Stephen Downes

This is quite an interesting article on how we understand AI, though it feels like it ends too soon. I recommend reading it from the perspective of human learning. What I mean is, imagine the interviewee, Naomi Saphra, is talking about humans, not AI. Most of what she says still makes sense. For example, "Just as biologists must understand an organism's evolutionary history to fully understand the organism, she argues, interpretability researchers should pay more attention to what happens during training." And also, "The model already wants to learn the easy thing. Your job is to keep it from learning the easy thing right away, so that it doesn't just start memorizing exceptions. That might make it hard to generalize to new inputs in the future."