AI RESEARCH
The Hot Mess of AI: How Does Misalignment Scale With Model Intelligence and Task Complexity?
arXiv CS.AI
•
ArXi:2601.23045v2 Announce Type: replace As AI becomes capable, we entrust it with general and consequential tasks. The risks from failure grow severe with increasing task scope. It is. therefore.