AI and politics put scientific expertise under pressure

Artificial intelligence is making knowledge more readily available, while political pressure is challenging scientists’ authority, SRF reports. Researchers say they must work harder to ensure reliable expertise remains visible and properly assessed.

Climate physicist Reto Knutti of ETH Zurich said the growing availability of data and AI was “fantastic,” but warned that quality control and context were often lacking. “Everyone feels they are an expert,” he said. In the OECD’s latest PISA study, 43 per cent of young people said they trusted their own common sense more than science.

Knutti said science was also facing political pressure in Switzerland, though the situation was not as extreme as in the United States under President Donald Trump, where funding has been cut and institutions dismantled. In Switzerland and elsewhere in Europe, he said, studies were selectively interpreted, publications delayed and press releases rewritten at the request of political authorities.

Silke Fürst, a communications researcher at the University of Zurich who studies AI’s influence on scientific expertise, said researchers needed to make sure AI systems could find and relay their work. Reliable information also needed to remain part of public discussion, including in the media, and AI tools should clearly disclose the data behind their answers. Users, she added, should remain sceptical and the data sources used by AI should become more transparent.

AI could also create opportunities for experts, said political scientist Michael Hermann of the Sotomo research institute. While AI can summarise studies, he said, it is less likely to replace a recognisable person with relevant experience. Experts who speak publicly and build a profile can offer people more than anonymous AI-generated information.

Hermann said criticism of science in some political circles had gained momentum through social media, where forceful personalities can polarise debate with extreme positions. He sees AI as potentially moving in the opposite direction, arguing that it tends to affirm users rather than make them feel belittled. The experts interviewed, as well as the AI tools SRF consulted, did not consider AI an existential threat to human experts.