OpenAI’s o3 Scores an “A” on ARC’s AGI Test
o3 Models are going to get a lot more expensive, here’s why.
Michael Spencer,Dec 27
Apparently OpenAI’s o3 Model scores 87.5% on the ARC challenge (arcprize.org) – the key thing about this benchmark is that it is impossible to pre-learn, as every test has new conditions, models were stuck at 30-55%. Humans are particularly good at and LLMs were bad at it.

Adam Brown – How Future Civilizations Could Change The Laws of Physics
Dwarkesh Patel
We discuss: destroying the light cone with vacuum decay, holographic principle, mining black holes, & what it would take to train LLMs that can make Einstein level conceptual breakthroughs.

How AI Understands Us: The Secret World of Embeddings
Making Language Click with Numbers
DiamantAI, Dec 27
Embeddings convert language into numbers that preserve meaning. So, how do we create this map?

LLMs Fight With Both Hands Tied Behind Their Back
We haven’t yet given them access to knowledge-in-the-world
Steve Newman, Dec 27

Measure Up
An Ode to an Instrument
Dean W. Ball

Note to Our Energy Sucking Overlords
Datacenters and the AI Infrastructure race impacts on Energy. Are you ready for higher energy prices?
Michael Spencer
Dec 25