0
Large-language models such as ChatGPT are still liable to distort the meaning of what they are summarising, says academic Seth Lazar