The AI Knowledge Trap - Omitted Information May Be Lost Forever - CleanTechnica

The article "The AI Knowledge Trap - Omitted Information May Be Lost Forever" from CleanTechnica highlights a critical concern about the limitations and biases inherent in current large language models (LLMs) and generative AI systems. While prominent figures like Elon Musk and others promote AI as a transformative force capable of solving major global issues, the article presents a contrarian perspective emphasizing that these AI models largely exclude vast bodies of human knowledge, particularly oral histories, indigenous languages, and non-Western epistemologies. Deepak Varuvel Dennison, a PhD student at Cornell, argues that because AI is trained predominantly on digitized content dominated by English and Western sources, significant knowledge from less represented cultures and languages—such as Siddha medicine from Tamil Nadu or languages like Hindi and Swahili—is marginalized or omitted entirely. Dennison warns that this exclusion risks entrenching existing power imbalances in knowledge representation and could lead to the irreversible loss of diverse cultural wisdom and traditional practices that have not been digit