The most interesting and challenging problem: align future all-outsmarting entities with conflicting human principles, using undefined rules for a technology we can’t grasp, ensuring flawless safety against unpredictable risks. The problem seems intractable in short timescales and the stakes couldn’t be higher.
Al Alignment, which we cannot define, will be solved by rules on which none of us agree, based on values that exist in conflict, for a future technology that we do not know how to build, which we could never fully understand, must be provably perfect to prevent unpredictable and untestable scenarios for failure, of a machine whose entire purpose is to outsmart all of us and think of all possibilities that we did not.
Once a Big Training Run is done, they test its behaviour to discover what new capabilities have emerged.
© 2025 Lethal Intelligence – Ai. All rights reserved.
Your email will not be shared with anyone and won’t be used for any reason besides notifying you when we have important updates or new content
