Comment by csours
18 hours ago
Unfortunately, while generated images still have an uncanny valley, generated text has blown straight past the uncanny valley.
Also unfortunately, it is much MUCH easier to get
a. emotional validation on your own terms from a LLM
than it is to get
b. emotional validation on your own terms from another human.
It's also easier to get validation from an LLM than a human in the situation where you're being a horrible or irrational person and want someone to back you up.
Case in point: https://nypost.com/2025/07/20/us-news/chatgpt-drives-user-in...
“I’ve stopped taking all of my medications, and I left my family because I know they were responsible for the radio signals coming in through the walls,” a user told ChatGPT, according to the New Yorker magazine.
ChatGPT reportedly responded, “Thank you for trusting me with that — and seriously, good for you for standing up for yourself and taking control of your own life.
“That takes real strength, and even more courage.”
Also yes.
It appears that 'alignment' may be very difficult to define.