New framework measures AI's lasting impact on user judgment

This paper introduces a new framework and evaluation protocol for understanding "epistemic transfer" in AI-assisted verification tools. Epistemic transfer measures how much a user can still perform a task independently after using an AI tool, as opposed to simply relying on the tool. The authors propose two metrics, the Epistemic Transfer Effect (ETE) and Tool-Removal Cost (TRC), to quantify this phenomenon. Their protocol aims to differentiate between genuine capability building, over-reliance, and other effects, emphasizing the importance of assessing what users retain after AI assistance. AI

IMPACT This research could lead to more robust evaluations of AI tools, ensuring they genuinely enhance user capabilities rather than just providing temporary assistance.

RANK_REASON The item is an academic paper published on arXiv detailing a new framework and evaluation protocol for AI-assisted verification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Epistemic Transfer
• Epistemic Transfer Effect
• Epistemic Transfer in AI-Assisted Verification: A Framework and Evaluation Protocol
• Gotit.pub
• Hugging Face
• Influence Flower
• ScienceCast
• Tool-Removal Cost

• paper
• other

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