Measure Once, Model Everywhere: Model-Based Per-Request Resource Consumption for HTTP
2026-07-03T07:23:51Z•234ef0f07ed3b615bc6973260dc8859c1038773453ca5d1187530e91acc10432
AI-alignmentAI-governanceCO2eHTTPLLM-gradingacademic-integritybenchmarkingcollective-empiricismdermatology-AIdisagreement-resolutioneducation-technologyenergy-disclosureevaluation-scarcitygenerative-AImodel-registrynginxpartial-credit-promptingper-request-metricspractice-auditingpseudo-rational-cognitionradiology-futuresscalable-oversightsociotechnical-alignmentsustainabilityuniversity-policy
What happened
This RSS bundle contains nine new arXiv papers (07/03/2026) spanning practical sustainability telemetry for web servers, empirical evaluations of LLMs in education, conceptual frameworks for AI governance and auditing, new oversight protocols for model disagreement, sectoral studies of AI adoption in clinical practice, workforce impacts in radiology, and meta-issues in AI evaluation and university policy. Key contributions: an nginx extension and offline/online model pipeline to emit per-request energy and CO2e estimates; a multi-model study showing liberal partial-credit prompting improves LL
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_cy
- Record identifier
- 234ef0f07ed3b615bc6973260dc8859c1038773453ca5d1187530e91acc10432
- Enrichment time
- 2026-07-03T07:23:51Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.