Editor’s note

It's the holidays. Let's keep it short.

Please enjoy!

Unrelated-but-interesting

Brains come with a blueprint for understanding the world. To hell with Locke's tabula rasa! And British empiricism! We may not be born blank slates. There are spontaneous electrical patterns observed in brains which have never received sensory inputs, as they haven't yet developed the ability to. The stranger part is how the UC Santa Cruz scientists conducted their experiment: by growing brain organoids from human stem cells—or brains in a vat — ucsc.edu

Parakeets test the water when they first meet strangers. And the idea is that you should probably do that too. If no one's being weird, you can get closer and touch beaks, as it were. This is what scientists discovered when they studied how social relationships form among wild-caught monk parakeets. If you just moved to a new city, figure things out gradually and slowly hop your way to riskier interactions — nautil.us

AI & software

The AI correction could be a wildfire, not a bubble burst. Wildfires are natural functions of ecosystems: they clear out the dry brush (moat-less GPT wrappers with no proprietary data or competitive edge) and leave space for new, better adapted sprouts to grow. The thick trees (hyper-scaler incumbents such as Google and Nvidia) will benefit as they will shed their less valuable offerings. Compare with the 2000s dotcom crash. All the built infrastructure, like fiber optic cables, eventually benefitted players born later, like YouTube and Netflix. Today, it is the compute power to serve AI (inference) that humanity can't get enough of. The bottleneck remains energy — ceodinner.substack.com

Loose ends

A very short self-administered ADHD test. Merely respond to the questions and evaluate your score. Join the neurodivergence! — add.org

15,000 word essay arguing NVIDIA will soon be dead. It’s long and repetitive but the arguments are sound — wheresyoured.at

Super cool and informative interactive website that shows how transformers work. The transformer is the neural network architecture underpinning how contemporary large language models work. It's the T in ChatGPT — poloclub.github.io

How to work with GPT 5 Pro as a junior hire. In this long article, the author, who is a solo founder, explains how they use ChatGPT as their colleague. The author runs a music studio and the different use cases can teach anyone how to better use these models — soundformovement.com