Lab AI policy

AI Policy of the Restoration Ecology Group

Introduction

Artificial intelligence (AI) tools may be used to support research activities in the Restoration Ecology Group; however, they do not replace scientific reasoning, professional judgment, or accountability. AI should be treated as an assistant or analytical tool, similar to advanced search engines or specialized software, rather than as an independent agent.

This document defines the permissible and prohibited uses of AI within the group. It aims to ensure that all researchers maintain scientific integrity, rigor, and transparency while leveraging AI to enhance efficiency, clarity, and productivity. Compliance with this policy is essential for the development of professional expertise and responsible conduct in ecological research.

1. General Principles

AI is a tool to support your work. Responsibility, critical thinking, and scientific reasoning remain with the researcher. Researchers are fully responsible for:

  • Understanding the science and logic behind all analyses and text. AI output depends entirely on the quality of your prompts and your understanding of the field. All reasoning, hypotheses, and interpretations must be grounded in your knowledge. AI cannot justify your scientific decisions or defend your thesis.
  • Verifying methods and results. Suggested transformations, tests, or analyses must be independently validated before application.
  • Ensuring accuracy, reproducibility, and traceability. AI-generated references, code, or summaries may contain errors or unverifiable statements. You must confirm all sources and results independently.
  • Ethical and transparent use of information. Declare all AI use, specifying the type and extent (e.g., code assistance, figure layout, text polishing).

2. Allowed (DOs)

AI may be used as a supporting tool under the following conditions:

Scientific Support

  • Translate text from foreign languages.
  • Summarize scientific papers after reading and drafting your own notes.
  • Explain statistical methods, ecological concepts, or laboratory techniques.
  • Provide guidance on coding (e.g., R or Excel) that you independently implement, verify, and adapt.

Improving Clarity and Structure

  • Refine your own text for grammar, readability, and logical flow.
  • Edit emails or documents for professional tone.
  • Check for internal consistency in your text.

Routine Tasks

  • Generate templates for data sheets, lab notebooks, or metadata tables.
  • Create preliminary diagrams or figures based on specifications you provide.
  • Produce example code that you verify line by line.

Brainstorming

  • List potential hypotheses or research questions based on your data and drafts.
  • Identify potential logical weaknesses (which you must independently evaluate).
  • Suggest sampling strategies or statistical approaches for your consideration.
  • Guide further literature exploration (verify all suggestions).

Important: All AI-assisted outputs must be understood, verified, and fully accountable by the researcher.

3. Not Allowed (DON'Ts)

AI must not be used to replace scientific responsibility.

Scientific Content Generation

  • Do not use AI to generate substantial scientific text that compromises authorship or academic integrity.
  • Avoid copying AI text directly into documents without revision and verification.

Hands-on Scientific Work

  • Do not use AI to generate lab data, measurements, or observations.
  • Lab protocols or safety instructions generated by AI require supervisor review and validation to ensure safety.
  • Do not delegate reasoning, data checks, or experimental decisions to AI.

Understanding and Justification

  • Only use AI output if you understand the underlying biological mechanisms, ecological theory, and statistical models.
  • You alone must justify all methodological and analytical choices.

Scientific Integrity

  • Do not use AI to obscure errors or compensate for poor-quality work.
  • All citations must be verified and drawn from peer-reviewed sources.
  • AI outputs are not peer-reviewed; responsibility for accuracy, scientific validity, and integrity lies entirely with the researcher.

4. Confidentiality and Reproducibility

  • Do not upload unpublished manuscripts, sensitive data, or confidential materials to public AI tools unless explicitly permitted.
  • Document all AI-assisted code and workflows to ensure reproducibility without AI access.
  • Follow all institutional and legal data protection requirements.

5. Required Transparency

When preparing a thesis, manuscript, or report, researchers must clearly specify:

  • Which content was authored independently.
  • Where AI assistance was used and its extent.
  • Whether every sentence can be explained and justified without AI.
  • Whether the choice of analyses can be fully justified without AI assistance.
  • Whether the reasoning and logic of the text can be independently explained.

Work that does not meet these transparency requirements is unacceptable. If uncertain, consult your supervisor.

6. Final Message

The purpose of this AI policy is to support scientific development, not replace it. AI may assist with tasks, but it does not substitute for scientific understanding, interpretation, or judgment. Reading and critically evaluating literature, drafting abstracts, and working through analyses yourself are far more valuable than taking shortcuts, because personal effort ensures that the results truly belong to you and that the knowledge and skills you gain will serve your growth as a researcher. Your time in the lab is an opportunity to develop expertise, reasoning, and professional judgment. These are qualities that no tool can replace.

Example of Transparency

This document was prepared with AI assistance:

  • Authored independently: The current version of this document is based on the sample document provided by Zsófia Horváth https://metacomlab.com/lab-ai-policy/ further improved by the Restoration Ecology Group
  • AI assistance: ChatGPT was used to grammar-check, refine phrasing, improve clarity, and generate alternative wording options.
  • Independent understanding: Every sentence can be explained and justified by the author without AI.
  • Analytical choices: All reasoning and methodological choices are the author's responsibility.

This demonstrates acceptable use of AI according to the group's policy.

Note: This document complements, but does not replace, university and institutional regulations. All legal and institutional requirements supersede these guidelines.