Like any crappy human writer, AI chatbots have a tendency to overuse specific words — and now, scientists are using that propensity to catch their colleagues when they secretly use it in their work.