Emma Hoes, Fabrizio Gilardi. "Existential risk narratives about AI do not distract from its immediate harms." Proceedings of the National Academy of Sciences (2025). https://doi.org/10.1073/pnas.2419055122