Mathis Grosjean

Mathis GrosjeanAug 5, 2026

“do you want me to do that next?” someone sold a travel guide they did not read we know because chatgpt was still asking what to write next inside the printed book..... this is funny until you realize the customer paid to become the editor this is what ai slop actually does one writer saves time then every customer has to 1. spend their own time checking if the places exist 2. if the history is real 3. if the advice can be trusted we count the time saved by the person using ai we never count the time wasted by everyone receiving their work think about it in your company when people are sending you ai slop, they make you do the work! then we call it productivity if your ai saves you time by making people verify what you made, you did not automate the job you made your customer & collaborators do it for you

Mathis GrosjeanJun 12, 2026

if you're not hitting this you're part of the permanent underclass they said so i went and did the thing did you?

Mathis GrosjeanJul 30, 2026

sometimes chatgpt does not summarize the internet it returns a Reddit thread with NO modifications if LLMs trust reddit, marketers manufacture reddit LLMs are becoming a main entry point for discovering products so, naturally, people want to sell through them. one tactic is already obvious: flood reddit with posts and comments that look like genuine recommendations it feels similar to what happened with google first, useful websites ranked then everyone learned the rules and search results filled with listicles written mainly to satisfy the algorithm now the same optimization game is moving into places that LLMs treat as human evidence but if reddit becomes full of fabricated opinions, what comes next? probably a shift from “which source said it?” to “what evidence supports it?” - verified purchases - long-term reputation - independent sources - real product usage - claims that are expensive to fake ... or something else? every distribution channel eventually gets optimized the interesting question is how (and if) LLMs will learn the difference between a recommendation and a recommendation-shaped advertisement...