Prominent literary figure Margaret Atwood has issued a pointed critique of artificial intelligence, emphasizing that large language models are fundamentally constrained by their training data. Speaking at the Babylon Literary and Cultural Festival in Porto, Portugal, Atwood addressed the growing reliance on AI systems, describing the technology as a classic example of garbage in, garbage out. During the event, she recounted a single interaction with Anthropic’s Claude chatbot, noting that her attempt to retrieve information regarding the television series Father Brown yielded unsatisfactory results. Beyond the technical limitations, Atwood directed sharp criticism toward individuals who increasingly depend on automated tools, labeling them opportunists seeking to circumvent the rigorous demands of original thought and research. Her assessment underscores a persistent concern in the technology sector: while generative models have achieved remarkable scaling, their output quality remains inextricably linked to the accuracy, relevance, and recency of the datasets used for training. By highlighting the risks associated with deploying machine learning systems trained on scraped and potentially outdated web content, Atwood’s comments reinforce calls for greater transparency in data sourcing and model validation. The remarks arrive amid ongoing industry debates regarding AI reliability, copyright compliance, and the practical utility of consumer-facing generative tools. As technology firms continue to optimize model architectures, Atwood’s intervention serves as a cultural reminder that computational speed and scale cannot substitute for verified information and human critical judgment.