In traditional Anglo-American copyright jurisprudence, the fair use doctrine has long evaluated the "transformativeness" of a secondary work by assessing whether it adds new expression, meaning, or message to the original creative artifact. For decades, this inquiry centered almost exclusively on semiotic and interpretive alterations recognizable to human audiences—such as parody, critical commentary, or artistic pastiche. However, the rapid emergence of automated large-scale text and data mining (TDM) has challenged this anthropocentric paradigm, compelling legal theorists to articulate a distinct doctrine of "computational transformativeness."
Proponents of this emergent paradigm contend that using copyrighted corpora to train machine learning models constitutes a transformative use not because the output necessarily comments upon or alters the underlying text, but because the ingested works serve a purely functional, non-expressive purpose. In this view, text is treated not as literary discourse to be read and aesthetically appreciated, but as raw statistical data representing linguistic patterns. By mining these statistical regularities to construct predictive semantic models, computational processes exploit the factual and structural dimensions of language—elements that copyright law has historically excluded from private monopoly.
Nonetheless, several legal scholars caution against uncritical judicial adoption of this non-expressive use rationale. They observe that unlike earlier, non-generative computational tools—such as search engine indexing algorithms or plagiarism detectors that utilize texts solely as search coordinates without generating competing market substitutes—generative artificial intelligence systems frequently produce synthetic outputs that directly supplant the market for the very expressive works on which they were trained. Consequently, these critics argue that decoupling transformativeness from its traditional communicative and human-interpretive mooring risks eviscerating the primary economic incentive structure foundational to statutory copyright protection.
Ultimately, the debate exposes a fundamental tension within modern intellectual property law: reconciling the public interest in technological innovation with the economic viability of creative labor. Rather than adopting an unyielding binary between expressive and non-expressive use, courts must develop a more granular doctrinal framework—one that scrutinizes not merely the mechanical function of data ingestion during model training, but also the degree of market displacement engendered by the resulting synthetic expressions.
Based on the passage, what is the author's primary purpose?
Answer: To evaluate competing perspectives on computational transformativeness in copyright law and propose a nuanced legal framework. / evaluating competing views on computational transformativeness and proposing a more granular framework that considers market displacement / to analyze the debate over computational fair use and recommend a balanced doctrinal framework / evaluating the debate surrounding non-expressive use in machine learning and advocating for a more granular judicial approach / to examine the conflict between computational transformativeness and copyright protection and propose a nuanced solution / analyze the debate on computational transformativeness and propose a balanced legal framework