How can I tell if text is written by ChatGPT?

Look for patterns like the rule of threes, monotonous sentence structure, hedging, and unusual punctuation like excessive em dashes, though these are becoming less common.

Are there tools to detect AI-generated writing?

Yes, tools like ZeroGPT and Grammarly's AI Detector can help identify AI-generated text, but they are not always 100% accurate.

Why did ChatGPT stop using em dashes?

OpenAI reportedly removed ChatGPT's fondness for em dashes in response to user feedback and to make the AI's output more customizable and natural-sounding.

Home / Technology / AI Writers' Em Dash Habit Ends: Detectability Plummets

AI Writers' Em Dash Habit Ends: Detectability Plummets

17 Dec

•

Summary

ChatGPT will now avoid em dashes when prompted, making AI text harder to detect.
New AI detection tools are available but not entirely foolproof.
Key AI writing tells include the rule of threes and hedging language.

AI Writers' Em Dash Habit Ends: Detectability Plummets

Recent advancements in AI text generation have made it increasingly difficult to distinguish between human and machine writing. A notable characteristic of older AI models, such as ChatGPT's reliance on em dashes for stylistic effect, has reportedly been phased out following user complaints and a commitment to customization. This shift poses challenges for those needing to verify authorship.

Fortunately, detection tools like ZeroGPT and Grammarly's AI Detector are available, offering a means to scan text for AI origins. However, these tools are not entirely foolproof, underscoring the continued importance of human discernment. Recognizing specific linguistic patterns can aid in identifying AI-generated content more reliably.

Several telltale signs of AI writing have emerged, including an excessive use of the "rule of threes," arguments framed with contrasting language, monotonous sentence structures, and the frequent insertion of short, unnecessary rhetorical questions. Additionally, AI often employs constant hedging and indirect language, leading to vague responses despite an appearance of nuance.