���ϳԹ���Դ

AI image generators can produce fake images and videos that are often indistinguishable from real ones and large language models can output false information that looks credible. Recent examples are the feature in Grok that could “undress” people in uploaded photos, and the use of fake sound and images in the portrayal of political candidates. The worry is that such uses of generative AI can harm the integrity of political processes and reduce social trust. The last Eurobarometer on disinformation and fake content finds that 69% of Europeans are highly concerned about misleading or false information online, and a 68% are highly concerned about AI‑generated fake content such as deepfakes. In the seminar we will present results from questions we ran in the last round of the NCP, on how worried people are about the potential harm of generative AI, both for themselves and for society. We discuss how the level of worry is influenced by experience with AI, self-assessed competence in distinguishing fake from real images, and age.