Simply search user prompts for curse words and then measure hostility sentiment. User hostility rises as agents fail to meet expectations.

preuceian 3 hours ago | parent | next [-]

Maybe im overlooking something obvious but how do you 'simply' scan the content of Claude users their prompts?

	▲	gordonhart 2 hours ago \| parent [-]
		GP was making a joke, but Anthropic could implement this if they wanted to. Not a bad metric actually if you can measure it cheaply enough.

mrbananagrabber 4 hours ago | parent | prev | next [-]

I uh might be skewing that as I generally just use a lot of curse words with Claude by default

Trufa 4 hours ago | parent | prev | next [-]

I'm glad I'm not the only one.

	▲	sejje 3 hours ago \| parent [-]
		One time I cussed Claude out so hard that it actually quit his doom-loop and fixed the thing. It's the only time cussing worked, though.

ctxc 4 hours ago | parent | prev | next [-]

I feel bad about it but sometimes it's so daft, I can't even xD

It's not my fault, they set high standards!

smotched 4 hours ago | parent | prev | next [-]

there are many times where I just do it myself and it thinks it did well.

mhl47 2 hours ago | parent | prev [-]

Or there are global events that stress people out .. or their expectations change over time. Not that simple ;)