Researchers using AI agents
Progress in using AI agents to empower researchers is rapid but still very unevenly distributed. Many researchers have been reporting great benefits while others are still complaining about ChatGPT hallucinating references. This is a collection of public accounts by researchers who shared their usage and lessons learned. It was compiled by AI agents at the direction of Dominik Lukeš who also provided some of his own use cases.
Find an account
Browse all 41By field
By kind of work
Recent examples
mathematics
Exploring a question in knot theory
Lipshitz chose an algebraic problem about Khovanov local equivalence groups that he thought Claude had a chance of solving and that he could check line by line. He reports that Claude did much of the exploratory research. He began with Fable 5, then used Opus 5 and later Opus 5.5. Some sessions assigned coding work to Sonnet. The project kept notebooks recording progress and failed approaches.
Robert Lipshitz. Lipshitz chose an algebraic problem about Khovanov local equivalence groups that he thought Claude had a chance of solving and that he could check line by line. He reports that Claude did much of the exploratory research. He began with Fable 5, then used Opus 5 and later Opus 5.5. Some sessions assigned coding work to Sonnet. The project kept notebooks recording progress and failed approaches.
- calculation and modelling
- writing and editing
- review and feedback
Lesson 1
digital archiving
Tracing summaries to a radio archive
I used Codex and local models to build a searchable catalogue of my radio recordings. The catalogue links summaries to passages in the transcripts so the wording and attribution can be checked.
Dominik Lukeš. I used Codex and local models to build a searchable catalogue of my radio recordings. The catalogue links summaries to passages in the transcripts so the wording and attribution can be checked.
- software building
- data analysis
- review and feedback
Lesson 5
Schwartz used Claude to assemble BootLoops, a set of numerical and symbolic tools that began with calculations in particle physics. As a first test, he asked Fable 5 to implement methods from his published work and related papers. He reports that Claude reproduced a published result before he asked it to look for unsolved calculations. The tools could check exact expressions against numerical evaluations at high precision.
Matthew Schwartz. Schwartz used Claude to assemble BootLoops, a set of numerical and symbolic tools that began with calculations in particle physics. As a first test, he asked Fable 5 to implement methods from his published work and related papers. He reports that Claude reproduced a published result before he asked it to look for unsolved calculations. The tools could check exact expressions against numerical evaluations at high precision.
- calculation and modelling
- software building
Lessons 1, 4
economics
Checking an R port against Stata
Francis DiTraglia reports using Claude Code and the GPL-licensed Stata source to port ivreg2 to R. The R package, ivreg2r, was published on CRAN on 21 July 2026. Package and development notes · CRAN record
Francis DiTraglia. Francis DiTraglia reports using Claude Code and the GPL-licensed Stata source to port ivreg2 to R. The R package, ivreg2r, was published on CRAN on 21 July 2026. Package and development notes · CRAN record
- software building
- data analysis
Lesson 4
Replication attempts
20 studies and benchmarks in which agents tried to reproduce published research.
- LLM workflow run over 4,452 economics replication packages1 Sep 2026, mixed
- Coding agents replicating four scientific machine learning papers with an evidence workflow2 Jul 2026, mixed
- LLM reanalysis of 180 social and behavioural science studies11 Jun 2026, partly succeeded
Key papers
Reading notes on 5 papers about agents doing research.
- Agentic AI systems for scientific discoveryFeng et al., 22 Oct 2026
- Reimagining research papers as interactive and reliable AI agentsMiao et al., 16 Sep 2026
- A multi-agent system for automating scientific discoveryGhareeb et al., 19 May 2026
- Accelerating scientific discovery with Co-ScientistGottweis et al., 19 May 2026
- Towards end-to-end automation of AI researchLu et al., 25 Mar 2026
Guides and tools
Things to open and use now.
- Council of modelsWorkflows for research
- Blattman’s starter kitWorkflows for research
- Sant’Anna’s workflowWorkflows for research
- ivreg2r validation guideTools and worked examples
- ivreg2r on CRANTools and worked examples
- AudioArchive case studyTools and worked examples
- ICMS workshop programmeWorkshop recordings and slides
- Science at the FrontierHow researchers used OpenAI’s tools
Every account by name
June 2026 onward18
- John H BoyerPhilosophy
- Thomas D. GrantLaw
- Robert LipshitzExploring a question in knot theory
- Dominik LukešTracing summaries to a radio archive
- Matthew SchwartzUsing physics tools to study a forest
- Chris BlattmanEconomics
- William BaudeLaw
- Lawrence EvalynLiterary studies
- Simon GoldsteinPhilosophy
- Terence TaoMathematics
- Gus SkorburgPhilosophy
- Francis DiTragliaChecking an R port against Stata
- Ke Xiao, Han Li, Teng Zhang, Yangchen Xu, Runze Mao, Zhi X. ChenEngineering
- Robert LipshitzMathematics
- Pedro H. C. Sant'AnnaEconomics
- Aaron TayLibrary science
- Scott CunninghamEconomics
- Adam JohnsonMedicine
Earlier accounts and uncertain dates23
- Joshua SternfeldHistory
- Jia Yu, Weiwei Yu, Pengfei Xiao, Fukun XingLinguistics
- Seyed Yahya ShiraziNeuroscience
- Paul Goldsmith-PinkhamEconomics
- Cameron BlevinsHistory
- Siddharth Mishra-SharmaPhysics
- Matthew SchwartzPhysics
- Max Zimmer, Nico Pelleriti, Christophe Roux, Sebastian PokuttaComputer science
- Seth LazarPhilosophy
- Zeke HausfatherClimate science
- Sam IllingworthHigher education
- Benjamin J. ShieldsChemistry
- Rob PatroGenomics
- Rayed AlGhamdiInformation systems
- Shawn GrahamHistory
- Tim DettmersComputer science
- Andy HallPolitical science
- Koen van GilstDigital humanities
- Kevin YangField not stated
- José R. PenadésMicrobiology
- Matthias MittnerPsychology
- Tiago VenturaPolitical science