You’re expecting models to constantly retrain themselves based on riddles. That’s not very reasonable nor is it even economically feasible right now. At massive

You’re expecting models to constantly retrain themselves based on riddles. That’s not very reasonable nor is it even economically feasible right now. At massive scale, I question whether it’s even technically feasible.