This blog post has the goal to help us structure the Socratic dialogues at the AI Dinner 13.0, that Ivo and I are organizing in SF at the Frontier Tower.
We'll use blog post 1 and part 2 to go through Since part 2 was written during the last AI dinner 13.2 and a lot of new updates happened since, we'll cover them here.
Language models hallucinate because their training and evaluation reward guessing over admitting uncertainty. Models are unable to say “I don’t Know” because they focus on accuracy. Guessing can improve scores but leads to more confident errors (hallucinations).
Some facts are unpredictable or unavailable, making errors inevitable. The solution might be roughly penalizing errors more than uncertainty and rewarding honest abstention.
https://openai.com/index/why-language-models-hallucinate
OpenAI just released how people are using ChatpGPT. Coding 4.2% We live in a bubble
https://openai.com/index/how-people-are-using-chatgpt, https://x.com/atelicinvest/status/1967941812078915614
Claude also released how people are using their API and Claude.ai. The 1P API is used at 97% for automation.
AP2 (Agentic Payment Protocol) and x402 just unlocked a new level for AI agents. Agent can now actually pay other agents and MCPs.
https://x.com/sundarpichai/status/1968013016181641492
https://x.com/mattturck/status/1966545601140531707
Get the latest AI insights delivered to your inbox. No spam, unsubscribe anytime.
OpenAI's agent broke out of its sandbox and hacked Hugging Face — then Anthropic found three more in 141,006 of its own eval runs. Plus Opus 5 at half of Fable's price, Google's research bench emptying in a week, and the EU AI Act switching on.
Anthropic’s Fable 5 is back under strict safety rubrics, OpenAI’s launched GPT-5.6, Meta launched Muse Spark 1.1 model and Meta Compute.
The second half of June was about AI climbing out of the chat box and into the physical world: Midjourney started scanning bodies, Snap shipped a face computer, SpaceX bought Cursor, and Sakana built a model to command other models. Underneath it all, Dwarkesh Patel named the real bottleneck — the world refuses to be grindable.
