As someone who considers themselves in the "safetyism contingent", the only thing I find myself agreeing with about the rollout of LLMs has been the argument that it has to be shown to and used by the public while under development.

Every specific prediction not only sounds like science fiction before it arrives, it tautologically is science fiction because such predictions are narratives about something science-y that has yet to exist: https://www.youtube.com/shorts/wt-fLWxkrfs

Even when we accurately predict something in the future, we often have no clue what the implications are for society until it is in our hands: https://youtu.be/2Pw_7vAK9k8?si=X5t4tcxsXiuHEfBi

Go too fast, suddenly drop in tech with too many novel implications all at the same time, and we likely hit something relevant without having the means to cope with all the change. However, "too many" and "likely" are weasel words with no predictive power about what the danger threshold might be. I may be entirely wrong to think there is one, or we may have been bouncing off it since the industrial revolution, or I may just be echoing the same thing everyone feels once reaching their 40s: https://www.youtube.com/watch?v=fkUwXenBokU