Have Papers by One of History's Most Famous Physicists Been Retracted?
- The Curious Retraction of Max Planck's Papers
- Retraction Guidelines, Committee on Publication Ethics
- The Structure of Scientific Revolutions, Thomas S. Kuhn