Designing stronger research with AI: how institutions can support responsible AI use from the start

Much of the discussion around the impact of AI on research has focused on scientific writing and data analysis. However, AI is increasingly influencing an earlier (and equally critical) stage of the research cycle: experimental design.   

As more researchers integrate AI tools into their experimental design, institutions face a growing challenge. How can they enable innovation while ensuring that research quality, integrity and trust in scientific results remain uncompromised?

Responsible AI use starts with research design

Strong research starts with strong experimental design. Well-developed research questions, robust experimental design and appropriate methodologies are fundamental to conduct impactful research.  

Many researchers are already using AI to support this process, from brainstorming ideas and exploring literature to identifying knowledge gaps and refining study protocols. However, key risks around bias, hallucinations, copyright, transparency, reproducibility and over-reliance on AI have also raised many concerns within the scientific community.  

For research institutions, the increasing adoption of AI presents both an opportunity and a responsibility. Supporting researchers to use AI effectively requires more than providing access to technology. It requires guidance, training and a clear understanding of how AI can enhance their work without compromising scientific rigour, research integrity or critical thinking. 

As expectations around responsible AI use continue to evolve, institutions now play a critical role in helping researchers develop the skills and confidence needed to navigate this changing landscape. 

Practical guidance from distinguished panellists

To help address this need, Nature Masterclasses convened a panel on Designing Your Research with Assistance from AI with experts from journal policy, editorial and research practice spanning multiple disciplines:  

• Ellie Gendle, Head of Journals Policy – Research Integrity, Springer Nature  
• Oliver Graydon, Chief Editor, Nature Photonics 
• Tony Hu, Zhao-Yi Endowed Professor and Founding Dean, School of Biomedical Engineering, Tsinghua University 
• Ching-yu Huang, Chief Editor for Immunology, Nature Communications 
• Helder Nakaya, Associate Professor, Universidade de São Paulo  
• Samraat Pawar, Professor of Theoretical Ecology, Imperial College London 
• Veda Storey, Tull Professor of Computer Information Systems and Computer Science, Georgia State University 
• Nicki Tiffin, Professor, University of the Western Cape 
• Fatemeh Vafaee, Professor, University of New South Wales 
• Alexia-Ileana Zaromytidou, Chief Editor, Nature Cancer 

Together, the panellists provide diverse perspectives on publishing policy, research integrity, editorial standards and real-world applications of AI in experimental design.  

Their discussion addressed question submitted in advance of the session providing direct answers to the concerns researchers are increasingly voicing around responsible use, current limitations and emerging expectations. 

 Find the webinar recording here. 

What researchers and research-led organizations need to remember  

The webinar provides researchers with practical insights on how to:  

• Use AI to support research design, not to replace scientific judgement  
• Verify and leverage AI outputs to explore, strengthen and challenge their thinking  
• Protect sensitive information when developing research ideas or study materials  
• Build transparency, bias assessment and reproducibility into AI-assisted study design  

The biggest takeaway from the webinar is that AI should assist research decision-making, not replace it. Human judgement and subject expertise remain essential for designing strong, sound science.      

For research-led organizations, this means creating a supportive environment where researchers can safely adopt new technologies without compromising research quality or integrity by learning where AI adds value and where human judgment remains critical. 

Providing access to expert guidance from experienced editors and subject specialists helps researchers understand how to responsibly integrate AI into their workflows, while still adhering to publishing, editorial and research integrity standards. By empowering researchers to work more confidently and effectively in complex research environments, institutions can strengthen the impact of their outputs. 

Supporting responsible AI use during experimental design  

This new webinar recording complements Nature Masterclasses’ broader training portfolio, including Experiments: From Idea to Design, which guides researchers through developing, planning and refining impactful experiments.  

Together, this training content helps institutions support researchers at a time when both strong study design and AI literacy are becoming increasingly essential to research success. 

Why a proactive approach matters  

As AI capabilities continue to evolve, institutions have an opportunity to move beyond reactive guidance and towards supporting researchers proactively.  

By investing in researcher development, institutions can help their research community: 

• Use AI more confidently and responsibly  
• Align with emerging publishing and policy expectations 
• Strengthen research quality and integrity  

Investing in researcher training helps institutions build a culture of responsible AI use, where researchers can balance efficiency with their own critical thinking and subject expertise.