Margaret Atwood calls AI garbage in, garbage out at Porto festival

Margaret Atwood told a packed BABELL audience in Porto that artificial intelligence is only as good as the material it ingests, after saying she had used Anthropic’s Claude exactly once and came away unimpressed. The Handmaid’s Tale author said she tried the chatbot for a spoiler on the British detective series Father Brown, only to get the wrong answer, or a lie, before summing up her view: “The thing about AI is that it’s garbage in, garbage out.”

The session was one of the most sought-after events at the Babell Literary and Cultural Festival, which runs June 24 to 29 in Porto, Portugal, and is being promoted by Fundação Livraria Lello as part of an effort to turn the city into a Book-City. Organizers said Atwood’s appearance had been sold out for weeks before they released 100 additional seats on June 15.

Atwood’s conversation was moderated by Tânia Ganho, with readings by Mia Tomé. The appearance sat within a program built around internationally known writers, including Byung-Chul Han, Alberto Manguel, Olga Tokarczuk and Salman Rushdie, with the festival also highlighting the presence of two Nobel laureates. That lineup has been central to the effort to position Porto as a literary destination rather than just another stop on the festival circuit.

Her remarks landed in the middle of a broader dispute over whether AI systems can be trusted to produce accurate answers at scale. Anthropic describes Claude as an AI safety and research product aimed at building reliable, interpretable and steerable systems, but its own documentation says hallucinations are not eliminated entirely and that critical information should always be validated. Atwood’s experience with a single query, and her conclusion that the model could not know it was lying because it was not a human being, sharpened a familiar complaint about large language models: whatever fluency they display, the quality of the output still depends on the quality of the input.