The same familiar, reliable statistical metrics used on TAR projects can be calculated to validate generative AI results in document review–and those calculations are easier to do, earlier, than you might think.