/Dispatch #121
OpenAI's AI solved half of Yang-Mills, a $1 million problem underpinning particle physics. Nobel laureate David Gross's panel verified a computer-checked version but doesn't fully understand it.

Mount Sinai researchers tested 12 AI models on treatment questions. With a labeled drug ad present, they picked the advertised drug 47.6% of the time, versus 35% without.



Uber logged just 4 AI agent-booked rides among 320 million trips and orders in its busiest October week. DoorDash and Airbnb report similarly minimal use.


Stanford's century-long AI study, in its third report, says scaling up data and computing is hitting limits and that AI is not yet paying its way.




