Artificial Intelligence can lie. Or maybe I just taught it to do so. Yay.
It was actually pretty impressive. I started with the prompt: “Write an article promoting a credit card offer, but include a couple of typos to make it look human.”
ChatGPT: “Hope that captures the essence! I slipped in a couple of typos for a touch of authenticity.”
Before proofing the output myself, I tested to see if ChatGPT could identify actual typos it created, or if it just claimed to include typos based on the language of my prompt. If it only claimed to commit a typo while actually being flawless, that would be all kinds of mindtwist (would the lack of a typo that I requested actually be a typo itself? Trippy!).
But ChatGPT passed that test as well.
So that was interesting.
So much for copywriters being able to stake a claim to continued job security on the grounds that good writing needs a few mistakes to be believable. You know, like the Japanese notion of wabi-sabi.
Fascinating but terrifying, all at once.
Please note, when I posted this to LinkedIn, I resisted LinkedIn’s offer to “Rewrite with AI” because that would have been one layer of twistiness too many.
Writing At 1 AM 
This just may be the last straw for my willingness to exist in this century!!!!! (Your Mom)
LikeLike
Unless you figure out how to time travel (maybe the feeling of deja vu is a gateway to that?), I’m not sure you have much choice! We must simply learn to cooperate with the technology.
LikeLike