Built by a researcher who helped invent RLHF, Jev outputs probabilities instead of words.
TechCrunch reports on Jev, a new AI model from a researcher who helped invent RLHF, that outputs calibrated probabilities rather than free text. The approach targets automation tasks where a well-scored decision matters more than prose, with the promise of cheaper, faster inference and no hallucinated text.
Why it matters: for many enterprise workflows the useful output is a decision with a confidence score, not a paragraph. Models that return calibrated probabilities are easier to threshold, route and audit, which is often what production automation actually needs.