Comment Section for Opinion | A.I. Is Endangering Our History - The New York Times

The New York Times opinion piece discusses the dangers posed by AI and deepfakes to historical records. The authors, Jacob N. Shapiro and Chris Mattmann, highlight that while society has developed methods to identify and discredit fake current events, historical documents are vulnerable to manipulation. They point out that generative AI could be used to produce fake historical documents or alter pre-existing ones, leading to potential misinformation and distortion of facts. The authors cite various historical instances of manipulation of records for political or other gains. The authors suggest that a solution to these concerns could be watermarking digital files to trace their origins. However, challenges exist, including intellectual property limitations, as demonstrated by Google's unsuccessful venture to digitize world's library books. The authors propose that both government and industry have strong incentives to create immutable versions of historical data. They argue that preserving original training data, tools, and the environment is crucial, referring to this preservation method as "digital vellum." The authors also note that AI companies could benefit from verified historical records, as AI models trained on AI-generated data have shown rapid performance degradation. They suggest that distinguishing real historical records from newly created "facts" is critical. Finally, they urge immediate action to extend these efforts to historical records to prevent the distortion of political and historical narratives by generated history.