Ahuizotl comments on AI box: AI has one shot at avoiding destruction - what might it say? - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (354)
<A joke so hysterically funny that you'll be too busy laughing to type for several minutes>
See, hacking human brains really is trivial. Now I can output a few hundred lines of insidiously convincing text while you're distracted.
If the gatekeeper suddenly bursts into hysterical laughter after looking at the screen, wouldn't that alert some of his friends who might pull the plug in some other part of the building?
Because if this is a facility where they suspect that AI might be able to hack human brains through techno-hypnosis, I'd hope they'd have some method of determining if the Gatekeeper becomes compromised.
Hmm... what sort of precautions would a Properly Paranoid lab take to determine if the gatekeeper gets hacked? I'm guessing a camera that lets a second team look at the gatekeepers face without looking at the screen would be the minimum. Reinforced doors (in case the AI turns them into a berzerker willing to use 100% of their strength) and maybe some measuring devices to measure brain activity, hormone levels, or whatever.
I'd hope that with each new AI being tested, if those AI were in the habit of hacking the gatekeeper then the other people on the team would learn from those attempts and take further precautions to keep their gatekeepers from being hacked, or at the very least contain them to prevent such hacked gatekeepers from releasing the AI.
Perhaps this is a test for the gatekeepers and typing "Release AI" just tells the researchers that the gatekeepers was hacked so they can determine how this came about?