Tau
AI for researchers & scientists · 11 lessons

AI for researchers who want to use it honestly

Use AI to map the literature, check citations, write analysis code and respond to reviewers, while meeting journal and funder rules. For researchers, PhD students and lab scientists.

Who it’s for: Researchers, PhD students, lab scientists and research managers.

AI can help researchers read more, code faster and write more clearly. It can also invent citations that look entirely real. This course shows you how to get the benefits while meeting the standards of journals, funders and ethics boards.

You'll map a field's literature in an afternoon with citation-mapping and search tools, then confirm each source exists. You'll catch fabricated and misattributed references by checking each DOI, the permanent identifier on a paper, and reading the full text. You'll use AI to pre-screen titles and abstracts for a systematic review, with two human reviewers deciding. You'll write analysis code that others can rerun, and draft a point-by-point response to reviewers while the scientific decisions stay yours.

Later lessons focus on integrity. You'll question the statistics AI suggests before trusting them. You'll disclose AI use correctly to your target journal, and follow funder rules on AI in grants and peer review. You'll decide which research data may go into which tools under your ethics approval. The course is for researchers, PhD students, lab scientists and research managers.

By the end you’ll be able to…

  • Map a literature field and verify every AI-suggested citation
  • Write and document reproducible analysis code with AI help
  • Disclose AI use correctly to journals and funders
  • Set a lab or group policy for responsible AI in research

Short on time? Start with the essentials

These 4 lessons cover the most useful ideas in this course. Each one takes about a minute.

Essential 1

Where AI helps across the research cycle

Map AI tools to each stage, from question and literature to analysis, writing and peer review, and spot the integrity risks.

Essential 2

Map a field's literature in an afternoon

Use citation-graph and semantic search tools to map key papers, clusters and gaps, then confirm each source exists.

Essential 3

Catch fabricated and misattributed citations

Verify that AI-suggested references exist and actually support the claim, using DOIs and full texts.

Essential 4

Write reproducible analysis code with AI

Use AI to write, explain and test analysis scripts, with version control, pinned environments and seeds so others can rerun them.

Every lesson in this course

This course has 11 video lessons of about a minute each, from the basics to expert habits. Each one ends with a question, three takeaways and its sources.

Foundations 2 lessons

  1. Where AI helps across the research cycleEssential

    Map AI tools to each stage, from question and literature to analysis, writing and peer review, and spot the integrity risks.

    “AI can help at every research stage, but major journals won't accept it as an author.”

  2. Map a field's literature in an afternoonEssential

    Use citation-graph and semantic search tools to map key papers, clusters and gaps, then confirm each source exists.

    “Start with one seed paper; finish the afternoon with a map of the whole field.”

Building skills 4 lessons

  1. Catch fabricated and misattributed citationsEssential

    Verify that AI-suggested references exist and actually support the claim, using DOIs and full texts.

    “Real journal, real authors, plausible title. The paper doesn't exist.”

  2. Screen and extract data for a systematic review

    Use AI to rank and pre-screen titles and abstracts for a systematic review, with two human reviewers deciding, a logged protocol and a stated stopping rule.

    “Thousands of abstracts to screen. AI ranks them; two humans still decide.”

  3. Write reproducible analysis code with AIEssential

    Use AI to write, explain and test analysis scripts, with version control, pinned environments and seeds so others can rerun them.

    “Your AI-written script works today. Will a reviewer get the same numbers?”

  4. Draft a point-by-point response to reviewers

    Use AI to split reviewer reports into numbered points, draft a courteous reply to each and track every manuscript change, keeping the scientific decisions yours.

    “Reviewer 2 wrote three pages. Turn them into numbered points and a change log first.”

Advanced 4 lessons

  1. Check the statistics AI suggests before trusting them

    Interrogate AI-recommended methods for assumptions, multiple comparisons and power, and get expert review.

    “Ask AI to find 'a significant result' and it may oblige. That's p-hacking.”

  2. Write with AI and disclose it properly

    Use AI for language editing within your target journal's rules (e.g. ICMJE recommendations, Nature or Elsevier policies) and write an accurate AI-use statement, checking the journal's current policy before you submit.

    “Nature won't accept an AI as an author. Here's what you must disclose instead.”

  3. Follow funder rules on AI in grants and peer review

    Apply funder policies such as NIH's (US) on AI-generated applications and reviewer confidentiality, or UKRI's in the UK, and check your own funder's current rules before you write or review.

    “NIH now caps each PI at six applications a year, partly because of AI.”

  4. Protect participant data and IP when using AI

    Decide which research data may go into which AI tools under your ethics approval, consent, data agreements and law (e.g. GDPR in the EU and UK, HIPAA in US health research), and check with your ethics board.

    “Your consent form never mentioned AI. Can participant data go into a chatbot?”

Expert 1 lessons

  1. Recheck a paper's reported statistics with AI

    Have AI recompute p-values from reported test statistics and degrees of freedom and run GRIM-style checks on means, in your own draft or a published paper, then judge which mismatches matter.

    “"t(28) = 2.1, p < .01"? That p is about .045. AI can recheck every one.”

Apps we recommend

These are the tools this course uses, and what each one is best for. We have no partnership with any of them.

Elicit

Best for: Systematic reviews and data extraction

Searches papers, screens studies and extracts data into tables with sentence-level links to the source text.

Freemium

Consensus

Best for: Evidence-based answers from papers

Answers research questions from peer-reviewed literature and shows how strongly the studies agree.

Freemium

ResearchRabbit

Best for: Visual literature and citation mapping

Builds citation maps from seed papers, recommends related work and alerts you to new studies.

Freemium
Hands-on project

Run one project through your group's AI rules

On a current or recent project, with no participant or confidential data in unapproved tools, map the literature with verified citations, make one analysis rerunnable and write an AI-use statement. Then draft a one-page AI policy for your group: approved tools, data classes, disclosure and supervision.

What you’ll make: A verified reference list, a rerunnable analysis repository, an AI-use statement and a one-page group AI policy.

Check your own work

  • Every reference has a DOI or stable link and was checked to support its claim
  • A colleague reran the analysis from the repository and matched the numbers
  • The AI-use statement meets your target journal's or funder's current rules
  • The policy says which data classes may go into which tools, matching your ethics approvals
  • The policy names who supervises students' and junior staff's AI use

Try a ready-made prompt

Copy a better prompt for a real task, and see why it works better than the version most people type.

Further reading and viewing

We’ve chosen these articles, research papers, videos and guides from other trusted sources.

Guide · European Commission · 30 min

Living guidelines on the responsible use of generative AI in research

EU guidance for researchers, institutions and funders on using generative AI responsibly.

Guide · US National Institutes of Health · 10 min

Supporting Fairness and Originality in NIH Research Applications (NOT-OD-25-132)

NIH's policy on AI-developed grant applications and the six-applications-per-PI annual cap.

Guide · Nature Portfolio · 5 min

Artificial Intelligence (AI) editorial policy

What a leading publisher allows: no AI authors, disclosure rules and limits on AI-generated images.

Research · Nature · 12 min

AI and science: what 1,600 researchers think

Survey of researchers' real uses of AI and their top worries, from misinformation to reproducibility.

Guide · ICMJE · 60 min

Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work

The authorship and disclosure standards most medical journals follow, including AI-assisted work.

Questions about this course

Can AI do a literature review?

It can speed one up a lot, but it can't do it for you. Tools such as Elicit, Consensus and ResearchRabbit help you find papers, map citations and pull out data. You still need to read the key papers, judge their quality and confirm every source exists. For a systematic review, use AI to pre-screen, with two human reviewers making the decisions.

How do I check if an AI-generated citation is real?

Look up each reference by its DOI or title in a database you trust, and open the full text. Then check two things. Does the paper exist, and does it actually support the claim it's cited for? AI can invent convincing references or attach real papers to the wrong claims. The course gives you a quick routine for this.

Do journals allow AI in research papers?

Many allow AI for language editing but ask you to say how you used it. Rules differ, so check your target journal's current policy, along with guidance such as the ICMJE recommendations. Funders have rules too, such as NIH's in the US and UKRI's in the UK. Keep participant data and confidential manuscripts out of unapproved tools, and check with your ethics board before using AI on research data.

Do I need coding skills for this course?

No. Tau is for people who use AI in their work, so you don't need to be able to code. Some lessons involve analysis code, but AI writes the first draft and you learn to check it. Each lesson is a video of about a minute, followed by one question, three takeaways and the sources. Your first 5 lessons are free, and Tau Pro gives you every course and lesson.

Get early access to this course

Tau opens soon. Join the waitlist and we’ll email you as soon as you can start this course.