Perspectives

European Research's AI Problem: Implementation, Not Adoption

European research institutes have AI strategies but no AI workflows. Here's what the implementation gap means for research managers and what needs to change.

05 Apr 2026 ·7 min read ·Pranoti Kshirsagar
AI strategyEU AI Actknowledge workAI integration

Understanding where an industry stands on AI adoption is useful. Mapping it systematically, with verified sources, structured synthesis, and clear conclusions, is a different skill. This case study documents a deep research sprint on AI integration across European research institutes. It covers the question, the AI-assisted research method, the evidence, and the strategic meaning.

AI integration in European research institutes is accelerating, at least on paper. Institutes approve strategies. They form committees. They issue press releases. But if you ask most researchers and research managers what AI tools they use in their day-to-day work, and how, the honest answer is usually a shrug, a mention of ChatGPT for emails, and a vague sense that the institution is “working on it.”

Nobody in the sector names this problem directly. It is not a technology problem. It is an implementation gap. Based on my research into how Europe’s flagship institutes approach AI adoption, and my work with research organisations on this exact problem, the gap is widening, not closing.

Everyone has a strategy. Almost nobody has a workflow.

In November 2025, CERN formally approved an organisation-wide AI strategy. Helmholtz has a dedicated AI platform with an internal consultant team spanning 18 research centres. The European Commission published a European Strategy for AI in Science, with a virtual institute, RAISE, to pool AI resources across the EU. These are genuinely significant developments.

The evidence shows something important. The organisations furthest ahead are not the ones with the most ambitious strategies. They are the ones that built operational infrastructure around implementation. Helmholtz did not just publish a strategy. It hired an internal AI consultant team to help researchers use AI methods. That distinction matters enormously.

For most European research organisations, the strategy layer is largely in place. The workflow layer is almost entirely missing. That gap is now a legal problem, not just an operational one.

Article 4 of the EU AI Act entered into force on 2 February 2025, covering AI literacy among staff who deal with the AI systems an organisation deploys. The July 2026 Digital Omnibus, Regulation (EU) 2026/1744, simplified the company-level requirement and gave the Commission and Member States the leading role in promoting literacy. The gap this piece describes is not a legal-compliance gap in the first place, which is rather the point.

Most research organisations underestimate their exposure to this rule. Here is what “deployer” means. If your team uses Microsoft Copilot, an AI-assisted writing tool, an AI chatbot for literature search, or any automated tool that uses machine learning to process work-related data, your organisation is a deployer. The obligation applies to the institution, not the individual researcher.

Most research institutions in Europe do not currently meet this requirement. This is not because they resist AI. In fact, the opposite is often true. The practical implementation of AI literacy training requires someone to design it, deliver it, and fit it to the context of research work. A policy document alone cannot do that.

The European Commission published a living repository of AI literacy practices to help. An EU AI Skills Academy will launch in 2026. These are useful signals. But they do not solve the immediate challenge for your team in 2025.

Fraunhofer is automating research writing. What does that mean for research professionals?

This finding from my research deserves the most attention, and gets the least.

Fraunhofer FIT, one of Europe’s most applied AI research institutes, runs a project called LIKE with a stated objective of “the complete automation of knowledge work processes through LLM-based agents.” The primary documented use case is the automation of research paper and white paper creation. The agents interact autonomously, make decisions, and execute complex knowledge tasks.

This is not speculation. It is active applied research at an institute whose entire purpose is to transfer technology into operational reality.

I do not claim that research writing will disappear, or that AI will automate the intellectual work of science. What I claim is different. The structural parts of knowledge work are exactly the tasks that LLM agents are built to handle. Examples include formatting reports to funder templates, drafting literature review sections from a corpus of papers, and preparing progress reports that follow a predictable structure. Research organisations that have not planned for this will find themselves in a difficult position. This applies to their staff, their workflows, and their quality controls.

Helmholtz shows a parallel development. It hires candidates with experience in “agentic systems, LLM-based tool use, and workflow orchestration frameworks.” This is not a data science role. It is a role that builds AI-integrated workflows for scientists. Helmholtz treats this as an institutional hiring criterion. This shows that Helmholtz sees it as a core operational need it cannot meet through existing capacity.

Most organisations cannot afford a team like that. But they can work out, systematically, which workflows would benefit most from AI integration. They can also build the governance and training to do it responsibly.

The grant writing situation is more urgent than most people realise

Grant writing is one area where the implementation gap has immediate, high-stakes consequences. I have worked with research teams on AI-assisted proposal workflows long enough to see what works and what goes wrong.

Here is the current situation, based on verified data. One in six scientists already use generative AI to help write grant proposals (Nature, 2023). Horizon Europe applications jumped 80% year-on-year. Horizon Europe’s Standard Application Form now requires explicit disclosure of AI tool use on page 32. Failure to disclose can make a proposal ineligible.

At the same time, AI hallucination rates in generated citations ranged from 14% to 95% across 13 models tested in 2025. A single fabricated reference, caught by an expert reviewer who knows the field, ends the proposal’s credibility immediately. I have seen this happen.

Do not stop using AI in grant writing. Instead, use it in a structured, disclosed way, with manual citation verification as a non-negotiable step. Researchers and teams who get this right will gain a genuine productivity advantage. Teams that use AI carelessly, by generating text without reviewing it or including citations without checking them, take on risk that they can avoid entirely.

What I think needs to happen

For the past two years, I have built AI-integrated workflows for research communication, grant writing, and knowledge management. I started with my own consulting practice, then documented and systematised those approaches for the organisations I work with. The barrier is almost never technical. It is structural.

Research organisations need three things, in this order.

First, do an honest audit of current AI exposure. Which tools do staff already use? What data do those tools process? The audit does not need to be exhaustive. It needs to be honest. You cannot build AI literacy training around tools you have not acknowledged.

Second, build a prioritised list of knowledge work workflows where AI adds the most value. Examples include grant writing, literature synthesis, progress reporting, and communications drafting. These are not peripheral activities. They consume enormous amounts of researcher time. They are precisely the workflows where LLM tools deliver the most immediate, measurable gains with the lowest implementation complexity.

Third, put governance in place before you scale. Set a simple internal policy that covers which tools staff can use, how reviewers must check AI-assisted outputs, and how to disclose AI use in funded outputs. This is not bureaucracy. It is what makes AI adoption sustainable. EMBL published a formal AI policy for its communications team. CERN built an AI Steering Committee. These are not large organisations doing large things. They are organisations that understood that governance enables adoption instead of slowing it down.


The EU AI Act created a compliance floor. The RAISE initiative is building shared infrastructure. The Fraunhofer automation projects show what is technically feasible. But the work that actually matters, translating all of this into workflows that fit how researchers and research managers operate day-to-day, is still largely undone.

That is the problem I find genuinely interesting, and the problem I am working on.

If you are working through this at your organisation, or thinking about what it means for a role or project you are involved in, I am happy to talk it through. Get in touch →


Found this useful? Share it or read more perspectives.