The issue is that @lessig is very very new to AI and doesn't understand it at all. He understands many things extremely well and has often had extremely good policy recommendations about those things. But on AI he's not competent, but for some reason has strong opinions anyway.