ChatGPT fixes one of its most recognizable language tics

Deal Score0
Deal Score0

ChatGPT gets closer to humans! If it was easy to spot a text written in AI, the task turns out to be more complicated, because the chatbotOpenAI gets rid of his language tics.

More precisely, it is the em dash (or long dash) which disappeared with the arrival of ChatGPT 5-1. This development makes generated texts more natural and difficult to distinguish from those written by a human.

The end of the ChatGPT dash

Please note, this text below is written by artificial intelligence, and there is a very easy way to guess it.

Les Numériques is a French media dedicated to high-tech and digital technology — smartphones, computers, connected objects, audio and household appliances. Each product is tested according to technical and practical criteria to guide consumers. The site also covers technological news – conferences, trade fairs, innovations such as AI or 5G – and analyzes their impact on daily life.

It is unmissable! The em dash floods the texts with ChatGPTor rather flooded, because he is now missing since update 5-1.

A short grammar lesson to put the character in context: originally, the em dash (—) is used to mark a strong pause or introduce an explanation into a sentence. Some writers or Internet users like to use it, but it is normally quite rare to come across it in everyday life.

It is the complete opposite of its cousin: the en dash (–), also called the 6 dash, often confused with the first. It is shorter and is used for intervals, to connect two elements, or even in a list.

A mocked, but emblematic dash

As we told you, normally, the em is rare, but with ChatGPT, it had become almost systematic and revealing of the use of AI. Pointed out, he was the center of much mockery and even nicknamed the “ChatGPT hyphen“.

GPT chat on phone

The em dash, which has become the symbol of texts generated by ChatGPT, has been derided on the networks.

© Shutterstock

OpenAI humorously acknowledged the em-dash problem going viral, going so far as to have ChatGPT issue an apology. The company now ensures that the model follows personalized guidelines for this dash, allowing users to control its frequency.

For the moment, access to GPT-5.1 remains limited: OpenAI deploys it first on paid versions before extending to free accounts.

More Info

We will be happy to hear your thoughts

Leave a reply

Bonplans French
Logo