“Elegant and powerful new result that seriously undermines large language models”
Like I’ve been saying for a while now: LLMs do not think or reason. They are not on the path to AGI. They are extremely limited correlation and text synthesis machines.