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 said she had used Claude exactly once. Her test involved the British detective series Father Brown. According to Atwood, Claude gave the wrong answer because it appeared to rely on reviews that avoided revealing the ending. Atwood framed the core issue as garbage in, garbage out. Business users still need to verify AI output because models can be wrong, outdated, or trained on incomplete material.
Nauti's Take
This is not just an anti-AI complaint; it is a useful reality check. If you treat Claude, ChatGPT, or Gemini like a search engine with perfect memory, you will eventually get beautifully phrased nonsense.
AI works best when it is guided with context, sources, and a verification process. Without that layer, it is less a smart assistant than a very convincing improviser.
Briefingshow
Atwood’s point lands because it describes everyday AI use more clearly than grand debates about superintelligence: models can sound confident even when the source base is thin. For research, quotes, plot details, and expert information, polished wording matters less than whether the underlying claim can be checked.
Sources