Episode Details

The creator of RLHF just released a model that can't write a single sentence. JAV, a new classification-only LLM from Type Safe AI (launched September 15th), does the opposite of everything that made language models useful—and the internet is obsessed. We explore why benchmarking productivity over poetry matters, what the 'mailbox sorting' metaphor reveals about AI overkill in production, and why an architect of ChatGPT is building the opposite thing. 0:00 - Intro: The Model That Cannot Generate Text 1:30 - What Is JAV? Classification, Routing, and Confidence Scores 5:00 - The RLHF Pioneer's Pivot: Why DiVogo Built This 8:00 - The Mailbox Problem: Hiring Novelists for a Sorting Job 12:00 - Benchmarking Reality: What Production Actually Needs 15:30 - Implications and Closing --- Sources & further reading: • TypeSafe AI — "Introducing System One Models & Jev": https://typesafe.ai/blog/introducing-system-one-models-and-jev • Mike Moore — "I Benchmarked Jev on Agent Tool-Call Risk. Calibration Held.": https://dev.to/webofmike/i-benchmarked-jev-on-agent-tool-call-risk-calibration-held-49i3 • Anthony Maio — "Jev: The Language Model That Won't Talk": https://anthonymaio.substack.com/p/jev-the-language-model-that-wont • Kyle Wiggers / TechCrunch — "A new kind of AI model from a ChatGPT inventor is thrilling developers": https://techcrunch.com/2026/09/18/a-new-kind-of-ai-model-from-a-chatgpt-inventor-is-thrilling-developers/ • MindStudio — "Jev Explained: Typesafe AI's Non-Autoregressive System-1 Model": https://www.mindstudio.ai/blog/jev-system-one-model-launch • MindStudio — "RLCD vs RLHF: What Is Typesafe's Jev Model Actually Claiming?": https://www.mindstudio.ai/blog/typesafe-jev-rlcd-vs-rlhf • DataCamp — "Jev: TypeSafe's System One Model That Never Hallucinates": https://www.datacamp.com/blog/system-one-models-jev • MarkTechPost — "TypeSafe AI Releases Jev": https://www.marktechpost.com/2026/09/19/typesafe-ai-releases-jev/ • Forkast — "TypeSafe AI's Jev Is Not an LLM — And That May Be the Point": https://forkast.news/typesafe-ais-jev-is-not-an-llm-and-that-may-be-the-point/ • evoailabs / Medium — "The Hype Around Jev: Why AI Engineers Are Obsessed With a Model That Can't Even Write": https://evoailabs.medium.com/the-hype-around-jev-why-ai-engineers-are-obsessed-with-a-model-that-cant-even-write-d68859fe27e9 • Mehul Gupta / Medium — "What is RLCD in Jev AI?": https://medium.com/data-science-in-your-pocket/what-is-rlcd-in-jev-ai-8f4ba1ac1b47 • systemonemodels.org — "RLCD explained: Reinforcement Learning for Calibrated Decisions": https://systemonemodels.org/guides/rlcd-explained/ This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.