> But intelligent beings are fundamentally fallible?

Not fundamentally, only until they're compelled to learn from it. The current crop of AI understands neither compelling nor learning.