The AI narratives project – a joint endeavour by the Leverhulme Centre for the Future of Intelligence and the Royal Society – has been examining how researchers, communicators, policymakers, and publics talk about artificial intelligence, and why this matters. This write-up presents an account of how AI is portrayed and perceived in the English-speaking West, with a particular focus on the UK. It explores the limitations of prevalent fictional and non-fictional narratives and suggests how practitioners might move beyond them. Its primary audience is professionals with an interest in public discourse about AI, including those in the media, government, academia, and industry. Its findings have been synthesised from discussions at four workshops, held in Cambridge and London between May 2017 and May 2018, and organised by the AI narratives project. This project had its origins in questions emerging from public dialogue carried out as part of the Royal Society’s report on machine learning.