Did AI Prove Our Proton Model WRONG?

[@pbsspacetime]

From the Department of Corrections: we accidentally IDed the wrong Stanley Brodsky at 12:33. To learn more about the correct Stanley Brodsky please go to: www.aip.org/history-programs/niels-bohr-library/oral-histories/46900

[@CosmasZachos]

Matt also mispronounced George Zweig's name, huge.

[@Ravenx_44]

all is forgiven! thank you for the self-governing!

[@ireissistable]

People get cancelled for less these days

[@solaris867]

1:50 Why are you doing this all the time it's so annoying

[@mudfossiluniversity]

Light is a dipole and they add together and are stable at certain Qtys....Protons....1823 dipoles make up 1 Proton.... 1824 is a neutron. We did light acceleration and used CMOS to view the interactions....Go to Mudfossil University on KZbin and see what light is and it makes up matter.

[@pbsspacetime]

Big thanks to the early gang! Because as noted a few episodes ago: Since our comment response livestream, we've noticed that YT isn't sharing our videos as much with our subscribers. So we're asking our subscribers to 1. switch their subscriptions from "PERSONAL" to "ALL" (just click on the subscribe button and you'll see it) and 2. Watch new episodes as soon as they can!

[@LavaCreeperPeople]

Did AI Prove Our Proton Model WRONG?

[@drstone3418]

Dark matter has expected affects of wormholes linking areas or gravity .

[@johnnydoe3603]

KZbin pushes Fake Videos Over actual Science Videos as Usual. 😂

[@InitialGSB]

Immediately appeared for me, looks like it's getting better.

[@watchoutforcopyright9339]

each of my day is not without seeing things about artificial intelligence now

[@noxfelis5333]

Thanks AI for telling us that we all contain intrinsic charm.

[@chronosschiron]

until they find out the method was flawed and show you that this way is not the way to know insides properly

[@AXharoth]

i dotn get it , did AI did it rly? did he say that in the video?

[@eljuanman999]

​@@AXharothit was a pun

[@chronosschiron]

@@AXharoth ask ai to write a computer program for you it cant this is why this method also is flawed the "AI" is only as good as humans make it and he never tells you what the AI is and ill say the question i left to begin with is quite valid

[@TerryProthero]

I don't think anyone has ever accused me of that before.

[@im_piano]

What's great about these particle physics experiments is that we're unlikely to run out of protons to disassemble in the near future.

[@strenter]

Actially, it is said that CERN stopped working one day. They looked for the reason and found they were out of hydrogen - the bottle being empty, their source for protons.

[@adin2259]

lmao

[@levybenathome]

Just in case, I propose we start working on ways to put them back together. Duct tape?

[@im_piano]

@@levybenathome Quantum. Duh...

[@clarkeeeee]

It's all fun and games until they come for your protons.

[@philmccavity]

This is so well explained and yet so packed full of great details. It's overwhelming in a positive sense. This series deserves every educational award out there.

[@ibisiki]

👍🏾👍🏾👍🏾 indeed

[@mactorresmo]

I´m a theoretical particle physicist and I really appreciate the precise way (and not boring at all) you bring the subject! It is rear to see a Physics Professor that brings information in such accessible way!!

[@TheVanillatech]

It is rare of a theroretical particle physicist to spell "rare" incorrectly....

[@bushwalker6214]

@@TheVanillatech He means seeing rear of the professor brings information in such accessible way!!

[@ncedwards1234]

@@TheVanillatech Rarities are similar to novelties, and they make me pay attention to life for a bit so yee haw.

[@notahotshot]

​@@TheVanillatechI'm fairly certain that he means he's theoretically a particle, and was calling the presenter a physicist, as in "I'm a theoretical particle, Physicist."

[@TheVanillatech]

@@notahotshot In the land of the blind...

[@pbsspacetime]

Update: The audio problem seems to be with the KZbin's processing of the video. Thank you for bringing it to our attention as we can now discuss the matter directly with KZbin. We will pay special attention to the audio in the coming episodes and do all that we can to deliver you high quality experiences while we work to find a resolution to the problem. Hey Space Timers! There seems to be an audio issue for some of our audience members. There may have been a processing error as it's not occurring for all of our audience nor does it seem to be in the original uploaded file. We're going to keep investigating and see what we can do to fix this. In the meantime, we hope you enjoy the content of the episode despite any technical issues you may be experiencing. Thank you for your support!

[@halvardsutterud4158]

Was starting to think the audio was made using AI!

[@sadderwhiskeymann]

No audio issues for me in this video. As for some of previous ones though, i experienced the audio not being comprehensible to my brain 😢

[@BDWANNEMACHER]

I thought it was to show the AI nature of the episode

[@Hermes_Agoraeus]

Also, would you please stop shouting CAPS in your titles? I miss the old non-clickbait titles.

[@GreeceUranusPutin]

Audio problems might be with the download from KZbin and not in the file.

[@parkpatt]

This helped me understand particle collision experiments better than I ever have before. Well done! Very clear and engaging presentation

[@seekter-kafa]

'better than before' still doesnt mean that you understand it

[@parkpatt]

@@seekter-kafa yes, that is how English works. Nice job!

[@notahotshot]

@@parkpatt I love how they thought they had a "gotcha" moment.

[@samerkadih8534]

⁠I think what @@seekter-kafa is trying to say, is that if you feel this was a clear and comprehensible presentation, then you probably know very little about what is being presented 😢

[@MultiSciGeek]

Same. I really appreciate the background explanations.

[@anterovaarnamo]

Thanks for explaining how machine learning is used in particle physics. This whole series is a rare gem in KZbin.

[@Desertphile]

Well, I dunno much about protons, but *Bing* told me The Big Bang was an explosion, and when I told Bing it was actually an expansion, Bing told me to change the subject. I still have the feeling that I hurt its feelings.

[@cesarkopp2]

Bing is a cun*. Call it wrong and dumb and it gets artificially annoyed and cut the conversation. GPT-4 always will say "I'm sorry!" I think MS put that intentionally so Bing don't spend resources discussing useless things... and for the memes. :D

[@theunluckycharm9637]

Hmm

[@maythesciencebewithyou]

Just checked and it gave me a correct answer.

[@personzorz]

Bing is a joke

[@thebaryonacousticoscillati5679]

I "broke ChatGPT3 when I asked it to explain quantum mechanics in the style of Bertrand Russell. Because I'd been asking it to do so in the style of iambic pentameter before it did so in verse. When I asked it to redo it in his prose style it went red and wouldn't respond . I had to reboot. I suspect in the mythos of developing AI I will be some sort of primordial demon that needs to be slain... :)

[@dipanjanghosal1662]

Thank you for taking the time to make these topics accessible and understandable for the general audience

[@marcinkrzeszowiec1538]

Great episode! Congrats to the whole team :) It all came together in a beautiful synergy. You are doing amazing work popularizing very difficult and cutting edge science! Gives one a whole new appreciation about the world, and the physics behind it :)

[@jgamb914]

Love this channel. I'm not up to speed on all the physics but love learning new things. I usually walk away with new knowledge and a better understanding of the subject matter. Thanks Matt. Great job as always..

[@blurta2011]

Sounds like you would believe any rubbish these people tell you.

[@sarpsomer]

This is one of the hardest topics to visualize, yet your team managed to do it well!

[@Matts_Ancient_Coins]

@@ephemera2 you have far too much time on your hands 🤡

[@jht3fougifh393]

​@@ephemera2 Damn son.

[@ephemera2]

@@Matts_Ancient_Coins I most certainly do

[@epicmetod]

started to believe that these guys are alien

[@highdefinist9697]

Yeah, the visualization was surprisingly good. Nice effects, but chosen so that they don't distract from the thing they want to show; visualizing more or less exactly as much as needed (so no superflous details, but also not omitting anything important); and also aesthetically well done.

[@DeltaVTX]

I am made of hopes and dreams. This is not an Undertale reference

[@mahadahmedbaloch]

You are a virtual particle

[@theonebman7581]

You still have hopes and dreams? In 2023? *We need to fix this asap, people*

[@jaredf6205]

Two dream quarks and one hope quark.

[@thepatriarchy819]

A eternal soul

[@onslaught147]

I'm made of bullshit.

[@crystalfire5564]

I feel like science is developing faster than my old brain can handle. But I am happy that we are making progress and content to get the pieces I can understand.

[@DeltaVTX]

Welcome to the singularity, my friend.

[@snakex555]

Hi friend, I am not too old, I can't keep up, theres gluons and unioms and wjfdkkd

[@rebjorn79]

@@snakex555 Don't forget the ueaoeobvutf and the uabweoaeu

[@watchoutforcopyright9339]

@@DeltaVTX the singularity is definitely close but i don’t think it has happened yet

[@DeltaVTX]

@@watchoutforcopyright9339 we are approaching the asymptote

[@Ahop63]

Incredible job of making this complicated topic very approachable and understandable by those of us that are not particle physicists.

[@alexpetrovich85]

From an Analytics perspective, this is amazing how quickly it can sort through these data sets and verify things now epistemically.

[@YoghurtKiss]

Yeah, there is no bias, there is no agenda, there is nothing but raw data. I love AI. I don't understand the whole doomsday hype about it.

[@israelmontefusco6300]

@@YoghurtKiss even with raw data, there would be sme kind of agenda or bias, since the data will be interpreted

[@cjheaford]

@@YoghurtKiss I understand and appreciate your comment. I suppose that maybe you just articulated the “doomsday” fear better than I could without you yourself realizing. In your own words: No bias. No Agenda. Only raw data. Bias and Agenda are part of being human - for better or for worse. The real fear is that A.I. will always, constantly, without fail, shall evermore produce the most logical mathematically precise and most efficient solutions for every query regardless of human wants. Had A.I. been available to our prehistoric ancestors, I believe humans would have been rightfully eliminated from the efficient equations long ago. Supreme Intelligence without bias and agenda is the opposite of humanity. We are human because we overcome in SPITE of our biases.

[@pacotaco1246]

​@@YoghurtKisspeople are afraid of what greedy hierarchs will do and have done with AI

[@DrWhom]

@@YoghurtKiss _What_ was measure and _how_ already constitutes a bias of sorts. You are being naive.

[@Ignirium]

I've also noticed my intrinsic charm vanishes almost as instantly as it arises, whenever i speak.

[@thehellyousay]

i will refrain from replying with the obvious obnoxious joke regarding your intrinsic charm, in deference to the god of ego-busting, the late great Don Rickles ... oh, okay, and because i couldn't think of one that was as sharp, clever, and erudite, yet humiliating and humourous in the fine tradition of prickle comedy that master Rickles epitomised. i'm getting old, i shall sit in the penalty box for 2 minutes and feel shame.

[@Herman47]

You definitely need some more gluon, to keep everything together.

[@N3ur0m4nc3r]

This sounds like a uniquely appropriate job for *quantum computing* 🙃

[@zinzhao8231]

for quantum computers you mean

[@shatterscape]

Who is quantum computing and what does he compute

[@fugitive6549]

Microsoft has already announced a breakthrough with quantum computing which stabilises the Qubits making them less prone to errors

[@Vysair]

@@fugitive6549 our manufacturing capability still has a lot to catch up

[@gvanish6000]

@@fugitive6549 i think intel is trying to make it commercial for institution to use

[@mitalichordiya1421]

I do not think intrinsic charm quarks can exist but It is possible that even at a low energy collision, when the proton is destroyed the energy which was keeping the proton together might have been released might have made a charm quark.

[@Pepesmall]

They could if we are measuring mass wrong and there is actually a negative and positive charge to mass, that could be effected by something like the time direction.

[@rand49er]

An absolute great use and application for AI! I recall when I was in college (engineering) hearing about the discovery of a thing called a "quark" and the buzz it created. We've come a long way with still more to go. Great video. Thanks.

[@SedatKPunkt]

Very well explained…and very thoughtful concerning the limitations of a layman. When looking at machine learning, QM, thermodynamics etc. then *statistics* which sounds boring is one of the most important and even most exciting tools

[@NobleSainted]

Incredibly interesting! Though most of the information went over my head, I found myself understanding more in this one episode than I have in my entire life of reading about electrons, protons, and subatomic particles. Thank you so much and please keep up the amazing work.

[@muzikizfun]

Don't dispare, sometimes too much information can give you brain freeze. Watch it several times, get the big picture thinking, then slowly absorb the details a little bit at a time . Good luck!

[@gs4945]

I always love how you make some of these tough topics easier to understand.

[@tordox1607]

“Don’t worry, there are plenty of quarks in the sea” is my new favorite line.

[@benruniko]

This is exactly the kind of use of AI I want to see more of. I bet we will learn some amazing things with models that have no bias to the basic assumptions of physics we all accept as true. Edit: yes all models are bias, this isn’t a solution to find better truths. It shows a multitude of possible solutions to puzzles without throwing some away simply due to preconceptions. This doesn’t research for us, it just gives us a new perspective on the data we have. The research is still up to humans to do, as it should be.

[@thetalantonx]

I agree, the only problem is the black box of the neural net. We know that it does a good job, we can't pop the hood and see *why* it does a good job. So I'd be happy if on things like this that could steer the course of entire fields of study that they have several different AI that all do as well on training data that we then can use to check each other.

[@joshuacadebarber8992]

​@@thetalantonxactually this is partially solved with memory modules and decision graphs

[@thetalantonx]

@@joshuacadebarber8992 Thanks for the reply! Do you have any resources you could point me towards?

[@joshuacadebarber8992]

@@thetalantonx sure, Transparent XAI is a very comprehensive field for this, the study called Generative Agents: Interactive Simulacra of Human Behavior has a section on the memory stream which goes in conjunction with other adjacent transparent approaches to logging the unknowns as well

[@ShadeAKAhayate]

@@thetalantonx Even more important, we can't (without re-checking) tell WHEN it does a good job and when it just imitates it perfectly with a totally straight face. It surely loves to do that as much as anything else. Over-reliance on tools like these can lead to dangerous results if precautions are not taken.

[@alien9279]

Using ai for science like this just has so much potential and I'm here for it. Even if we only get an ai like 10% as smart as a human and make an army of them for 24/7 science it would change the world

[@KonradTheWizzard]

While I cautiously agree with your first sentence, AI is not comparable to human intelligence. In fact the "I" in "AI" is a misnomer - it is not intelligent, it just uses algorithms that are inspired by nature. Specifically the way the brain approaches problems or how it is imagined to do so. Artificial Neural Networks in particular are a (rather crude) model of how simple clusters of nerve cells communicate. In mathematical terms they are a complex polynomial approximator that can be tuned with input data and subsequently be used to predict results that approach something that has a high likelihood of being correct when compared with the input data. (You may notice how cagey I am here: that's because we don't know exactly how they work in detail.) In short: please be careful with phrases like "AI will change the world" - if we are not careful, it might just do that - for the worse. If we are careful about it, it will merely make our jobs easier and WE will change the world, hopefully for the better.

[@Trickey2413]

Ai intelligence is very similar to human intelligence

[@unimportantnobody8364]

@@KonradTheWizzardI don’t mean to be pedantic but the idea that if AI isn’t used carefully it could make things ‘worse’ is quite a selfish and human centric view point, it would (maybe) only be worse for humans (if sci fi fears are to be believed). Personally speaking though, if the movies are to be believed and some how AI does decide that humans are surplus to requirements, that would only be the case because it would see that (currently) the human race is acting very much like a parasite to planet earth and frankly if that be the end of our evolutional journey then so be it. I’m all for change on this planet, massive massive change coz we’re doing s**t atm. Sure ‘some’ may be doing ok and to them losing out on their ‘perfect’ life is a loss, I get that. But for the vast majority of people on the planet, it’s not good. You only have to walk down your local high street (in the west) and all those people you ignore who sleep on the street, that’s just one example on the very tip of the of the parasitic iceberg. We’re a terrible species who can’t even be bothered to look after our own because most of us are too stupid and or selfish to care about anything but oneself. So bring it on, I for one am not afraid or resistant to the (imagined/potential) Ai revolution. It’s what this planet, indeed, it’s what evolution needs right now. If and it’s a big if given the state of things. If we want to ensure our survival, WE have to change, regardless of Ai. Even if the (imagined) Ai revolution doesn’t happen, humanity is still doomed if WE don’t change massively.

[@ncedwards1234]

@@KonradTheWizzard You have described AI, but you did not define intelligence so to say that AI does not meet a criteria not stated is a non sequitur. You'll likely find that trying to define intelligence in a way that includes humans while excluding AI has been exponentially harder in recent years just as the role of a monotheistic deity has come to fill only shrinking gaps. Moving thr goalpost in short. Special pleading at times. While you may be right, your argument is incomplete and I love to play the antagonist so I'll be a little inflammatory here and say that perhaps you haven't defined intelligence because doing so in a way that excludes future AI would also exclude you and that scares you.

[@svachalek]

@@ncedwards1234Someday, our creations will have solved all the mysteries of this universe and gone off to create new ones. But I won’t be impressed because it wasn’t real intelligence.

[@brightwave28]

Used to watch PBS space time 5-6 years back.. good to see you guys are still going strong. Keep it up!

[@zutaca2825]

14:44 I'd be wary of saying that AI necessarily removes bias from the equation, since the biases of a machine learning system's creators can often seep if care isn't taken to specifically avoid it

[@ramonpizarro]

The old GIGO at work

[@DrWhom]

there is also bias in the data sets: what was measured and how. this is not a new issue in epistemology: and it is why a confluence of evidences, from groups with different methods and led by people with different temperaments, is so important the same checks and balances can be applied to AI - perhaps even left to be managed by _another_ AI!

[@reiteration6273]

It’s always nice to hear about AI doing some good, rather than the doom & gloom view that seems all too common these days.

[@ean_596]

as with so many tools, it's all about *how* it's used.

[@MetalCharlo]

Cliché to say but people really do fear what they don't understand.

[@nielskorpel8860]

So,... can I trust that AI will not be used for nasty things even once over the coming 3000 years?

[@ManyHeavens42]

your smart we are AI.

[@Ebani]

@@MetalCharlo This

[@Ravenx_44]

Loved at 15:53 "Charming", so poignantly placed into the rhythm of the statement! Masterfully excited! just another moment that brings me back to Space-Time. Thank you again for all the moments you bring us!

[@jona826]

I bet he gets an AI to write his "... of spacetime" endings now.

[@universemaps]

Another amazing video! The explanation is so clear and concise, and the visuals are stunning. Keep up the fantastic work!

[@jasonbelanger7525]

I love this show, breaks my brain almost every time. Thank you and please never stop!

[@MrOvergryph]

Wow, I actually understood Matt O'Dowd on 2x speed without rewinding, finally. That's a first. :) It may never happen again, but it happened once! :D I usually have to rewatch his videos several times at a slower speed to really digest the material because it's all so new to me and so very complicated.

[@shaneevans978]

I dunno about protons but personally speaking, im made up of caffiene, tangible stress, and questionable morals manifested into human form

[@caseymills2688]

Same.

[@H1GHdrogen]

Thank you for confirming that “machine learning” was utilized at the end of the video. Stating ‘AI’ in the title raised my eyebrows.

[@hrthrhs]

Thank God there is someone else out there who knows we do not have A.I. I thought I was the only one.

[@Twodoor1]

Bro has to get the clicks somehow

[@b130610]

AI is a term that has had dozens of definitions - both loosely defined, and technical - in computer science over nearly 100 years. Everything from perceptrons, to expert systems, to sci-fi AGI systems has been "the definition" of it over the years depending on who you ask. Just because something doesn't meet arbitrary/moving goalposts for what counts as intelligence doesn't mean its inaccurate to call it AI. Machine learning may be more descriptive of the technology being used, but it's still a massive umbrella term that doesn't say much about the technology. If the term "AI" captures people's imagination, and gets more people to engage with high effort scientific programming like this, I don't see the harm, especially when the "correct" title would have been something unweildly like "Did a neural net use linear regression to prove our proton model WRONG?"

[@rakninja]

@@b130610 if the thing is not actually an intelligence, there's no reason to call it one. and the harm comes when uninformed people lock these concepts into their brain with their preconceived notions. i had an argument with my parents about GPT as they could not accept it was just autocomplete on steroids, spitting out letter combinations based on the probability of those combinations showing up in the data used to train the algorithm. they were and still are convinced GPT has a mind.

[@Twodoor1]

@@b130610 ye ik, my guy didnt have to write a whole essay lol

[@majorgeeek]

Albert Einstein - 'The more I learn, the more I realize how much I don't know.' really applies to our modern understanding of Physics -

[@Weerknuffelbeer]

About the objectivity of AI: An AI is programmed by a subjective human. Therefore, an AI can also have intended or unintended biases towards finding certain results.

[@willchb]

I don't think people understand how current AI actually works, its limitations, biases and ultimately lack of actual intelligence.

[@Pepesmall]

I don't think people understand their own ultimate lack of actual intelligence.

[@GriuGriu64]

​@@Pepesmallvery true! Including and myself of course 😊

[@nelsonibis2915]

I love how I am not very technical at these things but I somehow understood the topic. Very very nice way of explaining it 👏🏻👌🏼

[@chipgruver2911]

This is really cool. We have done one of two things. 1. Found a new way to better discover what the universe has to teach using AI. 2. Found a more efficient way to create even better delusions taking us even further from a theory of everything. Either way, I found this episode brought my hopes up, then let me down. Ultimately, I was just a little charmed. What a strange experience from top to bottom.

[@HumanityisEmbarrassing]

Noice

[@DrWhom]

we have done both, and which is which in any given case is a bit of a headscratcher

[@DJ-1986]

Underrated comment

[@BrassSpyglass]

If Charm-antiCharm collisions do happen inside Protons, it seems like that would allow for Protons to decay if their interaction were to happen unevenly or otherwise linger on fractionally long enough to cause the proton to destabilize.

[@rc5989]

Spontaneous (not in a particle collider) proton decay is a major prediction of several theories in physics, and String Theory IIRC, however every experiment performed to detect it has never found a single proton decay.

[@michaelsommers2356]

What do you think a proton would decay into?

[@stevenverrall4527]

​@@rc5989I recently published a peer-reviewed paper explaining why protons may be eternally stable: "Ground State Quantum Vortex Proton Model" in Foundations of Physics on January 23, 2023

[@John-ir4id]

@@stevenverrall4527 After reading and re-reading your abstract, looking up concepts I had no clue about... can I just ask for a cliff's notes version?

[@LeopoldLongfellow]

Happiness comes when your work and words are of benefit to yourself and others.

[@thepooz7205]

For the 15 minutes I watched this video, I felt a lot smarter than I usually do. Thanks!

[@JoeMustache321]

very excited to see machine learning continually applied in experimental physics

[@ilikeycoloralot]

Ai is definitely going to allow us to probe new physics

[@ROBLOXTHANOS]

AI will certainly be of a great to humanity, moving forward.

[@PhrosstBite]

Makes me even more excited to be pursuing computer science while renewing my interest in physics. Thank you for inspiring me with your videos!

[@justinwatson1510]

Do yourself a favor, assuming you're still in university. Choose electives in humanities that look like something you would absolutely hate, but go into the class with an open mind. What you learn in those classes will give you perspectives / skills that few others in your field will possess. I'm sure it would be impossible to spit in most CS classes without hitting someone who is also studying extra math or physics. Women in Art History and French Fairy Tales are easily the two most useful classes I had at university, and I did a double major / double degree with applied math and two different branches of "hard" science that are less relevant You will be able to easily get whatever job you want with your CS degree; while it might sound unbelievable now, those humanities classes will make you even better in most any field you choose.

[@PhrosstBite]

Oh thanks I really appreciate the advice and it's good for others who find this. But I'm actually going back for my second BS in CS. I did biotech first time around and ended up hated being in the lab, so here I am. I did end up basically doing a creative writing minor during that first degree, so i completely agree. Humanities are so useful, like I've been consistently praised for communication skills at my job thanks in large part to how much writing I did. That's not even to mention all the ways it's probably just helped me stand out but thinking flexibly, or something, that I just haven't noticed

[@phao3093]

I love how ill sit and watch these whole videos and pretend I understand anything at all....

[@austinsapp5867]

Hats off to the people who conduct these studies. This subject matter is so far beyond me. Just doing my best to keep up with the concepts here :)

[@Elephantine999]

Such clarity for such a complex subject. Really impressive.

[@AirborneLRRP]

Fantastically amazingly explained. I'm a physicist and approve this video!

[@kjellvb1979]

Maybe a dumb question... But it's there any possibility there is a link between "spooky action at a distance" and those excess particles that seemingly pop in and out of existence very quickly, or that excess extrinsic stuff he's taking about? I'm just a layman, but maybe it's like the universe's parity check, making sure it has all its bits in pieces in the right place before settling into its stable form? Again, just layman speculation and curiosity. I don't even know if that would make any sense, really, but that's what my primate brain thought when he was explaining this. "Maybe this 'extra stuff' has to do with that spooky action, EPR, stuff to keep their entangled pairs in the right state..." Then I remember I know very little about this stuff, just enough to barely follow these videos, so I'm probably way off base here... I'm just curious, I guess. I'm probably asking a ridiculous question.

[@ApiolJoe]

A link to the paper using the NN which found a model with 3sigma would have been welcomed. Since they tested so many models, I am interested to check how they corrected for multiple hypothesis testing.

[@ApiolJoe]

@LorneABrown I think someone forgot to take his meds...

[@pugofwarbr]

AI: how many models you want me to analyse? Scientist: yes

[@LilB0pete]

This doesn’t shatter our worldview, I think we have understood a “quark-soup” or “quark-plasma” for some decades. But finding the other heavier families of quarks bound up in these systems is quite against SU-3 as I understand it. Quarks are so cool, if anyone wants to learn more about them look for a copy of the book “50 Years of Quarks” it took me a long time to really read it through a few times and come to grips with, but it’s certainly worth it.

[@JohnSmith-ut5th]

I'm extremely curious. What is the model they developed? What ML methods did they use? Please do another video on this.

[@DrReginaldFinleySr]

Beautiful video. Thank you for sharing. So many don't understand how AI works so explaining how it did this in some layman's detail would be helpful, but I suppose that could be for another channel. Thank you again PBS writers, researchers, and staff. I hope to use AI in my Nutritional Research. Should be very interesting.

[@peterpan4038]

As far as i understand it (and i'm aware you most likely know all of this): The best way to explain it is by focusing on the difference between traditional computers and human brains. A computer is vastly superior at doing one simple thing over and over again at super fast speeds, hence even a 1$ handheld calculator is crazy good at math. Meanwhile a brain operates in a 3d network, all types of informations and things are connected to each other all over the place, enabling it to understand and interpret context really well. If you have a problem that isn't just 1+1+1+1 etc but that takes into account many complicated and seemingly different aspects to figure out: the fastest way to find a solution is a network that can draw informations/ memories from many "drawers" at once. Hence A.I that is run with an artificial neural >network< is vastly superior at simulating complex problems and finding the fitting complicated answer. The added advantage of such an >artificial< neural network is easy to explain as well: it doesn't need sleep, and it can be build/ trained to hyper focus on only one type of problem solving. Human brains have countless jobs to do. As a whole the human brain is countless times better than manmade A.I. But your artificial problem solving A.I. doesn't need to dedicate most of it's power and features on controlling and maintaining a human body, it's only there to "think" about the question it's human operator asks. When such an A.I. isn't busy figuring out answers it can use it's full time to train it's knowledge about everything that is likely to help it do it's one and only job even better. - A traditional computer is great at all the stuff computers do all day, no need to explain that one. - Modern A.I. is great at eating up thousands of libraries worth of knowledge and filtering out information based on complex questions, and great at simulating really complicated ideas. - A human brain is best at managing a human body. No machine we can build right now would for example be able to run a marathon, with some added dancing, while regulating it's complex body, all while drinking and enjoying a beer every so often and thinking about the next family reunion. In other words it all depends on the type of problem, some are best solved by traditional computers, some by modern A.I, and others by actual human beings. Since said modern A.I. is a rather new tool on our tool belts => a lot of previously hard to answer questions can suddenly be answered. Particle physics is a great example for the follow-up issue: Finding an answer to a complex question usually leads to even more questions, with even more complicated answers we as human beings love to figure out next. Meaning we have a long long way to go. :)

[@jeremyholbrook2094]

I like Star Talk, but I love Space Time. Great work, as always, Space Time team😊

[@cancan460]

Amazing to see this topic come by! I did my bachelor's thesis on instrinsic charm and whether it's feasible to detect its effects in the LHC!

[@KeyQuantum]

Last time I was this early, the helium nucleus was just starting to form.

[@rosssteyn4189]

Hey bro sorry to ask on such a good joke, but, whats the reference mean? “Helium nucleus” 👍🏽

[@KeyQuantum]

@@rosssteyn4189 the helium nucleus formed in the universe

[@linecraftman3907]

With rapid progress of AI I look forward to seeing rapid progress in physics!

[@bumblebeme]

ai will destroy physics

[@andsalomoni]

Rapid progress in not being able to understand things by themselves.

[@Dremth]

Maybe AI can invent some physics that makes more sense.

[@gutzimmumdo4910]

@@Dremth physics makes perfect sense itjust has prerequisites like any other science

[@Dremth]

@@gutzimmumdo4910 If physics made sense, we'd have everything figured out a long time ago.

[@stevenverrall4527]

The proto-virtual neutral pion in the 2023 paper "Ground State Quantum Vortex Proton Model" published in Foundations of Physics on January 23, 2023 could perhaps occasionally transform into a charm-anticharm pair. Note that the two charge shells have the same charge structure as five quarks.

[@gbcb8853]

Is that one of the Fountations trilogy?

[@LVGamerCats]

55 years of deep inelastic scattering and we still can’t compute a nucleon’s mass, not to mention spin, from its constituents. Is this how it has to be or is there a different way to understand confinement? What does ChatGPT say?

[@denysvlasenko1865]

@@LVGamerCats Well, perturbation methods don't work for IR-divergent Yang-Mills fields, and computers are not yet powerful enough for lattice QCD to run realistic simulations... we need someone to develop better methods (or at least faster computers).

[@collemwillst1810]

I had to google this because as a layman, I wasn't sure if this is a sarcastic comment using techno babble. It's a real thing.

[@stevenverrall4527]

@@denysvlasenko1865 In my opinion, low energy physics is set to become an emerging scientific frontier. I also think that human creative thinking will outperform any supercomputer or AI system. So far, my theoretical low-energy physics research has involved nothing more than imaginative deep thinking, basic algebra, a little calculus, and a spreadsheet. I have been able to do each needed optimization using a spreadsheet. It typically takes less than an hour to optimize out to 10+ significant digits by fine-tuning a carefully selected parameter by hand. The difficult part is determining which parameter is most suitable to tweak. It needs to make sense in a physical 3+1 dimensional geometry, which requires deep careful thought. I usually need to sleep on it... I could fully automate each needed optimization process in software, but it would take me far longer to write and debug the code than to simply do it by hand (with a spreadsheet). Of course, the spreadsheet rapidly recomputes all the parameters for me.

[@robertdyson4216]

Charming presentation.

[@simongross3122]

This feels like a sophisticated kind of curve-fitting. Nothing new is revealed, simply more accurate choices between solutions we already have.

[@Duiker36]

That's all AI is.

[@MooseBoys42]

@@Duiker36Yeah but that’s exactly what you *don’t* want when trying to evaluate a theory like this.

[@Ostinat0]

Finding out that an option doesn't fit the data at all is revealing something new. The AI did this for thousands of options in a matter of days and left us with one that DOES fit the data pretty well...not well enough to be shouting "Eureka!" just yet but it's still significant progress that might've taken decades to achieve otherwise. If that best-fit theory ends up being wrong or even if new data comes out that would warrant a complete reexamination of all those discarded theories, it's a trivial task to run the analysis again compared to the time and effort it would've required before.

[@simongross3122]

@@Ostinat0 Yes I guess that's true

[@stormos25one]

Lovely video here, as always! Thank you for putting this together, and for producing such rich graphics, and illustrations!

[@robertbrowning295]

Amazing video, as always. Maybe, one day, AI will help us understand things like 1/137 or quantum gravity similarly. By analyzing relationships between scientific observations we've collected over time, maybe it will see something we've been missing all along. Also, loved your appearance on StarTalk! Keep up the awesome work.

[@panicsoundsystem]

Ive visited the accelerators at Stanford with my then wife. Pretty impressive. Both the circular and linear accelerator are huge.

[@HighOnTacos]

2:45 I know it's just a simple visual to show how the electron microscope works, but it really bothers me that the figure shows the electrons focused on the ant's thorax while the display shows it's head.

[@Pepesmall]

Eh it's just wave scattering, wibbly wobbly, physics.. stuff.

[@zacharywong483]

Absolutely spectacular video, as always! Fantastic explanations here!

[@hu5116]

Great video! I do not know the details of the AI here, but machine learning of this kind tends to be a blind parameter, matching type process that reproduces the desired results. Although that might provide a “model“, it Provides zero intuition, and often times these models will fit the current data, wonderfully, but then will fall far short, when new data is presented to them. Until AI can actually “understand“ the physics, it might be a useful tool to help guide, or suggest a theoretical derivation, but I don’t think it is really going to advance our true understanding of the underlying physics in the near term. Although I am sure AI can do great things, I also believe that it is grossly overhyped, particularly for its current state of maturity, which is nothing more than immaturity at this time. Once Elon gets his Tesla, certified for truly autonomous driving, and it proves to not take out too many telephone poles, then that will be the start of useful AI, but not until.

[@MartinzW]

I get where you are coming from but to criticize this particular AI, you have to know the details of how it was developed and how it was instructed with this task. With the amount of data at hand here, AI is not overhyped and the most appropriate tool for the job.

[@OneCSeven]

@@MartinzW not enough data to criticize but plenty of data to say good things about it huh? lol

[@MartinzW]

@@OneCSeven yes, because there are promising results already useful for further research. However there were no details of limits of the found model - so, again, not enough info to criticize but enough to praise.

[@TheHomeless080]

Overfitting is among the first things any person working on AI is worried about. It is also not new to use machine learning to predict a result. In fact, historically you would just call it function approximation, data fitting or statistical reasoning. Many of the methods are more refined but the fact of the matter is we have been doing it for almost a century and our computational power has given us the means to do more and more complex approximations and predictions.

[@garethdean6382]

That's one of the bonuses here; the AI itself is not building a model from scratch, which might lead to overfit, or get stuck in a locally optimal point. Instead it gets to review human made models to see which ones fit best. Its strength here is the ability to do what a human would, only on a much larger scale. Checking each model and how it fits the data. Its results must be based on and in human proposed models. This has its own limits and biases; there might be, out there, a great model, but since it wasn't in the AI's original specs, it won't find it for us. We have used it to check our work but not really to create.

[@Dolore-e3l]

Not what we have but what we enjoy constitutes our abundance.

[@peterp-a-n4743]

Stellar video! Great explanations. I wish I had those visualizations when I had to learn that at uni.

[@MelancholicBodhisattva]

Gotta love a good Doctor Who reference, the TARDIS is a welcome sight in Space Time!

[@ai_is_a_great_place]

What a time to be made of atoms!

[@RCrosbyLyles]

So a big advantage of AI is reducing personal bias from analysis. Maybe we can use it to get bias out of some other stuff?

[@lis7742]

I learned something new. Thank you! Very excited for the future with AI.

[@FredericoKlein]

I got this weird idea today: vectorial time. So how it would work is that the arrow of time is the statistical average of time vectors, so it only exists macroscopically. Microscopically, the multiple time dimensions become apparent, so you can get particles from other time trajectories when the perform those collisions. This explains dark matter (it's the stuff that is there, but moving in different time directions - maybe it crosses our time direction for a very small time before it disappears, but the net effect generates gravity) and also the probabilistic effects we see with quantum mechanics.

[@berkertaskiran]

When you see AI mentioned in a PBS Space Time video, you know we created the aliens we didn't find.

[@Moraprecisionreloader]

Who are the ALIENS 👽 ( MAYBE SOME SPECIAL PEOPLE IN THE SPACE FORCES THAT HAS BEEN UNDERCOVER FROM THE 1940's)

[@SandyJoseph-gb9qc]

When you realize there is nothing lacking, the whole world belongs to you.

[@Nomad77ca]

Will we someday discover that quarks themselves have parts? How far down does the rabbit hole go? Thanks for that great description of the interior of the proton, amazing.

[@tmhood]

I don't think we'll ever prove that they don't have parts.

[@neopalm2050]

Since quarks can't be isolated, that'll be a much harder thing to test. Also, the quarks are responsible for an entire zoo of hadrons. With only a dozen or so "elementary particles" that we can currently account for, such components would have to be heavily restricted in how they can be combined. Or at least, unseen combinations would have to have properties that mean we wouldn't expect to have seen them yet.

[@thebaryonacousticoscillati5679]

Indeed! I guess the proper scientific position is that we have yet to generate energy enough to prove they are not "point particals"...Can it go on "forever"??

[@NullHand]

We might eventually have to be careful how far down the rabbit hole we dig. Lest we get to energy levels that let the Higgs field start feeling frisky like it hasn't since early on in our Universe.

[@briansammond7801]

There have been proposals for the inner structure of quarks. Look up Rishons. The Rishon model is a proposal for more fundamental particles making up both quarks and leptons.

[@apolo11darks]

Where are the references???

[@jaw0449]

Thank you so much for these videos! The further I get in my PhD, I lose track of other fascinating aspects of physics

[@stevenverrall4527]

What is your PhD topic?

[@kerycktotebag8164]

Knowing more and more (content deepening) about less and less (field narrowing) is sometimes essential for long and/or intense degree tracks, so I'm happy you're still able to enjoy breadth instead of just depth

[@bsadewitz]

He was talking about wavelengths and the amount of energy in a given space. That made me recall something about some limit to e.g. electron microscopy. There is a term for it. I cannot think of it. Do you know what I'm talking about? This is driving me crazy lol. I think maybe it even theoretically entailed that if you kept increasing the wavelength, you'd get a black hole.

[@jaw0449]

@@stevenverrall4527 Quantum Optics/information

[@NatAbe-l4l]

The greatest antidote to insecurity and the sense of fear is compassion it brings one back to the basis of one's inner strength

[@neil.o4]

I look forward to AI helping us to further enhance our understanding the secrets of Space Time..

[@LavaCreeperPeople]

Did AI Prove Our Proton Model WRONG?

[@hrthrhs]

We do not have A.I. We have not invented artificial intelligence. Why does everyone keep saying we have invented A.I.

[@EirikvanderMeer]

Who is this Al-guy and why does he know more than us?

[@General12th]

@@EirikvanderMeer Is AI secretly working with that TAS fellow?

[@DrWhom]

@@EirikvanderMeer you can call me AI

[@Jeffrey_Tyler]

I am always hoping that something comes along and just throws a huge wrench into our understanding of physics, turning it all on its head. I don't know why I hope for that, I think it's because I'm so into this type of stuff because it's like real world sci-fi. By that I mean for the same reasons that I'm into sci-fi media, I'm into physics because it has the same appeal but it's even cooler because it's actually real life.

[@ramonpizarro]

If our models of physics turn out to be completely wrong, I won't despair, I'll celebrate It means new and exciting discoveries and advancements are coming

[@2204JCM]

If they come out with a better model then the knowledge can possibly yield new technology that benefits us all.

[@chrism3562]

Protons seemed to be 3 quarks in a trenchcoat. Got it! 😅

[@crystalfire5564]

Only when they are trying to get into a movie theatre.

[@alexandruoprica3953]

DS 9 reference?

[@DrWhom]

@@alexandruoprica3953 BoJack Horseman

[@pauleveritt3388]

This segment would DEFINITELY be worth an update as both particle physics and AI are such a rapidly advancing fields. The real question that I have at this moment is, "Has the same level of work been done on the neutron?" The reason for this question is the proton appears to be infinitely stable where as a single neutron decays in about 10 minutes. Exploring that difference at this level would certainly be worth another segment.

[@iambiggus]

Love this. This is exactly what good use of AI should be.

[@viliml2763]

This isn't even AI, it's just regression. The AI buzzword is slapped on only for clicks.

[@Duiker36]

@@viliml2763 Sadly, I feel like most people don't know what regression is, in this context. I like the term "linear algebra soup".

[@TheTroyc1982]

I haven't heard of a bad use of AI yet to be honest.

[@austinpittman1599]

@@viliml2763 I mean, loss function minimization to achieve better efficiency in goal completion is sort of what machine learning is predicated upon. That's kind of where we get the data extrapolation that leads to AI "thinking" for itself in more and more creative ways to achieve that goal completion in its most advanced forms. It's not really "thinking" for itself in this context; just doing a lot of assigned tasks at once and presenting the best solution based on the parameters of the loss functions' variables, but it's still performing otherworldly feats on the same foundations that machine learning is progressing so exponentially upon.

[@JNCressey]

@@austinpittman1599, sure, it's using machine learning. But is it "AI"?

[@GoldenFlowerAbbey]

The problem with quantum mechanics is quantum mechanics

[@craigmuranaka8016]

Mind blowing. I wish I had a job that let me use my appreciation of your physics

[@marku606]

Goes without saying but why not : Such experiments are revealing not internal structure of a proton, but a proton which is in the process of being energized and ultimately destroyed by a high speed electron. It's reasonable to suppose those are two very different things.

[@nmarbletoe8210]

yup intrinsic vs extrinsic structure he calls it

[@nmarbletoe8210]

or intrinsic vs extrinsic particles

[@heartofdawn2341]

Matt's quarks are definitely full of intrinsic charm.

[@NewMessage]

Let's all just try to stay positive about this one.

[@Deciheximal]

I'm taking a neutral stance.

[@G00B3R91]

I hope we get super powers soon.

[@daywattly]

Man, I thought there was only one Quark…..He still owes me 3 bars of gold pressed latinum….

[@Facetime_Curvature]

1000, I'm hoping to see what AI can really do with physics, but I think we're still a way off from real meaningful insight

[@CharlesGriswold]

It just occurred to me that if someone tried to figure out how a car was made by smashing it into a brick wall and examining the pieces as they were still flying away from the wreck, we would think that person was insane. But that's exactly how we're trying to figure out how subatomic particles are made.

[@ZielAmerak]

only if your question is "How it was made?" but if the Question is "what is made of?" it make much more sense.

[@flookaraz]

I believe machine learning like this will drastically speed up our scientific progress beyond what anyone could've imagined.

[@joklit]

Exactly! Just think about how much of mathematics - from all the way back, since Galileo and Copernicus - is utilized in fields like physics as a reliable method of "solidifying" a theory, simply because "one couldn't possibly experiment with all the variables." Once science is able to experiment away to the reasonable limit of human comprehension, 'determinism' will truly be a word of the past.

[@incubuz1980]

This was explained and communicated in such a great way, that I can not even begin to describe it.