Margaret Atwood says AI chatbots can give wrong answers

In short: Author Margaret Atwood said she tried an AI chatbot once, got a wrong answer, and thinks people should not trust these tools without checking.

What happened

Margaret Atwood, the author of The Handmaid’s Tale and The Blind Assassin, spoke about AI during an interview at the Babell Literary and Cultural Festival in Porto, Portugal.

According to a recap from Deadline that was reported by The Verge, Atwood said she has used an AI chatbot only one time. The chatbot was Claude, made by the company Anthropic. She said she was looking for information about the British detective TV series Father Brown.

Atwood said Claude gave her the wrong answer, or what she described as a lie. She added that the system did not know it was lying because it is not a person. She described it as a “large language model,” meaning software trained by reading huge amounts of text, like a student who learns by skimming many books and articles.

Atwood said the chatbot likely relied on TV reviews that avoid spoilers, so it did not have the full details it needed. She also criticized people who depend on AI for an “easy way out,” and said users, including businesses, have to check the output because it can make mistakes.

Why it matters

Many people now use chatbots to answer questions quickly, like a search engine that talks back. Atwood’s example is a reminder that these tools can sound confident while still being wrong, especially when their training data is incomplete or outdated. For regular users, the practical takeaway is simple: treat AI answers like unverified tips from a stranger, and double-check anything important.

Source: The Verge AI