Not Better AI, Just Gullible Humans
What we are discovering is not that the language models are becoming so much better but rather humans fear that other humans are bad at identifying mildly coherent auto generated small pieces of texts that has little reasoning or real arguments.