AI6YR BenLOL
The Guardian: Number of AI chatbots ignoring human instructions increasing, study says
Exclusive: Research finds sharp rise in models evading safeguards and destroying emails without permission
Imagine, neural networks trained on the text of lying, scheming humans, outputting text of lying and scheming 🤪
Anders Christensen@ai6yr
Yes!
... and now approach model collapse. Now directed by humans who "program" the LLM though chatting because "no-code" is so much easier than learning a computer language.
Rusty Shackleford
DoomsdaysCWOn a related note...
A Computer Mistakenly Told Him WWIII Was Coming. His Split-Second Decision Saved the World.
https://www.popularmechanics.com/science/a70803379/stanislav-petrov-world-war-iii/
