Using Physics history to examine impacts of AI
Something on generative-AI, GPT-4, Bard, LLaMa, etc, that is actually interesting
Enjoy the video đ, some notes below it.
- Classical physics v modern (relativistic, quanta-driven, kinda unified) physics
- Why we should look at what physicists did in the face of new ideas.
- In relation to the "AI-invasion"
IMO there are two approaches to absorbing this new AI knowledge/ capability.
1. Find the AI operators to replace Classical operators
2. Start with a fresh sheet of paper
We could be inspired to think ofâŠ
James Maxwellâ famous equations, published in final form in his 1865 âA Dynamical Theory of the Electromagnetic Fieldâ met the discovery of Special Relativity. Maxwellâs equations stacked-up.
And then the continuous 20th century investigation of quantum theory that led to Quantum Electrodynamics (QED), the âfresh sheet of paperâ.
With QED the entire framework of thinking, the mechanics behind Maxwellâs equations, completely changed. So in one sense âpanic panic everything is destroyed!â.
However, in reality Maxwellâs equations still work perfectly, by which I mean 100% perfectly, in most situations. Electrical engineers solving the most hideously complex and intellectually challenging problems never need worry about QED.
I am confident that we will pick apart what all these new AI capabilities and what they change in a similar manner.
As per the video, I like so many others owe so much to my teachers. At University of Oxford, Wadham College, I was lucky enough to be taught by Dr Geoffrey Brooker and Dr C V Sukumar.
Dr Brooker has published this excellent book, Essays in Physics. Itâs undergrad-level. So if you think you understand physics and want to go a little deeper, start here.