Research integrity in an age of AI: Thought leadership webinar
Gen AI is becoming an integral part of research, raising concerns about research integrity, transparency, and trust. Explore the key takeaways in our most recent webinar in partnership with the European Alliance for Social Sciences and Humanities and leading experts in the field.
This post originally appeared on Epigeum blog here.
Generative AI is rapidly transforming how research is conducted, analysed, and communicated. From literature reviews and data analysis, AI tools are becoming increasingly embedded in the research landscape. While AI offers clear benefits, it continues to raise important questions about research integrity, transparency, and trust.
These themes were at the heart of our most recent webinar – in partnership with the European Alliance for Social Sciences and Humanities (EASSH) - which brought together leading experts to explore what research integrity means in an age of AI. Through short presentations and a live panel discussion led by Katie Metzler (Vice President, Social Science Innovation at Sage), Professor Jennifer Edmond (co-author of ‘The Trouble with Big Data’ and the recently released ‘AI for Democracy’), Nick Vangheluwe, PhD (Policy Officer at the European Commission's Directorate-General Research & Innovation) and Dr Rula M. Al Abdulrazak (Senior Lecturer and founder of the AI with Integrity initiative at the University of East London) explored how researchers and institutions can navigate the opportunities and challenges of AI responsibly.
The impact of AI on research practice
Opening the discussion, Professor Jennifer Edmond shared that, while technological change is nothing new to academia, the pace at which generative AI is developing presents a unique challenge of adapting quickly. One of the key tensions highlighted in the presentation was the balance between efficiency and effectiveness.
Edmond explored the practical concerns associated with AI use in research including hallucinations with data, potential biases in outputs and ‘an illusion of explanatory depth’ - where AI generated content can appear to passively recognise ‘something that looks credible’. Combined with the fact that many AI tools are developed by commercial organisations with strong profit motives – as seen by the amount of content increasingly being scraped to train models - these factors have contributed to a growing divide between researchers who embrace AI and those who are deeply sceptical.
Rather than viewing AI as the cause of challenges facing research today, Professor Edmond argued that the current evolution of AI has exposed and further highlighted weaknesses already present in the research ecosystems.
Policy evolving with AI advancements
Providing a policy perspective, Nick Vangheluwe shared how the European strategy for AI in science aims to empower researchers to utilise the potential of AI to accelerate scientific discovery, enhance the scientific discovery and amplify the impact of science through responsible use.
The living guidelines developed by the European Research Area (ERA) Action on AI in Science support responsible use of AI with considerations of the limitations, environmental and societal impact as well. Built on the core principles of reliability, honesty, respect and accountability, the guidelines provide a practical framework for researchers, helping them navigate the current ecosystem and AI technologies.
Empowerment beyond compliance: From human-in-the-loop to human ownership
Building on the discussion, Dr Rula Abdul Razak shared that while the policy gap is closing, a more pressing challenge remains - implementation. Many researchers still lack confidence in using AI because they do not yet have the clear guidance, practical tools or institutional support needed to navigate its risks responsibly.
Rather than passively reviewing outputs, researchers must own the process from defining the purpose and boundaries of the research question, to overseeing the final judgement including accuracy and compliance. By creating a consistent framework that researchers can follow and a safe space for disclosure, institutions can empower researchers with confidence to make the judgments. This is especially important for early career researchers, who may be more vulnerable to these risks. In conclusion, the conversation highlighted an important shift in moving beyond compliance towards human researcher-led ownership.
To explore more of these themes and watch the panel discussion, catch up on the webinar.