*AI in research needs standards, not just smarter tools.*
I recently came across an article in *Research Information* titled *“Developing standards for AI in research”*. It raises an important question: as AI becomes more common in academic research, how do we ensure that the information it provides can be trusted?
AI can help researchers explore literature, summarise papers and identify connections between studies. But speed and fluency do not guarantee accuracy.
A reference may look genuine but lead nowhere. A summary may miss important limitations. An answer may sound convincing while overlooking evidence that challenges its conclusions.
This is why responsible AI use in research needs to rest on three principles:
*1. Trusted sources* AI-generated claims must be supported by credible, verifiable research. Citations should always be checked against the original publications.
*2. Responsible use* Privacy, data protection, transparency and research integrity must be part of how AI tools are selected and used.
*3. Human judgement* AI can assist with analysis, but researchers must retain responsibility for interpreting evidence, questioning assumptions and drawing conclusions.
There is also a role for universities, research institutions and academic publishers. Clear standards, practical training and well-defined policies can help researchers use AI responsibly rather than leaving them to work out the risks on their own.
One point deserves particular attention: *a citation is not proof, and a confident answer is not necessarily a correct one.*
The real measure of AI in research should not be how quickly it produces an answer, but how effectively that answer can be examined, verified and used responsibly.
AI should support the research process without weakening the principles that make research credible in the first place.
*The question is no longer simply how to use AI in research, but how to use it in a way that deserves our trust.*
*Source: Research Information — “Developing standards for AI in research”* www.researchinformation.info/viewpoint/developing-standards-for-ai-in-research/