Understanding the spectrum from fully appropriate to clearly inappropriate.

Clearly appropriate:

• Using AI to improve clarity of writing you created
• Using AI to generate code that you review, understand, and take responsibility for
• Using AI to help brainstorm ideas that you then critically evaluate
• Using AI to extract information from papers that you then verify
• Using AI to generate practice problems or study materials for learning
• Disclosing AI use appropriately

Probably appropriate with disclosure:

• Using AI to draft routine sections (methods, results) that you then verify for accuracy
• Using AI to generate initial analysis code that you then validate
• Using AI to organize and synthesize literature that you then verify and integrate
• Using AI for literature search assistance
• All of these require: disclosure, verification, and your continued responsibility

Clearly inappropriate:

• Submitting AI-generated work as your own without disclosure
• Using AI to understand content without learning it (using AI summaries instead of reading)
• Using AI to answer assessment questions in a course setting without permission
• Using AI to generate research findings without conducting the research
• Using AI to plagiarize others' work (asking AI to paraphrase another paper)
• Failing to disclose substantial AI use
• Presenting AI outputs as your own analysis without verification

Rules at your university or organization for AI use.

Key points:

• Most universities are developing AI policies (policies from 2023-2025)
• Student use policies often differ from researcher/faculty use policies
• Some policies prohibit AI for coursework without instructor permission
• Some policies require AI disclosure in assigned work
• Some policies allow AI only for specified tasks
• Policies evolve as institutions develop guidelines
• Your institution likely has a policy; find it and read it
• If unclear, ask your advisor or department chair
• If your institution doesn't have a policy, more conservative interpretation is safer

What to look for in policies:

• Is AI use allowed? Under what conditions?
• What kinds of AI use are prohibited?
• Do different rules apply to coursework vs. research?
• Is disclosure required? In what form?
• What oversight/review applies?
• Are there different rules for different types of AI tools?

How to find your institution's policy:

• Check university website (academic integrity, research office)
• Ask your department chair or advisor
• Contact research office or academic integrity office
• Search: "[Your University] AI policy"

What journals require regarding AI use in published papers.

Key points:

• Major journals have published AI policies (Nature, Science, JAMA, The Lancet, etc.)
• Most major journals now require disclosure of AI use in papers
• Common requirement: "Authors must disclose use of AI tools and describe their role"
• Some journals prohibit AI-generated images or data
• Some journals have specific restrictions on what AI can be used for
• Policies vary by field; check your target journal's policy
• If policy isn't published, contact editor with questions
• Failure to disclose when required can result in rejection or retraction

Journal policy locations:

• Author instructions or author guidelines
• "Responsible reporting" or "research integrity" section
• Specific policy documents on journal website
• Email to editor if unclear

What disclosure typically includes:

• Which AI tools were used
• For what specific tasks
• Whether outputs were reviewed and how
• Contact email for verification if needed

Example disclosure statements:

• "Claude was used to improve clarity of the methods section; all scientific content was verified by the authors."
• "GitHub Copilot was used to generate analysis code; the code was reviewed for correctness and statistical appropriateness by the lead statistician."
• "ChatGPT was used in the literature review to identify related papers and organize by theme; all cited papers were verified against original sources by the authors."

Rules from organizations that fund research.

Key points:

• NIH, NSF, DOE, and other major funding agencies are developing AI policies
• Most encourage appropriate AI use for research efficiency
• Most require disclosure in funded research proposals or reports
• Some prohibit AI use for specific tasks (e.g., ethical assessment)
• Policies are still evolving; check your agency's website for updates
• Failure to follow funder guidelines can result in funding clawback or investigation
• Different funders have different rules; check your specific funding source

Major funding agencies' current positions (as of 2024-2025; these evolve):

• NIH: Accepts AI use with disclosure; requires acknowledgment in publications
• NSF: Accepts appropriate AI use; requires disclosure in proposals
• DOD: Careful restrictions on some AI use in classified research
• Foundations: Varying policies; check specific foundation guidelines

Beyond written policies, professional norms in your field.

Key points:

• Your research community has developing norms about AI use
• Some fields further along (computer science, some areas of bioinformatics) have more established norms
• Other fields are still developing standards
• What's normal in one field may be viewed differently in another
• Understanding your field's norms prevents being out of step
• Your advisor/senior colleagues are important guides to field norms

How to learn field norms:

• Ask your advisor: "How does [field] typically approach AI use in research?"
• Read recent papers in your field: do they disclose AI use?
• Attend conferences: are there discussions about AI in research?
• Join professional organizations: do they have AI use guidelines?
• Read opinion pieces in field journals about AI and research integrity

A fundamental principle: appropriate credit for all contributions to your work.

Key points:

• AI systems are tools, not authors, but their contributions deserve acknowledgment
• If AI made meaningful contributions, it should be disclosed
• If AI made minor contributions (grammar check), lighter disclosure is okay
• The test: Would I hide this from my advisor, editor, or institution?
• If the answer is no, you should disclose it
• Transparency builds trust; hiding AI use creates suspicion
• As an author, you're responsible for everything in your work, including AI-assisted content

—

Practical Research Use Cases

Scenario: You're in a graduate seminar. The assignment is to write a literature review. The syllabus doesn't explicitly address AI.

What you should do:

1. Check syllabus and course policies again for any AI mention
2. If still unclear, email professor: "Can I use AI tools to help with the literature review? Specifically, I was thinking of using Semantic Scholar for literature discovery and Claude to help organize findings."
3. Follow whatever the professor says
4. If permission granted, disclose: "I used Semantic Scholar for literature discovery and Claude to help organize findings; all content was verified by me."
5. If professor says no, don't use AI
6. If professor doesn't respond, default to conservative approach: use AI only minimally and disclose

Why this matters: You're not just following rules; you're establishing your own integrity reputation at the start of your academic career.

Scenario: You used AI assistance in multiple ways while writing your paper: ChatGPT for brainstorming, Claude for clarity of writing, GitHub Copilot for analysis code.

What you should do:

1. Check the journal's AI policy (in author instructions)
2. Create a disclosure statement accounting for all AI use:

"ChatGPT was used to help brainstorm implications during manuscript development. Claude was used to improve clarity of the methods and results sections. All scientific content was verified by the authors. Code for analysis was generated using GitHub Copilot and reviewed for correctness by [statistician name]. See [acknowledgments] for details."

1. Include disclosure statement as required by journal
2. If journal doesn't specify where, include in acknowledgments or methods
3. Send with manuscript
4. If reviewer asks about AI, provide details

Why this matters: Transparency prevents suspicion and demonstrates that you have nothing to hide.

Scenario: You want to use AI tools throughout your dissertation research. You're unsure what's allowed.

What you should do:

1. Check university dissertation guidelines and policies on academic integrity
2. Meet with your advisor: "I want to use AI tools [specify which] for [specific tasks]. What's the university policy? What's appropriate for a dissertation?"
3. Follow advisor guidance; they know institutional norms
4. Create a written account of which tools you used for which tasks
5. In your dissertation, include a section documenting AI use:

• "This dissertation used the following AI tools: [list]
• AI was used for: [specific tasks, e.g., literature organization, clarity improvement, code generation]
• For each task, describe how you verified or reviewed AI outputs
• Maintain responsibility for all content"

1. During dissertation defense, be prepared to explain your AI use if asked

Why this matters: Dissertation committees are looking for your original work. Transparency about AI use shows that the work is still yours while using modern tools.

Scenario: You're writing an NIH grant proposal. You want to use AI to help write sections.

What you should do:

1. Check NIH guidance on AI in proposals (check NIH website for latest guidance)
2. Current NIH position: Appropriate AI use is acceptable with disclosure
3. Use AI to help draft some sections but verify all content
4. Include in proposal where AI was used (typically in footnote or appendix)
5. Focus proposal reviewers on your ideas and research design, not on writing quality
6. If in doubt, call the NIH program officer and ask: "Is it okay if I use AI to help write this proposal? Here's how I'd use it..."
7. Program officers are usually helpful and can clarify

Why this matters: Funding agencies care about research quality, not whether you used tools. Transparency about tools shows sophistication about current research practice.

—

Hands-On Exercise

Objective: Know what your specific institution, journal, and funder require.

Steps (Time: 60-90 minutes):

1. Find and read your institution's AI policy (20 minutes):

• Go to university website
• Search: "[University] AI policy"
• Look in: academic integrity office, research office, provost office
• Document: What does policy say about AI use in research?
• Note: If policy doesn't exist or is unclear, note that

1. Talk to your advisor (15-30 minutes):

• Schedule time with your advisor
• Ask: "What's our field's norm about AI use in research?"
• Ask: "What's your comfort level with me using AI tools for [your specific planned uses]?"
• Ask: "How should I disclose AI use?"
• Take notes on their guidance

1. Check journal policies for your field (20 minutes):

• Identify 2-3 journals where you might publish
• Find each journal's author guidelines
• Search for "AI" or "artificial intelligence" in guidelines
• Document: What does each journal require about AI disclosure?
• Note: Any restrictions on AI use?

1. Check funding agency policies (15 minutes):

• Identify your funding source (NIH, NSF, NSF, foundations, other)
• Go to funder website
• Search for AI policy or guidance
• Document: What does your funder say about AI use?

1. Create your personal policy (15 minutes):

• Summarize what you learned
• Create a simple written policy for yourself:
• What AI tools will I use?
• For which tasks?
• How will I disclose use?
• What will I always verify?
• Keep this as your guide

Time required: 60-90 minutes

—

Common Mistakes and Misconceptions

Absence of policy isn't permission. Broader academic integrity standards still apply. Conservative approach is safer when policy is unclear. Ask your advisor or institution for guidance.

Norms are moving toward disclosure being expected. Not disclosing when AI made meaningful contributions creates risk. If in doubt, disclose. Overdisclosure is better than underdisclosure.

It doesn't. Disclosure of appropriate tool use is normal. It shows you're sophisticated about research practices. Hiding AI use creates suspicion.

It is. You take responsibility for all content regardless of how it was created. Verification is not optional; it's how you maintain responsibility.

Others' ethics aren't your guide. What you need to do is right regardless of what others do. You want a reputation for integrity, not for cutting corners.

—

Key Takeaways

• Academic integrity requires disclosure and verification: Use AI appropriately and disclose where it helped; verify that AI-assisted content is accurate and represents your thinking
• Institutional policies, journal policies, and funder policies all apply and may differ; understanding your specific requirements prevents violations
• The test for appropriateness: Would you hide this from your advisor, editor, or institution? If the answer is no, you should disclose it
• Disclosure statements are becoming standard: Most journals now expect them; formatting follows journal guidelines; transparency builds credibility
• Field norms matter: Understanding what's normal in your research community prevents being out of step; your advisor is a key guide
• The attribution imperative: AI made contributions to your work deserve acknowledgment; this isn't diminishing your work but demonstrating integrity

—

Reflection Questions

1. Your institution: What does your institution's policy say about AI use? Is it clear? If unclear, who should you ask?

1. Your field: What's the norm in your research community about AI use? How much do colleagues disclose? What's expected?

1. Your publishing plans: Where do you plan to publish your research? What do those journals require about AI disclosure?

1. Your ethical standards: Beyond policies and norms, what's your personal ethical standard for AI use? How would you want others to know about your use?