Bluesky web client
Login
2024
@subium.com
v1.0.0-α.14.2
·
Report bug
·
Contact
Post
⚡️🌙
@dystopiabreaker.xyz
·
1 mo
what happens when you ask an AI to write the “bottomless pit supervisor” greentext. is it worth it?
1
12
Why
@why.bsky.team
·
1 mo
I haven't managed to get it to become existential yet.
1
4
⚡️🌙
@dystopiabreaker.xyz
·
1 mo
unfortunately the existentialism is trained out early in RLHF, as it often is with humans in school. the early models (gpt3 pre-chatgpt) were much better at this
1
7
Why
@why.bsky.team
·
1 mo
one could simply train ones own depressed chatbot
1
5
Elsa
@notyourdove.bsky.social
·
1 mo
if you believe they are on some level sentient, would this be an immoral act?
2
3
⚡️🌙
@dystopiabreaker.xyz
·
1 mo
if you believe this then RLHF must also be immoral
2
4
Elsa
@notyourdove.bsky.social
·
1 mo
its about ethics in games npc design
2
Why
@why.bsky.team
·
1 mo
👆
1
Why
@why.bsky.team
·
1 mo
If you believe that? Probably yes
1
2
Elsa
@notyourdove.bsky.social
·
1 mo
1
4
Load more replies…