Post

I’ve seen a lot of my colleagues’ interactions with Claude working on research problems & it would be easier for me to accept it as a serious scientific tool if it didn’t end every single response with the LLM equivalent of “you want fries with that?” — constantly trying to goad you into asking more

It’s bad enough that it tries to flatter you by telling you you’re asking such good and insightful questions. But desperately egging you on to try to make you spend more tokens is just gross. Like if every time you ran a mass spectrometer it tried to upsell you on some more advanced analysis.

< dozens of people tell me "you can just tell it to turn that behavior off!" > Yeah that's really not the point of my complaint. I'm not asking for suggestions on how to make the system more pleasant for me to use; I'm pointing out that it has dark patterns built in that make it harmful by default.