AI image generation is transforming scientific illustration, but it raises important questions about ethics, transparency, and appropriate use in academic settings. Understanding these considerations is essential for researchers adopting AI tools.
This guide covers ethical guidelines, journal policies, and best practices for using AI-generated images in academic work.
AI tools are reshaping how researchers create scientific illustrations
The AI Revolution in Scientific Illustration
AI image generation tools have rapidly evolved:
2022: Basic AI image generators emerge 2023: Tools become viable for scientific use 2024: Scientific-specific AI tools mature 2025: AI illustration becomes mainstream in research
This evolution brings both opportunities and responsibilities for researchers.
Benefits of AI-Generated Scientific Images
Time Efficiency
- Traditional illustration: 4-8 hours per figure
- AI-assisted: 15-30 minutes per figure
- Time saved: 75-90%
Cost Reduction
- Professional illustrator: $200-1000 per figure
- AI tools: $0-20 per figure
- Cost reduction: 90-100%
Accessibility
- No design training required
- Democratizes scientific illustration
- Enables visual communication for all researchers
Iteration Speed
- Rapid prototyping of concepts
- Easy experimentation with styles
- Quick revisions based on feedback
Ethical Considerations
1. Accuracy and Misrepresentation
The core principle: AI-generated images must not misrepresent scientific reality.
Appropriate uses:
- Conceptual illustrations
- Schematic diagrams
- Visual metaphors
- Educational graphics
Inappropriate uses:
- Fake experimental data
- Fabricated microscopy images
- Synthetic photographs presented as real
- Manipulated results
2. Transparency and Disclosure
Best practice: Always disclose AI use
Disclosure elements:
- Tool name (e.g., "SciDraw", "DALL-E")
- What was generated vs. human-created
- Any post-processing applied
- How accuracy was verified
Example disclosure:
Figure 1 schematic was generated using SciDraw AI
with subsequent refinement in Adobe Illustrator.
Molecular structures were verified against
published crystal structures (PDB: XXXX).3. Intellectual Property
Key questions:
- Who owns AI-generated images?
- Can they be used commercially?
- Are there training data concerns?
Current consensus:
- Check tool's terms of service
- Most scientific AI tools grant usage rights
- Commercial use policies vary
- Keep records of generation
4. Training Data Ethics
Considerations:
- Was training data ethically sourced?
- Does output resemble specific artists' work?
- Are there copyright concerns?
Mitigation:
- Use tools with transparent training
- Avoid prompts targeting specific styles
- Generate original compositions
Journal Policies on AI-Generated Images
Current Landscape (2025)
Most major publishers now have AI policies:
Nature Portfolio:
- AI-generated content must be disclosed
- Cannot be used for scientific data
- Acceptable for conceptual illustrations
- Authors responsible for accuracy
Science/AAAS:
- Transparency required
- Clearly label AI-generated content
- Cannot replace experimental evidence
- Subject to standard figure guidelines
Elsevier:
- Disclosure in methods section
- AI tools not listed as authors
- Acceptable for appropriate uses
- Must meet figure quality standards
ACS Publications:
- Disclosure expected
- Conceptual use acceptable
- Cannot misrepresent data
- Standard TOC requirements apply
Disclosure Requirements
What to disclose:
- Use of AI tools (which ones)
- What was generated
- Human oversight/editing
- Verification methods
Where to disclose:
- Figure legends
- Methods section
- Acknowledgments
- Data availability statement
Best Practices for AI Use in Research
1. Use AI for Appropriate Tasks
Ideal applications:
- Graphical abstracts
- TOC graphics
- Conceptual diagrams
- Educational illustrations
- Review article figures
- Presentation graphics
Avoid for:
- Primary data representation
- Microscopy/imaging results
- Photographs of experiments
- Any "evidence" figures
2. Verify Scientific Accuracy
Verification checklist:
- Structures are chemically/biologically correct
- Processes are accurately represented
- Labels use proper nomenclature
- Scale is appropriate
- No misleading simplifications
3. Maintain Human Oversight
AI is a tool, not a replacement for expertise:
- Review all generated content critically
- Edit and refine outputs
- Add precise details manually when needed
- Verify against primary sources
4. Document Your Process
Keep records of:
- Prompts used
- Tools and versions
- Modifications made
- Verification steps
5. Consider Your Audience
For technical papers: Higher accuracy standards For public communication: Ensure accessibility doesn't compromise accuracy For education: Balance simplification with correctness
Implementing AI in Your Workflow
Starting Safely
Week 1-2: Learn the tool
- Practice with non-critical images
- Understand capabilities and limitations
- Develop effective prompting skills
Week 3-4: Internal use only
- Create images for presentations
- Generate drafts for lab meetings
- Get feedback from colleagues
Month 2+: Publication use
- Follow disclosure guidelines
- Verify accuracy rigorously
- Keep documentation
Quality Control Process
- Generation: Create initial image with AI
- Review: Check for scientific accuracy
- Refinement: Edit to correct any issues
- Verification: Have domain expert review
- Documentation: Record process and tools used
- Disclosure: Add appropriate attribution
Building Lab Protocols
Consider creating lab guidelines for AI use:
- Approved tools and uses
- Required verification steps
- Documentation requirements
- Disclosure templates
Addressing Common Concerns
"Is using AI cheating?"
No, when used appropriately.
AI tools are productivity aids, like:
- Spell checkers
- Reference managers
- Statistical software
- Image editing software
The researcher's expertise, verification, and judgment remain essential.
"Will journals reject AI-generated figures?"
Not if disclosed and appropriate.
Journals object to:
- Undisclosed AI use
- AI-generated "data"
- Misleading representations
Journals accept:
- Disclosed conceptual illustrations
- AI-assisted diagrams
- Properly attributed graphics
"What about reproducibility?"
Document thoroughly.
Include in your records:
- Exact prompts used
- Tool version
- Date of generation
- Post-processing steps
This enables reproduction or recreation if needed.
"Can AI replace scientific illustrators?"
No, but roles are evolving.
AI changes the illustration landscape:
- Simple illustrations: AI can handle
- Complex custom work: Still needs professionals
- Quality standards: Remain important
- Hybrid workflows: Increasingly common
The Future of AI in Scientific Illustration
Short-term (2025-2026)
- More journals establish AI policies
- Scientific-specific tools improve
- Best practices become standardized
- Disclosure becomes routine
Medium-term (2026-2028)
- AI integration into manuscript systems
- Automated accuracy checking
- Style consistency across figures
- Real-time collaboration features
Long-term (2028+)
- AI assists in figure planning
- Automatic figure generation from data
- Interactive dynamic figures
- Universal standards established
Resources for Staying Current
Policy Updates
- Monitor journal guidelines
- Follow publisher announcements
- Check professional society statements
Community Discussion
- Twitter/X academic communities
- Research integrity forums
- Professional society meetings
Training
- Publisher webinars
- Tool documentation
- Peer workshops
Practical Guidelines Summary
DO:
- Disclose AI use transparently
- Use for conceptual illustrations
- Verify scientific accuracy
- Maintain documentation
- Follow journal policies
- Apply human judgment
DON'T:
- Generate fake data
- Misrepresent AI images as photographs
- Use without disclosure
- Skip accuracy verification
- Ignore journal guidelines
- Abandon critical thinking
Getting Started with Ethical AI Use
Ready to use AI responsibly in your research?
- Visit SciDraw for scientific illustration
- Start with appropriate use cases
- Develop verification habits
- Implement disclosure practices
- Stay updated on guidelines
AI is a powerful tool for scientific communication. Use it wisely.
Related Articles:
