LLMs Echo Millennial Slang In Outputs

A Boston Globe senior correspondent reports that ChatGPT, Google Gemini, and other LLMs increasingly default to millennial-era language because they were trained on internet posts from roughly 2006–2020. The columnist cites repeated use of terms like "chaotic" and "unhinged" and notes multimodal models often reproduce dated visuals such as skinny jeans. The piece argues this reflects training-data bias and urges developers to update data and style controls.