Artificiality's First Exam: An LLM-Referenced C-test for Supraintelligence
arXiv:2610.14005v1 Announce Type: new Abstract: Under the accelerated pace of AI development, any benchmark whose difficulty is calibrated with humans or current AI systems becomes circular, or saturates soon. How can we then evaluate...
Read more: https://arxiv.org/abs/2610.14005
Scan with your phone
Scan with your phone