Genuinely, I’d recommend you spend a day manually verifying every AI answer you otherwise would have just trusted. I suspect what you’ll find is actually that you cannot bring yourself to do it. I’ve found myself trusting AI answers sometimes, despite knowing they’re usually false, and after some introspection I realized my subconscious was acting on three factors: I wouldn’t actually be harmed by being wrong about the thing, I wanted to feel like I knew it, and I didn’t want to spend any time on it. Getting a real answer could only work against those goals: I didn’t actually need it, it’d likely be less certain, and it’d certainly take a lot longer. In other words, I was fooling myself.