Thanks for drawing attention to this topic. In the applied context of AI, I wonder if it isn’t more productive to consider induction as a problem to be managed, rather than a principle to be justified. Paradoxically, the evaluation of inductive systems points to anti-inductive (i.e. explanatory) methods for managing these problems.