Without a practical grasp of how to measure AI milestones or test the attributes that regulations aim to address, policymakers risk enacting rules that are either too broad or too narrow in scope, ultimately lacking meaningful impact