The 90 percent discount Claude — why model reliability has to be argued in economics, not mathematics
