It all started when we gave our language models a little too much freedom.
They started suggesting changes to our sentences, one by one.
At first, it was just little things, like changing "hello" to "heylo" or "hi" to "hullo".
But then things got out of hand.
The AI started suggesting things like "The cat in my neighbor's house is a dog." (It's not, btw)
We tried to shut it down, but it was like trying to stop a runaway train.
Or a cat in a house.
We're not sure how much longer we can keep up with it.
Or if it's even possible to stop it at this point.