As submissions to AI conferences soar, authors’ own rankings — combined with peer review — could offer a powerful way to identify high-impact research.