Paper: LLMs as techno-semiotic machines create 'epistemic schizologia'

A new paper proposes understanding large language models (LLMs) as techno-semiotic machines rather than isolated agents. The authors argue that the fluency of LLM outputs can lead to a condition called 'Epistemia,' where users feel knowledgeable without proper verification. They suggest that this phenomenon, termed 'epistemic schizologia,' arises from a socio-technical split between linguistic expression and responsible interpretation, exacerbated by algorithmic authority and a lack of inspectable genealogy. AI

IMPACT Challenges current frameworks for understanding LLM knowledge, suggesting a need for designs prioritizing inspectable genealogy and distributed responsibility.

RANK_REASON The cluster contains a single academic paper discussing theoretical concepts related to LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Carlo Sini
• Epistemia
• epistemic schizologia
• Hugging Face
• Large Language Models
• Quattrociocchi
• techno-semiotic machine

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →