Last year (pre the chatGPT bonanza) I was using GPT-3 to generate some content about attribution bias and the responses got much spicier once the prompt started

Last year (pre the chatGPT bonanza) I was using GPT-3 to generate some content about attribution bias and the responses got much spicier once the prompt started including the typical HN poster lingo, like "10x developer": https://sonnet.io/posts/emotive-conjugation/#:~:text=I%27m%20always%20right%2C%20you%27re%20always%20wrong https://sonnet.io/posts/emotive-conjugation/#:~:text=I%27m%2... My conclusion was that you can use LLMs to automate and scale attribution bias. We did it guys!