Copyright’s chilling effect on AI training data worsens algorithmic bias

Evidence

Levendowski’s pioneering article (and subsequent book-length scholarship) on the relationship between copyright law and the implicit-bias problem in AI training data is foundational for understanding the equity dimensions of contemporary AI-IP debate. Levendowski argued that the chilling effects of copyright on AI training datasets contributes to bias, since developers default to public-domain corpora skewed toward older, whiter, more male material. — , The Braided Reading EPUB, 2242

Categorization

• XREF: Directly connects to the Annotated Bibliography’s AI-and-IP theme and the book’s recurring equity-through-IP thread. — connects to: Machine-learning bias often stems from unrepresentative training data; AI training data is invisible yet permanently shapes deployed system behavior — Brian Christian, The Alignment Problem, 150; Even massive, comprehensive text datasets reflect biases present in the real world (+2 more)
• SEED: Article angle: copyright-induced bias — how fair-use rules are quietly selecting the demographics of AI’s memory.

Related

• Machine learning systems are not inherently fair or just by default (similarity 0.94)
• Even massive, comprehensive text datasets reflect biases present in the real world (similarity 0.90)
• Machine learning increasingly intertwines technical problems with human and social questions (similarity 0.79)
• Privacy rules can prevent engineers from even measuring model bias (similarity 0.79)
• AI training data is invisible yet permanently shapes deployed system behavior (similarity 0.79)

Broader Connections

• Machine-learning bias often stems from unrepresentative training data #corroborates (0.75)
• AI training data is invisible yet permanently shapes deployed system behavior #expands (0.55) — Brian Christian, The Alignment Problem, 150