Among the other things AIs will get better at is recognizing AI-generated text. Presumably writing generated by this year's models will be easy to catch in a few years. Prediction: This will cause huge scandals in academia, as published papers are later exposed as AI-generated.