Higher Education & ResearchIssue 482 records
Research Universities
Two newly inspected sources from 26 October 2026 on research-enterprise AI. A Columbia-led PNAS audit of 500,000 science, engineering and computing papers found suspected fabricated references in about 1 in 435 papers from early 2026, up from 1 in 1,600 in 2023, with experts confirming 85.8% of a 400-reference sample; the authors call for automated reference checks at submission. Cornell launched CoPAIR, a program of discipline pods and a core-facility pilot to test AI tools and set norms for use and non-use in research. CoPAIR reports no outcomes yet.
- Evidence records
- 2
- Cross-source patterns
- 1
- Evidence classes
- 1 academic research1 vendor claim
- Outcomes
- 1 cautionary1 emerging
- Source freshness
- 2 new this fortnight
- Research completed
- 2026-10-27
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Research AI adoption is being paired with discipline-level norms and automated verification, and the integrity evidence says those controls are needed
Interpretation across two 26 October sources. Cornell's CoPAIR program sets up discipline pods and a core-facility pilot to test AI tools and write guidance on responsible use and non-use of AI. The same day, a Columbia-led PNAS audit reported fabricated references in about 1 in 435 science and engineering papers from early 2026, up from 1 in 1,600 in 2023, and argued that automated reference checks at submission are a practical control. CoPAIR reports no outcomes yet, and the audit does not attribute causes to AI directly, so this is a direction of travel, not measured effectiveness.
Operating questionAs researchers adopt AI tools, which automated checks (such as reference verification) and which discipline-level norms are in place before use scales?
Supporting evidenceColumbia University School of Nursing and Data Science Institute (Topaz, Gupta, Zhang) with University of Eastern Finland and University of Turku (Peltonen)Cornell University (Cornell AI Initiative and Office of the Vice Provost for Research; Biotechnology Resource Center pilot)
Evidence ledger
Full record · every source keeps its link and limitations
Columbia-led PNAS audit: about 1 in 435 science and engineering papers from early 2026 cite references that cannot be found, up from 1 in 1,600 in 2023
The team drew 500,000 chemistry, physics, materials science, engineering and computer science papers published January 2023 to March 2026 from OpenAlex. An automated tool checked the 21.6 million references that carried a DOI and classified 1,135 references across 619 papers as fabricated when the cited title could not be found online. Two independent experts reviewing a random sample of 400 flagged references confirmed 85.8% as fabricated. Affected papers rose from 6.3 per 10,000 (about 1 in 1,600) in 2023 to 15.4 per 10,000 in 2025 and 23.0 per 10,000 (about 1 in 435) in Q1 2026. Field rates per 10,000 papers: physics 25.6, materials science 18.4, computer science and AI 11.1, engineering 10.3, chemistry 4.3. Most affected papers had one or two fabricated references, but 27 papers with five or more held over a third of all flagged references. The authors extrapolate about 8,500 affected papers among roughly 7 million in these fields; two of the 619 had been retracted at audit time. They recommend automated reference checks at submission, retroactive screening, and recording fabricated references as a distinct integrity category. A follow-on HHS OASH award (ORIIR260088, $249,999) funds screening of up to 12 million medical papers and development of screening protocols with editors and research integrity officers.
Why it matters, evidence and limitations
- Why it matters
- Directly relevant to vice presidents for research, research integrity officers, libraries and university presses. It extends the team's earlier biomedical audit to physical sciences, engineering and computing and gives a measurable, automatable control (reference verification) that institutions can run on theses, grant applications and faculty submissions.
- Evidence and measured results
- Read the full release text: corpus, DOI-only scope, automated classification rule, 400-reference expert validation (85.8% precision), yearly and per-field rates, distribution of fabricated references per paper, extrapolation, retraction count, recommendations and the follow-on HHS project. The PNAS article page returned 403 to the agent, so methods detail beyond the release was not inspected.
- Limitations and uncertainty
- The release attributes no cause to AI use directly; fabricated references are inferred from titles not found online, and 14.2% of the expert-reviewed sample were not confirmed. Only DOI-bearing references were checked, so non-DOI fabrication is unmeasured. Extrapolation to 8,500 papers is an estimate. The PNAS paper itself was not opened (403).
Cornell launches CoPAIR: research pods and a core-center pilot to give researchers AI tools, training and shared norms for AI use and non-use
Cornell announced Communities of Practice for AI-enabled Research (CoPAIR), co-sponsored by the Cornell AI Initiative and the Office of the Vice Provost for Research. It has two pillars. Research Pod CoPAIRs are small self-formed groups of domain researchers that test tools and methods under Cornell's "AI in the disciplines" framework. Core-Center CoPAIRs are anchored in existing centers and shared infrastructure; the first pilot, led by the Biotechnology Resource Center, runs about eight weeks in fall 2026 across genomics, proteomics and metabolomics, imaging and structural biology. Both pillars are meant to produce guidance, use cases, templates and best practices for responsible use and non-use of AI, curate shared repositories of AI agent skills and research applications, and help researchers integrate general-purpose and AI-for-science tools. Research Pod applications are due 4 December. The story reports no measured outcomes.
Why it matters, evidence and limitations
- Why it matters
- Offers a concrete operating model for research enterprises: discipline-level communities that evaluate tools and set norms, rather than central policy alone. Relevant to vice provosts for research, research computing and core facilities directors.
- Evidence and measured results
- Read the full Chronicle story: program purpose, sponsors, two-pillar structure, BRC pilot scope and length, outputs (guidance, templates, agent-skill repositories) and application deadline. Institutional announcement; no evaluation data.
- Limitations and uncertainty
- Institutional announcement of a new program; no participant numbers in this story, no outcomes, and pilot results are not yet available. Classified as vendor-claim because it is self-reported by the operating institution.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Academic research
- Research produced through an academic institution or peer-reviewed venue.
- Vendor claim
- A supplier-provided assertion that has not been upgraded to independent evidence.