Margaret Atwood says the problem with AI is ‘garbage in, garbage out’
TL;DR
Margaret Atwood discussed AI at the Babell Literary and Cultural Festival in Porto and gave a blunt verdict: the core problem is garbage in, garbage out. She said she had used an AI chatbot exactly once: Anthropic Claude. When she asked about the British detective series Father Brown, Claude gave her a wrong answer. Atwood did not frame the mistake as a human-style lie, but as a system issue with large language models: they generate plausible text without knowing whether it is true.
Nauti's Take
Atwood’s point is sharp but fair: the problem is not simply that AI makes mistakes. The problem is that it often delivers mistakes in language that sounds authoritative.
For users, the lesson is practical: chatbots are useful tools, not truth machines. Anyone using AI for research needs sources, cross-checks, and the discipline to throw away a polished answer when it is wrong.
Briefingshow
Atwood points at a practical failure mode: AI can sound confident even when its source material is incomplete. The problem is not only hallucination, but source blindness: a model can turn partial, outdated, or deliberately incomplete writing into a polished answer. For teams, AI saves time only when verification, sourcing, and responsibility are part of the workflow.
Sources