A study conducted at the University of Chicago found AI detection tools varied widely in quality, and proposes a framework for use both inside and outside of education.