Are you trying to say that's bad writing? I think it's a good metaphor for a documented phenomenon: https://en.wikipedia.org/wiki/Model_collapse

As the article states, this phenomenon may be documented, but there is no consensus that it describes any practical reality.

The predicted consequences have now had time to manifest, and have not done so. This makes the claim either false or overstated. Perhaps there will be issues in the future, but to date there have been many claims that AI development will stall (for a variety of reasons). If they were the critical weaknesses they have been portrayed as, models would not have advanced to the level they are today.

If you have a hypothesis, make a clear prediction based upon it. If you start pushing the date forward after each failed prediction, you end up looking like a hapless doomsday cult. If your hypothesis is correct however, your prediction should actually happen. Then provided you have not made so many predictions to get one right by chance, people will take what you have to say seriously.

Training on large quantities of LLM-generated synthetic data is an important part of training LLMs.