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 summed up her concern as „garbage in, garbage out“. She said she had used Anthropic Claude exactly once to look up information about the British detective series Father Brown and received a wrong answer. Atwood stressed that Claude did not knowingly lie because a large language model is not human. Her point: the system can sound convincing while still being wrong.
Nauti's Take
This is not an anti-AI manifesto. It is a warning about beautifully packaged junk.
If you build AI products, sourcing, uncertainty, and verification need to show up in the interface. Otherwise you are not selling intelligence.
You are selling confidence with garbage in the tank.
Briefingshow
Atwood’s point lands on a central AI problem: output quality depends on training data, context and verifiable sources. In culture, history and literature, wrong details can look harmless, but they quickly shape research, teaching and public memory.
Sources