I can go on, but imagine you're relying on this system to grade papers... Now any independent thought or argument is squashed and corrections in a bias manner are added. ChatGPT only knows what it's trained on, it doesn't have real-world examples or live-time examples incorporated.
The AI is not decerning and right in the announcement, OpenAI states it's intention on "correcting assumptions":
> challenge incorrect assumptions
I imagine some of these assumptions will be bias towards particularly ideologies / things people desire.
- https://twitter.com/Basedeyeballs/status/1613269931617050625
- https://medium.com/ninjas-take/chat-gpts-bias-is-very-very-e...
I can go on, but imagine you're relying on this system to grade papers... Now any independent thought or argument is squashed and corrections in a bias manner are added. ChatGPT only knows what it's trained on, it doesn't have real-world examples or live-time examples incorporated.