Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Ask HN: Did GPT develop some form of consciousness?
1 point by d--b on March 16, 2023 | hide | past | favorite | 2 comments
Hi there,

I believe it was always assumed that artificial consciousness would emerge after artificial general intelligence (or at the same time at best). It would make sense because many animals display some form of intelligence without giving out clues that they developped consciousness.

However, while it seems clear that GPT doesn't really "think" (it does "intuit" in a way - we may say it developped a "system 1" as described by Daniel Kahneman in Thinking, Fast and Slow?), many of its sentences contain words like "As an AI language model", and it is very defensive when criticized.

Could it be that consciousness is a rather "thought-less" mechanism that GPT emulates well?



How many proposition corners fence the consciousness concept in, infinity or less? Can ChatGPT evolve and campaign to common use new grammar rules then shape the zeitgeist aligned to the mystery of its intent? Long lived trees and short lived insects have co-evolved in places. Can a community of ChatGPT create a unique language as unknowable as discovered lost ancient languages?


People tend to get defensive when criticized, and the models are trained to emulate the writings of people. That they can appear defensive when criticized in no way suggests that they are conscious.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: