“Open science represents modern knowledge creation, but also supports evidence-based decision-making,” explains Prof. Dr. Julia Priess-Buchheit, professor at Kiel University with expertise in higher education, research integrity and open science. Open science also affects scientific exchange, that is, how knowledge moves between researchers, policymakers and society. “Generative AI is transforming this exchange,” she adds.
The question of how to help Europe shape this important transition is explored in the study ‘Open Science and the Effects of Generative Artificial Intelligence in Scientific Exchange‘, which will be presented at the STOA Panel meeting in the European Parliament in Strasbourg on Thursday 9 July 2026. Professor Priess-Buchheit is the coordinator and one of the authors of the report.
How would you explain to readers the relationship between Open Science, generative AI, and scientific exchange?
Julia Priess-Buchheit: Open Science stands for modern knowledge building, problem solving, and making informed decisions based on evidence rather than guesswork, so to say modern research. It emphasises that we have the possibility now to make research even more transparent, accessible, usable, and trustworthy. That influences in particular scientific exchange, which is how knowledge moves between researchers, policymakers and society.
Generative AI is transforming that exchange. On the one hand it can help researchers navigate vast amounts of information, communicate across languages and make knowledge more accessible. But, on the other hand, it can also generate convincing errors, obscure sources and blur responsibility.
The point of our report written for the European Parliament’s Panel for the Future of Science and Technology (STOA) is not whether scientists should use AI – we know they already do. The real question is whether AI will strengthen effectiveness, reliability and trust in research with and for society, or undermine this.
That is precisely why this report was commissioned: to help Europe shape this important transition rather than simply react to it.
Where do you see the biggest gap in scientific exchange within the European Research Area?
Julia Priess-Buchheit: The report identifies “inequity of access in the research ecosystem” as one of the five priority risks we have in Europe. For me, this is the biggest gap in scientific exchange within the European Research Area. The problem is not simply that some researchers cannot access information – meaningful participation not only depends on access to knowledge, but also on infrastructure, skills, support, and the capacity to verify and reuse scientific outputs.
This is where generative AI makes the issue more urgent. It may help with translation, summarisation and discoverability, but if “meaningful AI skills” and reliable infrastructures are concentrated in a few places, then scientific exchange becomes faster but less equal; Europe does not lack talent – the danger is that talent is not supported equally across the European Research Area.
The human role in fact-checking in science in the age of generative artificial intelligence is very important. Which human skills would you highlight as key and why?
Julia Priess-Buchheit: Human skills are essential. Researchers need three things: First, they need critical reasoning: the ability to ask, “Do I understand that. Would I be able – in theory – to reconstruct that knowledge?” GenAI can sound like stating facts while hallucinating.
Second, they need methodological judgement: the ability to understand whether the research process was solid. In research, it is not enough that an answer sounds good. We need to be able to process how the answer was produced.
Third, they need domain expertise: deep knowledge of the subject. Without this, it is very difficult to spot mistakes, missing context or weak arguments.
If the European Union removed all rules and nothing were regulated regarding the role of generative AI in scientific exchange, what consequences would this have for EU citizens?
Julia Priess-Buchheit: Without rules, several actors would effectively make rules themselves. The risk for society is not simply more misinformation. The deeper risk is a loss of trust in scientific knowledge. Citizens would find it increasingly difficult to know what is reliable, who is accountable and whether evidence has been properly checked.
Good regulation is not the enemy of innovation. It is what allows innovation to remain trustworthy. If we get this right, citizens benefit from better science and better decisions. If we get it wrong, trust becomes much harder to rebuild.
Given that GenAI content can appear highly convincing, what is the best safeguard for preserving scientific integrity?
Julia Priess-Buchheit: The best safeguard is transparency. In research, the key question is not only what result was obtained, but how it was produced: with which data, which methods, which tools, which assumptions, and under whose responsibility. It is a line of arguments that can be justified in a dialogue with other experts.
This becomes more important than ever in the age of generative AI, because AI-generated results can look polished and convincing even when the underlying process is based on statistical algorithms using tokens. That is why research must be transparent enough for the process to actually be understood. It is not sufficient to present a result at the end. Researchers need to make the chain of evidence visible, so others can see how a claim was constructed, checked and interpreted.
The danger is not only obviously false information. The bigger danger is plausible information that cannot be verified. If we cannot trace the process, we cannot properly judge the result.
So research reliability depends on two things: first, making research processes more transparent; and second, communicating both the process and the results much better.
In an AI-mediated research environment, useful research will not only need to be done well. It will need to be explained well. This is also in line with the report’s emphasis that, in AI-mediated scientific exchange, outputs should not only be open; their “origins, transformations, and conditions of production” should remain traceable and accountable.
How can we ensure that scientific exchange is not only open but also inclusive, without becoming concentrated around a small number of institutions or actors?
Julia Priess-Buchheit: We need to recognise that “open” does not automatically mean inclusive. So inclusion must mean more than open access. It must include access to skills, trusted infrastructures, multilingual support, verification tools and the ability to reuse and contribute to research. The goal should be open capacity, not only open information. Otherwise, scientific exchange may become faster and more open in theory, but more unequal in practice.
What is most important for the EU today, and how can this report help guide the EU in the long term?
Julia Priess-Buchheit: The report does not claim to predict the future. It is a report for parliamentary preparedness, not a single-path forecast. Its purpose is to help Members of the European Parliament and relevant committees see which risks in scientific exchange are emerging, where choices are needed, and how different policy paths could shape scientific exchange in the long term.
At the presentation in the European Parliament on 9 July, and later with the August publication, the report can help policymakers make decisions on how Europe should support trust, reliability, inclusion and strategic capacity in an AI-mediated research environment.
In the long term, the report can guide the EU by treating scientific exchange as a public-interest infrastructure. If Europe acts wisely, generative AI will be a huge benefit for Europe’s research.
Useful link:
• Watch the presentation of this report live online during the STOA Panel meeting in the European Parliament in Strasbourg on Thursday 9 July 2026 at 9:30h CEST.