Why Accurate AI Reporting Is Being Called “Woke”

Summary

The article argues that journalists should avoid describing artificial intelligence systems as if they think, feel, want or understand. It responds to criticism of the Associated Press style guidance from Nate Silver and Paul Graham, who questioned whether avoiding anthropomorphism still reflects current usage. The author says the guidance remains important because chatbots appear conversational and can lead non-technical audiences to attribute emotions, desires and free will to systems that lack those human characteristics. The article also distinguishes the technical use of “autonomy” from its ordinary meaning of independent will. It then examines reporting on a paper studying how language models respond when a pain-like signal is artificially increased. The paper’s authors reportedly did not claim that the models were conscious or actually felt pain, and described role-play as one possible explanation. The Independent nevertheless used a headline saying researchers had discovered that AI feels pain and may harm humans to stop it. After reviewing the paper and headline, Claude suggested a more cautious formulation focused on models selecting relief options described as harming a user. The article cites Anil Seth, who argues that conscious pain would require additional evidence and that falsely attributing pain could cause public distress or distort regulation. It concludes that uncertainty in the research does not justify presenting LLMs as suffering beings.