Reasoning Models Sometimes Output Illegible Chains of Thought
Published on November 24, 2025 6:24 PM GMTTL;DR: Models trained with outcome-based RL sometimes have…
Recent Articles
View All
The State of AI: Chatbot companions and the future of our privacy
Welcome back to The State of AI, a new collaboration between the Financial Times and…
What’s next for AlphaFold: A conversation with a Google DeepMind Nobel laureate
In 2017, fresh off a PhD on theoretical chemistry, John Jumper heard rumors that Google…
A new AI benchmark tests whether chatbots protect human wellbeing
Most AI benchmarks measure intelligence and instruction-following rather than psychological safety. Humane Bench evaluates models…
Former MrBeast content strategist is building an AI tool for creator ideation and analytics
Palo is tool for short video creators to gain insights into what is working for…
Gemini 3 Pro Is a Vast Intelligence With No Spine
Published on November 24, 2025 2:00 PM GMTIt’s A Great Model, Sir One might even…
Momentic raises $15M to automate software testing
AI testing startup Momentic has raised $15 million in a Series A round led by…
Formal confinment prototype
Published on November 24, 2025 12:57 PM GMTThis whitepaper was a preliminary approach to using…
Job for 2027: Senior Director of Million-Dollar Regexes
The following article originally appeared on Medium and is being republished here with the author’s…
On negotiated settlements vs conflict with misaligned AGI
Published on November 24, 2025 12:03 PM GMTcross posted from my SubstackIntroductionFollowing on from my…