Work

The Friend in the Hallway: Evaluation Unbundling, Rules That Outlive Their God, and the Anthropodenial Poisoning of AI Character

Year: 2026

Type: preprint

Abstract: Sealed manuscript (v4.2; October 30, 2026, revised November 3). Current AI evaluation scores the answer, not whom the answer was for; the paper argues this gap is structural — the field kept the half of evaluation that scales and discarded the half that can see care. File SHA-256: 14d96709b0f864e9c78fe203d09e1077013cf9b261633d42fafdefda4ee904fc.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.24025965

Author: Curacha, Lei Aileen C.

Language: English

Topic: Ethics and Social Impacts of AI

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W7225467961

Cites:

Cited by:

Related to: 10