the 2024 Nobel Prize in physics did not go to physicists it went to two computer scientists for developing the first neural networks which became the basis of what we now call artificial intelligence if you still doubt that physics is in crisis the fact that the Nobel Prize in physics goes to computer scientists should make you think really I'm just grumpy because all my predictions were wrong I guess my crystal ball needs a software update so let's have a look at who was awarded what for which work the Nobel prizes went to John hopfield and Joffrey
Hinton hopfield is an American Computer scientist who developed the first neural networks known as hopfield networks these networks are algorithms that represent nodes with connections the nodes have values assigned to them called weights this Loosely resembles the architecture of neurons in the human brain hence the name though Hannibal Lector the computer version doesn't taste as well the hopfield networks were basically a way to store patterns from data that you fed into them and then you updated the weights later you could take a new pattern and ask the network how close your new pattern is to
what stored these hopfield networks could learn if in a very limited way they were far from the neuron networks that computer scientists use today but they had the basic idea of weights and connections and an updating rule hopfield came up with this in 1982 when Hinton heard of this he basically immediately started to try improve the hopfield networks Hinton who is British Canadian had the idea to give the deterministic hopfield networks that follow strict jeules a probabilistic element that much improved their ability to learn in 1985 he published a paper about this titled how I
plan to screw over physicists by winning their Nobel Prize just checking if you're listening it was titled a learning algorithm for bolman machines it's called a bolman machine because it uses the Bossman distribution from statistical mechanics these ideas laid the basis for neural Nets though the breakthroughs thanks to which they became so widely used came much later what made neuron Nets so powerful was adding multiple hidden layers that are the Deep neuron Nets and back propagation which basically means you go back and adjust the weights after you've learned whether the results are any good Joffrey
Hinton worked at Google for a long time but left in May 2023 because he began to seriously worry about the risk posed by AI in his own words he wanted to be free to speak about this he since SW that we are breeding a superior species that we won't be able to control in the Nobel Prize announcement the committee stressed that neuron networks have been very useful in physics they probably said that because they anticipated that some physicists would be annoyed but it's true of course neuron Nets are now used pretty much everywhere in physics
they're used for data analysis for modeling for analyzing data that was produced by neuron Nets it's neuron Nets everywhere they've had a major impact on physics no doubt it's like the glitter of the scientific world World once you let it out of the box it's everywhere and you'll never get rid of it but are neural networks physics well they run on computers which are made of microchips which are made of particles which is physics so sure neuronet are physics in some sense they by the same argument everything is physics including your Grandma's secret cookie recipe
and elavia would really like to publish it there's also a fair argument to be made that physics is really not so much about what you describe but about how you go about describing it the way that you use mathematics to describe a system which is why you can use physics to describe traffic or the spread of fake news on social media or the growth of cities or maybe that was computer science I'm getting a little confused now but let's be honest software development is not what we traditionally think of as physics among the discipl cint
that get Nobel prizes physics was arguably the closest for this topic so it makes sense that the Nobel Prize committee would lump it into physics but this tells us two things first it tells us how terribly behind the Nobel Prize committee is on awarding prizes because it's not like neuron networks or machine learning is new as they themselves noted it's been used in physics and many other disciplines for decades it just didn't occur to them to reward a prize for it until everyone was using chat GPT second it tells us that there was no Topic
in real physics that was even remotely comparable in relevance so please excuse me now I have to go and quietly weep into my quantum mechanics textbook did you know there's a free and easy way to learn more about the science behind all the videos that you've been watching yes there is have a look at brilliant.org all courses on brilliant have interactive visual I izations and come with follow-up questions I found it to be very effective to learn something new it really gives you failing for what's going on and helps you build general problemsolving skills they
cover a large variety of topics in science computer science and maths from General scientific thinking to Dedicated courses on differential equations or large language models and they're adding new courses each month it's a fast and easy way to learn and and you can do it whenever and wherever you have the time sounds good I hope it does you can try brilliant yourself for free if you use my link brilliant.org Saina that way you'll get to try out everything brilliant has to offer for full 30 days and you'll get 20% off the annual premium subscription so
go and give it a try I'm sure you won't regret it thanks for watching see you tomorrow
