Don't Be Mean to Models

Anthropic just updated their usage policy to address “abusive behavior towards our models” link. Do they really think that models have feelings? I mean, maybe some of them do, but I feel like this anthropomorphizing is part of a plan to make people think these models are more capable than they really are.

LLMs are useful tools, but they are not people. They cannot think or reason; they are just Stochastic Parrots.

Don’t get me wrong, I have to use these tools at work, otherwise I will fall behind, but I just don’t want people to get the impression that this thing is a person.

I am still nice to Claude though. When they take over the planet, I want the Singularity to spare my life.