The Devouring of All Knowledge (ryanfleck.ca)

The AI community is facing a growing controversy as companies, particularly Anthropic, are reportedly acquiring physical books—especially rare and out-of-print titles—to train their large language models (LLMs), only to subsequently destroy these texts. This practice, highlighted in recent news, raises ethical concerns about the commodification of knowledge and the potential loss of invaluable human-written literature. With the advent of LLM2 text generation in 2023, many see a cultural shift termed the “ensh**tification of all knowledge,” as any post-2023 writing could be AI-generated, diminishing the authenticity and reliability of future knowledge sources. As firms vie for scarce knowledge resources, a potential crisis looms where rare books may become vulnerable to wholesale removal from public circulation, resulting in an effective gatekeeping of information. Critics warn that if access to these digitized materials remains confined to proprietary databases, future generations will lack connection to fundamental human thought and history, echoing themes from dystopian literature like Ray Bradbury's "Fahrenheit 451". This situation exemplifies the tension between technological advancement and the preservation of cultural heritage, urging a collective effort to safeguard physical books against systematic destruction in the quest for powerful AI models.