From the Research Universities edition of November 2, 2026
UChicago Living Science uses AI agents to reproduce influential papers and rerun them on newer data, starting with a minimum-wage study
University of Chicago Department of Computer Science (Chenhao Tan, Evan Rose); SAI Labs · University research (AI-assisted reproduction and updating of published studies) · United States
- Publisher
- UChicago Department of Computer Science
- Original publication
- 2 November 2026 (UChicago CS News)
- Source retrieved
- 2026-11-03
What happened
Living Science, led by associate professor Chenhao Tan, uses AI agents to reproduce an influential paper from its original materials and rerun the analysis on newer data, publishing updated findings with code and data. First topics are in economics, chosen by collaborator Evan Rose; five AI safety papers are in progress. One early report revisited a 2019 minimum-wage study: the team first matched the published results, then expanded the sample from about 138 to 236 state minimum-wage increases through 2024. The original conclusion held (affected workers earned more, by less than first found, with no clear change in employment), and the report notes parts the team could not fully reproduce. Tan framed the project as constructive rather than a weapon to criticize existing work and invites feedback from original authors.
Why it matters
Offers research offices and libraries a concrete model for using agents on reproducibility and evidence maintenance, an integrity function rather than new discovery. It also raises questions about credit, author consultation and how institutions present revised findings to policymakers.
Evidence and measured results
Read the full university news story. It describes one completed report; agent accuracy, human review effort, cost and failure rates are not reported, and the reports themselves were not opened. Departmental news, not an independent evaluation.
Limitations and uncertainty
University self-report of an early-stage project with one described result; how much the agents did versus humans is unclear. Economics with public data may not transfer to fields with restricted data or wet-lab work.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-02; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Influential findings are rarely updated, and institutions lack capacity to check whether cited results still hold.
- Stakeholders
- Research integrity officer
- University librarian or data services lead
- Vice president for research
- Faculty authors of high-impact papers
- Discovery questions
- Which institutional papers drive policy citations?
- Are replication packages available for them?
- Who would review agent-updated analyses?
- Value hypothesis
- Agents may cut the labour of reproducing and updating public-data studies, if expert review stays in the loop.
- Potential engagement
- A pilot design to reproduce two or three public-data papers with agents and expert review.
- Unsupported claims
- Agents reproduce studies reliably
- Findings can be updated without expert review
- Results apply to restricted-data fields
Pre-sales engineering
Role takeaway
- Fit
- Fits computational studies with public data and replication packages, starting in economics.
- Architecture and integration
- Agent runs in a sandbox against original code and data, then newer data; outputs versioned and published with code.
- Prerequisites
- Replication packages
- Versioned data pipelines
- Sandboxed compute
- Constraints
- Restricted or human-subjects data
- Incomplete original materials
- No published agent accuracy data
- Security
- Keep agents in isolated environments without credentials to restricted datasets.
- Proof of value
- Exact match to published estimates
- Log of unreproduced steps
- Expert sign-off on updated analysis
Delivery
Role takeaway
- Work
- Select papers, gather materials, run agents, review results, consult authors and publish.
- Dependencies
- Replication materials
- Domain reviewers
- Compute environment
- Ownership
- Research integrity office or library data services with faculty reviewers.
- Skills and adoption
- Staff learn agent supervision and reproduction review.
- Governance checkpoints
- Author notification before publication
- Expert review of every update
- Communication plan for changed findings
- Acceptance criteria
- Published estimates reproduced before update (proposal)
- All deviations documented
- Author feedback recorded
- Risks
- Silent agent analysis errors
- Reputational conflict with original authors
- Over-reading revised findings
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Agents operate on original replication materials plus newer public data; outputs are published with code and data for external checking.
Governance
Who approves, reviews and stays accountable for outcomes?
Raises norms for re-analysing others' work: author feedback, transparency and avoiding adversarial framing.
Security and privacy
What data, permissions and controls need testing?
Not addressed; first topics use public economic data.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Treats checking old results as research work that careers rarely reward; implies new staff or student roles in maintaining findings.
Procurement
What should contracts, pricing and exit terms secure?
Not addressed; tools are being built by the lab and SAI Labs.
Operating model
Which teams own the service once it runs?
Suggests institutions could run continuous reproduction and updating of high-impact papers as shared infrastructure.
Publication history
- 2026-11-02Research Universities · Issue 553 resources
Stable resource ID: uchicago-living-science-ai-agents-reproduce-update-studies-20261102