People who swear at LLMs make me wonder. These systems have no feelings and no thoughts. Yet some of my colleagues report spending their energy on berating them. I have a few thoughts on this.
The practical case first: swearing at AI wastes tokens. It pollutes context, and it makes the model spend more effort to mollify you.
With the above out of the way, I still see more problems with this behaviour. Nobody judges you for swearing at a printer, but a printer does not answer in full sentences and say “sorry”. I propose that there is danger in getting used to swearing at something that is able to converse.
Fake it till you make it, right? If every day I get used to anger while chatting, I will acclimatise to that. I believe I would soon find myself getting angry while chatting with my colleagues.
I’m also reminded of a philosophy class I had as a kid, where I learned about the Ring of Gyges. It’s a magical artifact proposed by Plato, which grants the wearer the power to become invisible. The exercise was then to consider if you would still act justly or morally, even when no one can watch you.
Similarly, there’s the mom-wisdom of observing how one treats a waiter. One’s character is then revealed by how they treat someone who can give them no social or financial advantages.
The waiter could at least spit in your soup, but Claude cannot even do that. This waiter test nowadays has a subject with no leverage at all, and some of the smartest people I know appear to be failing it.
That said, smart people notice things eventually. So while I am not worried, I would still like to suggest my “fake it till you make it” approach: every exchange with the machine is a chance to practise.
Next time you find yourself filled with anger and frustration, pause. Try to at least remove the anger from your response. You may find yourself becoming calmer and kinder with the humans around you too.
Plato had to imagine the ring. We get to wear it every day.