Title: Decoding Semiotics: PEEL as a Framework for Epistemic Accountability in AI-Assisted Research

Original: arXiv:2606.04152v1 Announcement Type: New Abstract: The integration of large language models into research workflows is subtly undermining the epistemic responsibility of scholars. This commentary presents PEEL, an acronym for Protocols for Epistemically Engaged Literacy in AI, which serves as a conceptual scaffold. It merges the deterministic distant reading capabilities of Voyant Tools with interpretive analysis from Claude, relying on Peircean semiotics and abductive logic. When tested against AI-generated summaries of three distinct source documents, PEEL exposes consistent distortions in quantitative metrics, term frequency, and epistemic tone—flaws that remain undetectable without non-AI analytical tools. The study offers three key design recommendations: AI systems must be paired with deterministic instruments; linguistic fluency should not be conflated with accuracy; and epistemic authority needs to be intentionally engineered rather than left to assumption.

Source: arXiv Generated at: 2026-06-04 00:00:00 UTC