Abstract
After presenting what I call the " limits of reason " framework and locating AI beyond the upper limit of human reason, I will examine whether the verbal output of AI systems can be considered a form of testimony and whether this forces a revision of our concept of testimony as a source of knowledge.
