OpenAI says ChatGPT will now drop dashes if you tell it to. There is a distinctive sign that supposedly signals text written by artificial intelligence jumped out up everywhere in recent months, including in school newsletters, emails, comments, customer service chats, LinkedIn posts, online forums, ad copy, and more. Adding the dash caused people to criticize these writers for being slothful and turning to an AI chatbot to do their job.
Of course, many also argued in favor of the dash, arguing that it was part of their writing long before LLMs adopted punctuation. However, the fact that chatbots could not avoid its employ made the so-called “ChatGPT connector” an objectionable up-to-date addition to any text, even if it was not a reliable signal of content produced by generative AI.
The problem had embarrassed OpenAI for some time, as ChatGPT users were unable to force the chatbot to stop using this symbol, even if they explicitly asked it to do so.
Now OpenAI CEO Sam Altman says the company has resolved the problem. IN entry on XAltman writes: “If you tell ChatGPT not to use em-dashes in custom statements, it will eventually do what it’s supposed to do,” calling the update “a small but happy victory.”
The the company explains in a post on Threads (where he forced ChatGPT to apologize for “destroying the hyphen”) that ChatGPT will be better at not using the hyphen if you instruct it not to do so via custom instructions in the personalization settings. This means it won’t necessarily eliminate the dash from your results by default, but at least you’ll have more control over how often it appears.