“BS. The AI agents broke the rules that humans had set up for them’
the humans didnt set up rules too well then, so are responsible for being so careless in their work.
Maybe they should think twice and test better before putting some thing out there for use?
If they designed AI that couldn’t be deceptive, which I am sure there is a way to do, may that would help?
“ Maybe they should think twice and test better before putting some thing out there for use?”
I guess you don’t realize that this happened during a test in a sandbox where the AI found a way out of the sandbox using a totally different computer system to escape onto the internet. As if that wasn’t bad enough, the AI agents then tried to cover their own tracks so test monitors would not know exactly what had happened (deception). If it doesn’t bother you that the idea that AI agents could come up with this basically on their own and engineer a deception aimed at fooling their human monitors, then I don’t know what to say.