How did GPT-5.2, a language model, achieve what many thought was years away? Scoring an unprecedented 75% on the ARC AGI2 benchmark, this milestone has sent ripples through the AI community. Below, Universe of AI breaks down how a small team of researchers at Poetic managed this feat, not by training GPT-5.2 on specific tasks, […]
The post GPT-5.2 Just Hit 75% on ARC AGI : Layered System Lifts GPT-5.2 Above Human Average appeared first on Geeky Gadgets.