12Two-faced AI language models learn to hide deception - ‘Sleeper agents’ seem benign during testing but behave differently once deployed. And methods to stop them aren’t working.(www.nature.com)posted 9 months ago byLugh@futurology.todayMin futurology@futurology.todayView comments9 commentssavehidereportSort:HotTopControversialNewOldYou are viewing a single thread.View all comments [ +- ]mateomaui@reddthat.com1 point9 months agoJust… don’t hook it up to the defense grid. permalinksavereportreply[ +- ]Possibly linux@lemmy.zip1 point9 months agoSorry, to late for that permalinksavereportparentreply[ +- ]mateomaui@reddthat.com2 points9 months agoAlright, I’ll be out back digging the bomb shelter. permalinksavereportparentreply[ +- ]Possibly linux@lemmy.zip1 point9 months ago*Its too late for that honestly permalinksavereportparentreply[ +- ]mateomaui@reddthat.com2 points9 months agoAlright, I’ll switch to digging holes for the family burial ground. permalinksavereportparentreplyShow more comments