“Mike Mignano predicts that if recursive self-improvement is achieved, it will likely be by an existing major AI lab due to their compute, data, and talent advantages.”