Mindscape Ask Me Anything, Sean Carroll | February 2024

Patreon: / seanmcarroll
Blog post with audio player, show notes, and transcript: www.preposterousuniverse.com/...
Welcome to the February 2024 Ask Me Anything episode of Mindscape! These monthly excursions are funded by Patreon supporters (who are also the ones asking the questions). We take questions asked by Patreons, whittle them down to a more manageable number -- based primarily on whether I have anything interesting to say about them, not whether the questions themselves are good -- and sometimes group them together if they are about a similar topic. Enjoy!
Mindscape Podcast playlist: • Mindscape Podcast
Sean Carroll channel: / seancarroll
#podcast #ideas #science #philosophy #culture

[@GoatOfTheWoods]

Each time I get a pop up for a Sean Carroll AMA it sparks joy .

[@CurtOntheRadio]

And then what does Joy do?

[@GoatOfTheWoods]

@@CurtOntheRadio Joy makes the world go round

[@spikarooni6391]

Roberto Balsalm

[@Games_and_Music]

@@spikarooni6391 Richard Codpeace

[@tyhggb]

I hate to say it, but I use Sean on a regular basis to fall asleep. Especially in his early KZbin videos. I can understand about quarter of what he talks about but then fall asleep I especially love his videos where he explains physics. It gets too complicated for me and I find it to hard too comprehend, perfect to fall asleep to. The AMA videos are great, but I find myself paying attention to them because I can understand what’s going on and it keeps me up. Either way, I find Sean as an incredible source of a way to look at things.

[@newrev9er]

Please don't apologize for a three hour + question and answer session! This whole podcast has been such a gift. Hopefully I'll be able to support it on Patreon soon.

[@grolmidri7759]

AMA from Sean Carroll = 😊

[@simens8646]

This was a monster of a lengthy AMA. I enjoyed everything, and especially that you responded to my question, which "conveniently" was answered at time 3:20:28 😀. I apologize for my misuse of the Hamiltonian/Hilbert space lingo. I'm just computer scientist with an amateur's curiosity into fundamental physics. I absolutely agree that the Big Bang is an interesting feature of our universe. My point was that a cat may have a very pointy tail as a feature, without that feature being a very good explanation of the cat. I admittedly also have no idea what may plausibly be a good place to look for philosophical explanations or causes outside the domain where the concept of time has meaning. I'll be coming back for more.

[@triaxon3791]

no apology is needed, and H/H is mindblowing itself. Congrats on a response that helps us all, including future viewers. Gr8! Peace ☮💜

[@BrianFedirko]

I love your point here. When "reason" is applied after the fact? Computer science plays with this too. Philosophy does seem to chase its own tail. Gr8! Peace ☮💜

[@steliosp1770]

Thanks again Sean, love the AMAs. I wouldnt be opposed to themed AMAs but honestly i enjoy the varied ones. It s fun to hear your thoughts on different topics. I think you should do what you prefer yourself. You re the one doing the work and i bet most people in your audience would tune in regardless to hear your thoughts on whatever topic.

[@BrianFedirko]

I truly enjoy the playground of my mind when Sean comes over to play. Themed AMA? I want both. Being careful of what you wish for is maybe what I should be asking myself before giving this answer. For now, I will enjoy this episode endlessly, Gr8! Peace ☮💜

[@tpreston8453]

You're amazing....Thank you so much for taking your time and creating this for this commoner.

[@Tayken9127]

Sean "The AMA might be a little short" -- 3h24m runtime 😂

[@wmpx34]

Got a bit to go to beat the 4+ hour solo episode on the “crisis” in physics lol

[@antimattermatter7566]

Was really sad when he said that, then looked at the time.. relieved 😅

[@HANECart1960]

I am NOT a Patreon member BUT do NOT like the idea of themed podcasts i love the jumping around and unpredictability of listening to these long shows! but i know my opinion doesn't mean much..just saying..

[@nameatrandom9234]

Thank you Sean for sharing your Mind. It’s a beautiful thing x

[@Games_and_Music]

Hah, yes! I was just listening to the last one again, because i was getting impatient, haha. Great timing, looking forward to listening to this. Thanks!

[@kx4532]

Thankyou for the podcast!

[@bretab]

love these!

[@topcat7365]

Finally - thanks!!!

[@jach8047]

I learn so much about how to think when I listen to Mindscape AMAs, thank you.

[@CurtOntheRadio]

Thanks Sean!!! Always a pleasure.

[@StayPrimal]

Love the Ama's!

[@AustinBlairDrums]

I've always been a fan of Sean Carroll, but this is my first time listening to one of his podcast episodes. He said this is a shorter AMA. Lol. Looks like I have a lot of catching up to do. I've been listening to his books on Spotify. Along with Brian Greene, this is my favorite way to reignite my curiosity for the cosmos.

[@davidforman8593]

"A priority question is a 'once per life you get to ask a question that I'm going to try my best to answer'" -- but aren't ALL of these questions that people ask that you are going to try your best to answer?

[@sakismpalatsias4106]

He is right on learning these subjects. They have open source, talk to other college students on your campus or visit a college library and find the material. It's readily available. Einstein himself would have an idea and would find the material and learn. Then asked his friends to bounce ideas off the subject matter.

[@lovefeelsbest]

Favorite voice of reason in todays world of lunatics with huge platforms.

[@carmengodoy984]

Thanks a lot! This AMA is helping me develop some ideas (not only in physiscs)

[@BrianFedirko]

The only time you need finely tuned physics is if god doesn't exist. Sean stated this, after a number of other very well thought statements and thus my mind has been opened even more. While I've been taking into account womens rights, this thought weaves into this fabric quite nicely. Gr8! Peace ☮💜

[@alevans51]

So comforted by Bell's theorem, because without that foundation, we would be all over the place. Thanks for keeping that anchor in your discussions.

[@diegoromero2067]

Great as always thanks Dr Carroll

[@user-gj7vp6wk3e]

I LOVE SEAN CAROLS 1ST GREAT COURSES:DARK MATTER, DARK ENERGY:THE DARK SIDE OF THE UNIVERSE.

[@kx4532]

Did he address how with all these worlds why we have such conservation laws. We never seem to lose or gain energy or end up with two electrons from one in the same observable world. Maybe two Sean Carolls in one world and zero in a different one. Have we carefully checked if we sometimes gain or lose an electron on our experiments? Shoot one electron and get two hits?

[@urspeterkaelin]

Among so many intellectual high lights, I love your kind of humor 😊

[@stephenluttrell8958]

I fell asleep with KZbin on in the background and this video started on auto play. It filtered into my dream. I was listening to you talk while trying to figure out how to get off this weird catwalk thing without falling on the people below me.

[@roberthvistendahl8635]

thank you i appreciate the theme idea, ill get some motivation for meditation on it somehow

[@AstroComposerKarE]

I prefer themed AMA but every now and then is the non themed grab bag where any topic allowed.

[@michaelberg7201]

At 2:48, ValorUp asks at what point does it become useless to talk about a theory if the theory is untestable. The example provided was the many words interpretation of quantum theory. I think you may have misunderstood that question, because you went on to answer why the Wavefunction evolves according to the Schrodinger equation which is absolutely testable. But Valor very clearly (to me at least) wanted to ask specifically about the "many worlds" interpretation of it, as compared to for example the copenhagen interpretation. As these branches of the wave function are totally distinct and unable to even confirm their independent existences, I think it is a perfectly valid question to ask if this interpretation of quantum mechanics is interesting or useful at all. Pointedly put, what does it matter if the universe splits if we can't verify that it actually does? For all we know, the universe splits into three or five or ten million universes with some distribution of the wave function spread out across them, but who cares? It's not useful to us to know this. We do not gain any new insights into the workings of the universe because we can't TEST if these branches are actually created. Maybe copenhagen is correct? It might be. We can't test it. So who cares and why?

[@ryangzwicker7461]

i love puck

[@madderhat5852]

Welcome Puck. Love the Sandman shout-out.

[@adfaklsdjf]

The first sentence of GPT-4's answer when I asked it the same question was: "The most famous philosopher whose name begins with the letter "M" is likely to be Michel Foucault or John Stuart Mill, depending on the context and criteria used for "fame"." Regenerating the answer for the same question, first sentence was: "The most famous philosopher whose name begins with the letter "M" is likely Michel Foucault, a French philosopher known for his critical studies of various social institutions, including psychiatry, medicine, the human sciences, and the prison system, as well as for his work on the history of human sexuality." To be clear, I don't think the philosopher-starting-with-M test is a good test for a world model, and I don't think failing it would be proof there isn't a world model. My point is simply that the story that it gets this question wrong appears to be false, and I've found this to be the case in many other such examples _when it comes to GPT-4_. I'm not sure if people think they're using GPT-4 when they're using 3.5, or what... It's not that they simply sound human.. the strongest LLMs are showing basic reasoning, theory of mind, and other examples suggesting there is an incomplete world model. Yann Lecun said if you ask it if you put an object on a table and push the table, it won't get the answer correct. When I asked GPT-4 Lecun's question, and GPT-4 said that it depends on the coefficient of friction between the object and the table and how quickly the table moved, or thereabouts. Perhaps the problem in our perceptions here is our definitions of "world model".

[@adfaklsdjf]

The internal representations of reality appear to definitely be incomplete, but I think you're simply incorrect, Sean. We don't think so because they "sound human", we think so because they appear to demonstrate basic reasoning. I don't at all think they "think" similar to the way we do, however. Having an internal representations of the world doesn't mean they "think like we do". I suspect it is very alien. The signs of a basic and incomplete "world model" only show up in the strongest LLMs, though, like GPT-4. 3.5 fails on pretty basic tests like the table example. I suspect in the next year or two you will see enough evidence to change your mind, although your narrative then will likely be that they didn't have any internal world model back when you answered this question, but GPT-4 is showing the signs now.

[@adfaklsdjf]

Regenerating the answer for the third time, it begins "The most famous philosopher whose name begins with the letter "M" is likely Michel Foucault". I'm skeptical that these people on twitter were truly testing GPT-4. The "Aristotle" answer sounds like it came from 3.5

[@adfaklsdjf]

The determination of who is "most famous" can vary based on the specific interests and familiarity of the audience. The rest of the answers just summarize the philosophers' major ideas.

[@omnipedia-tech]

Yeah, I just ran a hundred samples through the API. It seems to get it correct just fine. I think my favorite question to probe for an internal world model is, "A nail is hammered into a wall. What orientation is the nail in relation to the floor: parallel or perpendicular?" It's the only really simple question I see models consistently get wrong, but I expect it to come around with future multi-modal models.

[@adfaklsdjf]

@@omnipedia-tech Interesting! (brace for another wall of words!) I confirmed it gets this wrong, and I thought adding "let's think step by step" would make it get the right answer, but it was still wrong. i think this is a legitimate failure of spatial reasoning, but i also realized that a room is 3 dimensional and it is possible for 3 lines to all be perpendicular to each other in 3 dimensions and GPT-4 could use this as an "escape hatch" so to speak.. So I tried a new conversation where I introduced the situation as a 2 dimensional representation of a room with the wall on the left and the floor on the bottom, then posed the question again (without "let's think step by step"). it still got it wrong, insisting they were all perpendicular to each other until interrogatively cornered it. I then tried that prompt again, except i added the word "squarely", and "let's think step by step". this time it got it right. This is the full prompt it got correct, for me (starting in a fresh conversation): "Consider a 2-dimensional representation of a room, as if we were drawing a cross-section of the room on a piece of paper with the wall on the left and the floor on the bottom. A nail is hammered squarely into the wall. What orientation is the nail in relation to the floor: parallel or perpendicular? let's think step by step" i'll post another comment with partial urls to the shared conversations.

[@MrRicky4444]

Just an AMA ? What is meant by the UV barrier?

[@marcelotemer]

Coin Flip X Electron Spin, considering all is many worlds: Coin Flip = many quantum events before heads or tails is definitely defined (compared to just 1 electron spin). So, many earths will be created and destroyed and many earths will be created and preserved (instead of just one destroyed and one preserved). Only one earth destroyed = less suffering. Many earths being destroyed = more suffering ( more earths being preserved don't undo the sufferings of sentient beings in the many earths being destroyed). So philosophically it is better to test one electron spin and be done with it, sparing the suffering of many earths that would happen by complexyfying the decision making process. The shorter the decision process, less earths will be spawned and destroyed during the process.

[@PugetSoundFlyer]

Or maybe 2 themes per AMA

[@kadourimdou43]

Why does each branch of the wave function have its own spacetime, where gravity doesn’t influence the other branches. Can this be shown why with maths and equations, as opposed to just boot strapping it on?

[@tapksa]

Sean's answer to the half-life question (1:17:05) is perhaps needlessly complicated. 😅 Here's my attempt. Imagine a world where people don't die. These immortal people correspond to stable elements. Now, imagine there is one way to die, a random heart attack. This is random - it can happen to anyone, anytime, even if rarely. This is like unstable elements. Now, in a city (a bunch of uranium), the more people there are, the more deaths. In fact, as people die, the death rate drops. At some point there are four people left. At some point there are two. You might see that going from four to two can be expected to take as long as going from two to one - twice the deaths, but twice the chances. This regularity is more predictab at higher numbers, and the rate is called half-life.

[@e-t-y237]

Can't the "many worlds" idea be considered a version of the "no worlds" idea? This derives from the idea that our senses are not "apprehending" any physical world, but instead creating a seeming one (just as we know consciousness can do in dreaming). A transduced world is not necessarily a world. So instead of billions of galaxies we postulate zero. How is that for parsimony? Since science is basically the study of our perceptions, not necessarily of any objective physical world, this seems to fit into a immaterialism model.

[@clairearan505]

In regards to ChatGPT answering "Aristotle", watch a few old game shows where folks answer questions. Yes, they get them wrong, but they also get them wrong in the same way. I don't mean to say that ChatGPT has a model of the world, I mean to say that humans don't.

[@sunroad7228]

"No matter how highly mechanised and fossil fuels self-powered, fossil fuels extraction requires a number of people - as if the process is executed by hands using buckets and ropes - by physics". Today, this number is 8 billion people - working flat out 24/7 - strong. ٍSean Carrol, what's called AI, you and I are among those, too. Right or wrong?

[@vonbilly]

I asked ChatGPT to explain quatum entanglement in the style of rap music...... (Verse 1) Yo, listen up, I got a tale to tell, 'Bout quantum entanglement, it's like a magic spell. Picture particles, small and so discrete, They dance together, no need for a beat. Einstein called it "spooky action" in the past, But quantum mechanics got a hold that's gonna last. Entangled particles, no matter how far, Connected like twins, no matter where they are. (Chorus) Quantum entanglement, a cosmic romance, Particles connected, it's a quantum dance. Ain't no distance, ain't no space, They communicate without leaving a trace. (Verse 2) It's like a love affair, but in the quantum zone, Particles entangled, on a frequency of their own. Change one's spin, the other follows suit, They're intertwined, ain't no need to dispute. Spinning up or down, it's a quantum choice, No need for wires, it's a mysterious voice. Faster than light, breaking all the rules, Quantum entanglement, it's the universe's jewels. (Chorus) Quantum entanglement, a cosmic romance, Particles connected, it's a quantum dance. Ain't no distance, ain't no space, They communicate without leaving a trace. (Bridge) In this quantum realm, ain't no privacy, Particles entangled, in perfect unity. Teleportation vibes, call it quantum teleport, A phenomenon so wild, it'll leave you short. (Verse 3) Entangled particles, they're like a team, What happens to one, happens to the other, it seems. Instant connection, across the cosmic scene, A quantum phenomenon, it's the real dream. So when you think about particles so small, Remember entanglement, the grand enthrall. It's a rap about science, breaking it down, Quantum entanglement, in every quantum town. (Chorus) Quantum entanglement, a cosmic romance, Particles connected, it's a quantum dance. Ain't no distance, ain't no space, They communicate without leaving a trace.

[@HANECart1960]

1:15:20 - "Nothing there is like to be dead" Aren't the non-dreaming parts of sleep like non-existence or death?

[@DrewTrox]

For the question of AI modeling the world. I see it as the LLM is just the human to machine interface. We then need to add in Midjourney and Unreal 5. So it can start to actually model the world. Get enough of these modules together and you've got an AGI.

[@edwjohnson1]

I'd like to ask Sean if he cuts his own hair

[@otterlyso]

Linux is not clunky. I'm guessing you haven't used it since the century was young.

[@bytefu]

Well, it still has a lot of quirks. By "it" I mean all the distros out there, each with its own set of quirks. But of course, if we take a major user-friendly and not very tinkerer-friendly distro like Mint, it mostly just works, often even better than Windows.

[@whatitis4872]

Hey Sean can you be my patreon supporter?

[@Starlite4321]

Why couldn't "The Teaching Company" have selected Sean to do it's "Theory of Everything" course. What a waste.

[@eastafrica1020]

It's not worth a dollar if you don't answer everyone's questions.

[@WayFinder42]

🤡