LLMs as Techno-Semiotic Machines: A Philosophical Challenge to Epistemia
In a recent paper on arXiv, Quattrociocchi and his team caution that large language models (LLMs) might lead to a scenario where linguistic plausibility overshadows authentic epistemic assessment, a phenomenon they label 'Epistemia.' While the authors acknowledge this concern, they critique the explanatory model that juxtaposes an embodied human knower with a detached generative model, placing epistemic validity in independent agents. Utilizing Carlo Sini’s philosophy concerning practices, signs, and techniques, they suggest viewing an LLM as a 'techno-semiotic machine' that streamlines a segment of written semiosis by generating credible linguistic arrangements from the accumulated history of human writing. Thus, Epistemia emerges as a result of this automation.
Key facts
- Quattrociocchi and colleagues warn that LLMs may allow linguistic plausibility to substitute for epistemic evaluation.
- The condition is called Epistemia: the experience of possessing knowledge without undertaking judgment practices.
- The article challenges the explanatory framework that compares an embodied human knower with an isolated generative model.
- The framework locates epistemic legitimacy in capacities internal to autonomous agents.
- The authors draw on Carlo Sini's philosophy of practices, writing, signs, and technics.
- They propose understanding an LLM as a techno-semiotic machine.
- The machine automates a phase of written semiosis from the sedimented archive of human writing.
- Epistemia is seen as one consequence of this automation.
Entities
—