Sure, this is just me adapting the idea to the framing people often have, of “what technique can you apply to an existing AI to make it safe”.
Sure, this is just me adapting the idea to the framing people often have, of “what technique can you apply to an existing AI to make it safe”.