“I” is first person.
“You” is second person.
“She” “They” “It” are all third person.
So far, so good.
But how can ChatGPT use the word “I”? And when we talk about ChatGPT, is it “he” or “she” or “they” or “it”? Because anything that is an “it” shouldn’t be able to say “I”.
We probably need a form of “it” that can be used by ChatGPT when it is talking about itself or on its behalf. Because “I” brings emotional and intellectual weight that confuses or deceives us.
As Kevin points out, the regulation to lead to this fix is really simple and easy to implement. When I say “we” I think we know what I mean. But when ChatGPT or other LLMs say “I”, what is being communicated here?
When we built the bot for this blog, I insisted that the bot not say “I.” Because it’s not me. It’s a bot.
Inventing new rules for how language works is fraught and regularly fails. But it’s only been a few months, and it sure seems like we’re getting comfortable with not distinguishing between text from a person and text from “it.”
It might be as simple as IT, with the second t being capitalized. Or ix, which is fun to say and will help my Scrabble game…
Just because the computer says “I” doesn’t mean that we’re not interacting with a computer. The uncanny valley is real and perilous.