Several key happenings in the space:
Ethan Mollick provides an update on a study run at Harvard, Stanford, and others that looked at patient assessment by physicians vs AI. They've updated their study to use o1 Preview, and it's beginning to pull away. Also worth noting that studies move quite slow, and we already have access to both o3 and what-will-become o4. (note: there was no o2 thanks to the European telecom)
https://x.com/emollick/status/1925362565946786206
It also appears that AI+instructor is giving children 1.5-2 years of advancement in only a few months, however students who were unfamiliar with LLMs are more prone to use it as a crutch, damaging their learning.
https://x.com/emollick/status/1925055450254385592
In what might be the coolest thing I've seen in some time, this group used AI to create an environmentally friendly coolant.
https://x.com/vitrupo/status/1924568771353841999