AI tools for research labs: one library, one style, shared rules

This guide is for principal investigators, lab managers and senior postdocs who want consistent, defensible AI use across a group. It is part of our guide to AI for qualitative research and research by field.

What AI tools does a research lab actually need?

A lab needs fewer tools than it thinks, used consistently. Choosing AI tools for research labs starts with the shared jobs. Standardize the jobs that touch everyone's work, and let individuals choose for the rest.

Everything else, such as which editor a student writes notes in, can stay personal.

Should a lab have an AI policy?

Yes, because your funders and journals already have one, and they apply to every member of your group. A lab policy turns their rules into daily practice. The main rules to build on:

Rules change. Assign one person to recheck these pages twice a year, and always follow the specific journal's or funder's current instructions.

What goes in a lab AI-use policy?

A lab AI-use policy is a short, written agreement on which AI tools members may use, for which tasks and data, and how that use is checked and disclosed. Keep it to one or two pages. These ten items cover what most labs need:

1. Scope. Who it covers (everyone, including visitors and undergraduates) and what counts as AI use.
2. Approved tools. A named list, with the account type (institutional or personal). Prefer tools whose terms rule out training on your inputs.
3. Data classes. Public, unpublished lab data, confidential (others' unpublished work), identifiable participant or patient data. Say which class may go into which tool.
4. Never upload. Peer review material, others' unpublished manuscripts, identifiable participant data outside approved systems, and invention details before a patent filing (see prior art search for researchers).
5. Verification. Every reference is checked to exist and to support its sentence; every number is checked against its source.
6. Disclosure. Record AI use as you work, and state it in methods sections, cover letters and grant documents as each venue requires.
7. Authorship. AI is never an author. Authorship follows the lab's usual criteria.
8. Images and data. No generative AI to create or alter figures, gels, micrographs or data, unless the AI is the subject of the research.
9. Theses. Students follow their program's rules on AI in theses and agree any use with their supervisor in writing.
10. Ownership and review. Who maintains the policy, and when it is reviewed.

A short template you can paste into your lab handbook:

[Lab name] AI-use policy (version 1, [date]; owner: [name]; review: every 6 months)
Approved tools: [tool, account type, approved uses]
Allowed: editing our own text; summarizing published papers; coding help; brainstorming.
Ask first: anything involving unpublished lab data.
Never: peer review material; identifiable participant data outside [approved system];
  invention details before filing; generating or altering figures or data.
Always: check every citation and number; keep a log of substantive AI use;
  disclose as the journal or funder requires. AI is never an author.
Questions: ask [PI or lab manager] before, not after.

How do you share a reference library in a lab?

Use one shared library that belongs to the lab, not to whoever set it up. A shared reference library is a single collection of references and PDFs that every lab member can read and add to, organized so anyone can find a paper later. Set it up in six steps:

1. Choose the home. A reference manager group library (Zotero groups are a common free option [7]) or a research workspace with shared projects.
2. Create a folder per project plus a "Core papers" folder every newcomer reads first.
3. Agree a tagging scheme. For example: method tags (method:qPCR), status tags (status:to-read, status:cited-in-paper-2), and initials for who added it.
4. Set a file-naming rule for PDFs: FirstAuthor_Year_ShortTitle.pdf.
5. Name a curator who merges duplicates and fixes broken entries once a month.
6. Respect licenses. Share PDFs only within the lab and as your institution's licenses allow.

Pick one citation style per output

Agree the style for each kind of output: the target journal's style for manuscripts, the funder's preferred format for grants, and one house style (for example APA or Vancouver) for internal reports and theses unless the university requires another. Store the style choice in the project folder so co-authors do not reformat each other's references.

How do you onboard a new student?

Make the first week about the lab's shared tools and rules, not about reading 50 papers alone. A checklist:

1. Add them to the shared library and the "Core papers" folder.
2. Walk through the AI-use policy together and have them sign or acknowledge it.
3. Share the project folder with the right access: "can edit" for their own work, "can view" for others'.
4. Show the search log format the lab uses for literature reviews.
5. Set up the citation style for their first output.
6. Agree how drafts reach the supervisor and how comments come back.
7. Book a check-in after two weeks to answer questions about AI use.

AI tools for research labs in AI for Research

AI for Research is a research workspace with shared projects, a Library and AI tools that cite real sources. The Team plan is built for labs, departments and writing groups.

• Team plan. Everything in Pro for each member, 2,000 credits a seat a month pooled across the team, shared projects and Library, one invoice from 3 seats, and onboarding support. Team is arranged on request; see pricing [8].
• Shared projects. A Project holds a thesis, review or paper and is shared by email as can edit, can comment (for a supervisor) or can view. Every task is saved to History and can be moved into a project.
• Library with page citations. Members save papers and PDFs to the Library and ask questions of a PDF; each answer cites the page. Zotero and Mendeley import brings existing collections in.
• One citation style. Citations are free, in 10,000+ styles, and the editor exports to Word, PDF and LaTeX.
• Checks that support your policy. Citation check confirms each reference exists and flags retractions, at 0 credits. The Reviewer scores a student's thesis or manuscript and quotes each passage with a suggested fix, 10 credits per 10,000 words.

A worked example (invented)

Example: a lab of five takes a 5-seat Team plan, so the pool is 10,000 credits a month. In a busy month they use: two Reviewer passes on 40,000-word thesis drafts (2 × 40 = 80 credits), 30 Writing Studio sections (30 × 3 = 90), 6 whole drafts (6 × 15 = 90), 400 Library questions (400 × 1 = 400) and 5 slide decks (5 × 10 = 50). That is 710 credits, well inside the pool. Searching, citations and the Library itself cost nothing.

Honest limits: AI for Research does not enforce your lab policy, so the rules above are yours to keep. Institution sign-on is not available yet.

FAQ

Should a research lab have an AI policy?

Yes. Funders and journals already have rules: NIH bars reviewers from using generative AI on applications, NSF treats uploading proposal material to unapproved AI tools as a confidentiality breach, and ICMJE asks authors to disclose AI use. A one-page lab policy turns those rules into daily practice: which tools are approved, what data may never be uploaded, how output is checked and how use is disclosed.

What should a lab AI policy include?

At minimum: scope, approved tools, data classes with what may be uploaded where, a list of things never to upload (peer review material, identifiable participant data, unfiled inventions), verification rules for citations and numbers, how to disclose AI use in papers and grants, authorship, rules for images, record keeping, and who owns and reviews the policy. Keep it to one or two pages so people read it.

How do you share a reference library in a lab?

Use one shared library with a folder for each project, a naming and tagging convention, and one person who keeps it tidy. Agree one citation style for each kind of output, and add new papers as you read them, not at submission time. Zotero group libraries are a common free option, and research workspaces with shared projects let the lab keep papers, PDFs and drafts together.

Can grad students use ChatGPT for research?

Usually yes, within rules. Most universities allow generative AI for tasks such as brainstorming, editing and summarizing published papers, but not for confidential or identifiable data in unapproved tools. Students must check every citation and claim, disclose substantive use, and follow their program's rules for theses. A lab policy and a supervisor conversation in the first week prevent most problems.

Does NIH allow AI in grant applications?

NIH does not ban AI tools in preparing applications, but its 2025 notice NOT-OD-25-132 says applications substantially developed by AI, or with sections substantially developed by AI, will not be considered the applicant's original ideas, and that NIH may use detection tools. The same notice limits each principal investigator to six applications a year. Use AI for editing, not for writing the science.

Next steps

• Handling interview or survey data in the lab? Read AI for qualitative research.
• Keeping individual reading in order? See how to organize research papers.
• Running a review as a team? Start with what a systematic review is.
• Lab results that might be inventions? Read prior art search for researchers.

Sources

1. National Institutes of Health. NOT-OD-23-149: The use of generative artificial intelligence technologies is prohibited for the NIH peer review process (23 June 2023). https://grants.nih.gov/grants/guide/notice-files/NOT-OD-23-149.html
2. National Institutes of Health. NOT-OD-25-132: Supporting fairness and originality in NIH research applications (17 July 2025). https://grants.nih.gov/grants/guide/notice-files/NOT-OD-25-132.html
3. National Science Foundation. Notice to research community: Use of generative artificial intelligence technology in the NSF merit review process (14 December 2023). https://www.nsf.gov/policies/ai/merit-review
4. Nature Portfolio. Editorial policies: Artificial intelligence (AI). https://www.nature.com/nature-portfolio/editorial-policies/ai
5. International Committee of Medical Journal Editors. Recommendations: AI use by authors. https://www.icmje.org/recommendations/browse/artificial-intelligence/ai-use-by-authors.html
6. Committee on Publication Ethics. Authorship and AI tools: COPE position statement (13 February 2023). https://publicationethics.org/cope-position-statements/ai-author
7. Zotero documentation. Groups. https://www.zotero.org/support/groups
8. AI for Research. Pricing. https://aiforresearch.app/pricing