When the Summary Replaces the Result: Machine-Generated Text, Provenance, and the Limits of Engagement
When the Summary Replaces the Result: Machine-Generated Text, Provenance, and the Limits of Engagement C. J. Tully Independent Researcher, Tamworth, NSW, Australia ORCID: 0009-0007-5661-7332 · Email: [email protected] 19 November 2026 Abstract This note examines the conditions under which AI-generated summaries cease to function as tools and become substitutes for mathematical or experimental claims. It draws on publicly posted examples: selective machine digests of research notes, AI-authored summaries of standard arguments posted as original papers, and exchanges in which caveats appear or disappear without new calculation. A single side-by-side contrast shows the precise step that is absent. A positive control illustrates a use that remains inside the boundary. The operational cost to a third-party reader is stated. The note is limited to publicly visible text; it does not infer private intent, platform logs, or clinical effects.
Authors
- Chloe Tully (ORCID: https://orcid.org/0009-0007-5661-7332)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24360370
- Primary Topic
- Academic Publishing and Open Access
- Type
- article
- Field-Weighted Citation Impact
- 0.00