Somehow YouTube thought I needed to see this. Now I can see why. The release of ChatGPT o1 is not an incremental update to AI, it is a paradigm shift.

I'll include just one channel of an astrophysicist PhD playing around the recently released to the public version of o1. The model was correctly answering never before seen PhD graduate test questions over and over, including many tricky multi-part ones. These were never posted anywhere.

This PhD spent over a year coding abstract physics involving astronomical concepts into part of his thesis…the code was over 1,000 lines. Without seeing the code and only reading a description in an academic journal submission paper, o1 wrote the code with only 6 shots in under an hour. The code worked and did the physics and did it in around 200 lines.

This is happening all over the world and it hasn't even been a week.

Grasp the follow-on implications of billions of people (many, many PhDs and geniuses) leveraging this.

Sauce: https://www.youtube.com/@KMKPhysics3/videos