Last December in Rome, at the First Global Consensus Conference on Y90 Radioembolization for Hepatocellular Carcinoma,
nearly a hundred experts from around the globe reached an agreement on 35 out of 36 clinical recommendations in just a day and a half. The credit goes to the “Assisted Consensus” system developed by Ex Machina. Powered by their proprietary COSMO42 framework, the system analyzed roughly 600 scientific publications and guided specialists toward common ground.
The process begins by crunching hundreds of scientific papers, all validated by human experts through an “Expert-in-the-Loop” approach. During the conference, the algorithm processes comments in real-time, reframing proposals to break through the typical deadlocks of debate. The results speak for themselves: 80% of the system’s suggestions were accepted by the experts, with accuracy peaking at 93.8% in its most refined automations. The AI acts as a neutral agent, smoothing over friction caused by individual egos or status by crafting language that satisfies the majority.
Ex Machina’s “Assisted Consensus” isn’t just for doctors. It could offer a “third way” for political bodies, such as the Ticino Grand Council, which is currently debating how to limit speech times in the chamber. AI could preemptively map out positions to identify areas of technical agreement, saving floor time for genuine political debate on the most contentious issues. The goal isn’t to replace human judgment, but to make it more effective.
Taiwan and Estonia have already proven that technology can supercharge democracy. Taiwan utilizes vTaiwan, a platform that maps consensus on complex issues, fostering cross-party agreements that lead to concrete government action. Meanwhile, Estonia has woven digital participation into its national fabric with Rahvaalgatus.ee, where citizens propose and sign legislative initiatives using their digital IDs—a modern take on the referendum.
The Ex Machina experience teaches a simple lesson: AI works best when it empowers the human decision-maker rather than replacing them. Whether defining medical guidelines, voting on a corporate budget, or debating a new law in Parliament, the objective remains the same: spend less energy managing complexity and more time on the quality of the final decision. Real efficiency doesn’t come from silencing voices or letting them talk in circles—it comes from helping them harmonize.
