AI poses risks to data accuracy in science and politics

Artificial intelligence is presenting new challenges for data accuracy across scientific and political sectors. In the field of citizen science, researchers from the Cornell Lab of Ornithology and iNaturalist have warned that AI-generated or AI-enhanced images are compromising biological data collection. While some users use AI to improve image clarity, the technology can inadvertently reconstruct areas with inaccurate information or add features from different species, leading to false wildlife observations.

In the political sphere, a study by the Fondation Jean-Jaurès indicates that generative AI is becoming a significant source of political information, particularly ahead of the 2027 French presidential election. However, tools like ChatGPT and Perplexity frequently commit factual errors despite maintaining a neutral and authoritative tone. The study found that these AI models rely heavily on traditional journalistic sources, such as Le Monde, rather than direct communications from political candidates or social media.