European research integrity code updated to reflect advances in artificial intelligence

A new version of the European Code of Conduct for Research Integrity has been published that includes guidance on artificial intelligence (AI), navigating EU data protection laws and how to approach changes to research impact assessments.

The code reflects recent advances in AI, in particular the public release of the large language model-powered ChatGPT in November 2022, which brought home realisation of the power of AI to facilitate plagiarism and falsification of data and to undermine research integrity.

There are two ways in which AI could influence research integrity, according to the code. The first is in using applications like ChatGPT to generate data and write papers, the second is journal publishers using it to run quality checks on articles, for example checking papers are not plagiarised and that graphs and figures are faithful representations of the underlying data. 

“In both of these instances, the focus is really on transparency,” said Mathijs Vleugel, scientific policy officer at ALLEA, the European federation of academies of sciences and humanities, which drew up the code. “These are tools that are available to the community, which in principle, have great benefits and should be used in the best way possible. But it's important that we are transparent that these tools are being used. That’s where the focus is from our side.” 

Vleugel said AI is likely occupy a larger place in the next code of conduct, which should be published in the next three to five years. “AI will be a big game changer in all different fields,” Vleugel said. “I think we really have to see how to find the best way of using these tools, but still, as human beings stay in control of what comes out of them.”

The code of conduct, which is recognised by the European Commission and other research organisations around Europe, was last revised in 2017 and another new component is guidance on how researchers and institutions should navigate the EU’s General Data Protection Regulation (GDPR) that came into force in 2018.

The code takes a broad stance on the legislation, advising researchers and institutions to follow rules set out in it, but is not prescriptive on the implications for research integrity. This is intentional, according Vleugel, who acknowledged the challenges that some researchers have faced in following GDPR. This is particularly true in medical research where the legislation makes it more difficult to share health data with collaborators outside the EU or European Economic Area.

Navigating GDPR and all other aspects of research integrity has to be a shared responsibility, Vleugel said. “I think this is one of the main things that has changed compared to the last version of the code,” he said. “Instead of putting the responsibility for this purely on the researchers themselves, we have a much stronger emphasis now on the institutions and their responsibility.”

Without this backing, “We will not have the environment and the infrastructure where researchers can actually follow the good practices that are in the code,” he said. 

Another area that the new code takes on is the ongoing attempt to reform how research impact is assessed in the EU, with a shift to putting more emphasis on quality of research, rather than quantitative metrics such as the number of papers published. Over 500 institutions have joined the Coalition for Advancing Research Assessment (CoARA) and signed an agreement pledging to implement new standards of research assessment.

Again, it is a shared responsibility to ensure the emphasis on quality over quantity happens in practice. “There is a lot of pressure on researchers, with funding becoming tighter and competition higher. So, there is an incentive to cut corners,” Vleugel said. “We really hope [our code of conduct] will help the individual actors, but especially, I think, the institutions, to focus more on research integrity.”