How AI is Changing the Game for Policymakers – But Why Humans Still Hold the Keys

Brussels, 23 June 2026 – Imagine you’re a policymaker facing a mountain of research papers, conflicting data, and a looming deadline to draft a new climate policy. Now, picture an AI assistant that can sift through thousands of studies in minutes, highlight the most relevant findings, and even draft a first version of your report, while still letting you make the final call. That’s not science fiction; it’s the reality being tested today in places like the European Commission’s Joint Research Centre (JRC) and the OECD.

The AI Sidekick: Faster Research, Fewer Headaches

Alessandro Zani introduced RASS, the JRC’s prototype AI research assistant, designed to act like a supercharged intern. Need to review 500 studies on renewable energy transitions? RASS can:

• Scan and sort relevant papers from databases like Semantic Scholar or EU publications in seconds.

• Flag key insights and even draft a structured report (complete with sections tailored to your needs).

• Pause at critical points to ask, “Does this make sense to you?”, ensuring humans stay in the driver’s seat.

The catch? RASS isn’t just about speed; it’s about trust. Unlike commercial AI tools that operate as “black boxes,” RASS forces users to verify every step. “We’re not just automating,” Zani explained. “We’re making sure the human can question, adjust, and ultimately own the output.” Think of it as a GPS for policy research: it suggests the route, but you still decide whether to take the exit.

The Human Edge: Where AI Hits Its Limits

Margarita Kalamova from the OECD drove home a crucial point: AI excels at production, but humans excel at judgment. Her team’s research shows that while AI can draft reports, translate jargon into plain language, or even predict stakeholder reactions, it stumbles in three key areas:

• Reading the Room: Negotiating a partnership between governments, businesses, and NGOs isn’t just about data; it’s about reading unspoken tensions, building trust, and navigating decades of institutional history. “No AI can replace a handshake, or detect a sceptical eyebrow raise,” Kalamova noted.

• Spotting the Gaps: AI might confidently summarise a study, but it can’t always tell if the study itself is flawed or outdated. Humans must certify accuracy, especially when lives or reputations are on the line.

• Taking the Heat: When a policy backfires, it’s not the AI that faces the press or answers to voters. “Accountability can’t be outsourced,” Kalamova warned.

The OECD’s data reveals a telling split: 80–90% of routine tasks (like drafting reports or searching databases) can be AI-assisted, but the remaining 10–20% (high-stakes decisions, ethical dilemmas, or politically charged communications) demand human leadership.

The Big Picture: AI as a Tool, Not a Replacement

Both experts agreed: the future isn’t about AI versus humans: It’s about AI for humans. Zani’s team is testing RASS in real-world scenarios, while Kalamova’s work at the OECD helps governments anticipate where AI can help (and where it might misfire).

So what’s the takeaway for the rest of us? Whether you’re a policymaker, a teacher, or just a curious citizen, the message is clear:

AI can do the heavy lifting, freeing up time for the work that truly matters. The best AI tools don’t replace human expertise; they amplify it. As Dr. Zani put it: “We’re not building AI to think for us. We’re building it to help us think better.”