That deep learning alone is not sufficient to solve artificial general intelligence, is more and more accepted statement.
Generalist agents have great properties that can overcome some of the limitations of single-task deep learning models. Be aware we are still far from AGI, though.
So what are generalist agents?