Mathematician explains Gödel's Incompleteness Theorem | Edward Frenkel and Lex Fridman
Вставка
- Опубліковано 12 кві 2023
- Lex Fridman Podcast full episode: • Edward Frenkel: Realit...
Please support this podcast by checking out our sponsors:
- House of Macadamias: houseofmacadamias.com/lex and use code LEX to get 20% off your first order
- Shopify: shopify.com/lex to get free trial
- ExpressVPN: expressvpn.com/lexpod to get 3 months free
GUEST BIO:
Edward Frenkel is a mathematician at UC Berkeley working on the interface of mathematics and quantum physics. He is the author of Love and Math: The Heart of Hidden Reality.
PODCAST INFO:
Podcast website: lexfridman.com/podcast
Apple Podcasts: apple.co/2lwqZIr
Spotify: spoti.fi/2nEwCF8
RSS: lexfridman.com/feed/podcast/
Full episodes playlist: • Lex Fridman Podcast
Clips playlist: • Lex Fridman Podcast Clips
SOCIAL:
- Twitter: / lexfridman
- LinkedIn: / lexfridman
- Facebook: / lexfridman
- Instagram: / lexfridman
- Medium: / lexfridman
- Reddit: / lexfridman
- Support on Patreon: / lexfridman - Наука та технологія
Full podcast episode: ua-cam.com/video/Osh0-J3T2nY/v-deo.html
Lex Fridman podcast channel: ua-cam.com/users/lexfridman
Guest bio: Edward Frenkel is a mathematician at UC Berkeley working on the interface of mathematics and quantum physics. He is the author of Love and Math: The Heart of Hidden Reality.
Gödel's first incompleteness theorem states that for any consistent formal system that is sufficiently powerful to represent arithmetic (which includes most foundational systems of mathematics), there exist statements within that system that are true but cannot be proven within the system. In other words, there are true mathematical statements that cannot be derived or proven using the rules and axioms of the system.
The key idea behind Gödel's theorem is the concept of self-reference. Gödel constructed a mathematical statement that asserts its own unprovability within a given formal system. This statement, known as Gödel's sentence or Gödel's formula, essentially says, "This statement is unprovable." If the system could prove this statement, it would be inconsistent because it would be asserting both its own provability and unprovability. On the other hand, if the system cannot prove the statement, it implies the existence of true but unprovable statements.
Gödel's theorems challenged the notion of completeness and consistency within formal systems and had a profound impact on the philosophy of mathematics. They demonstrate inherent limitations of formal systems and suggest that there will always be truths that lie beyond the reach of any particular system. These theorems have also influenced the field of computer science, particularly in the areas of artificial intelligence and algorithmic complexity theory.
Entanglement
5 formulas, : 1+tan(2)power=sec(2) power
I love that he mentioned Alan Watts, who had the best description of Goedel’s Theorem: “No system can define all of its own axioms.”
I was surprised, amazed how he connected them, but then remembered how wide ranging Frenkel's knowledge about things and people are (away from Mathematics).
I recall (and am now rewatching) his 2014 talk, when, at the beginning of his talk, the computer system breaks down, he tells the tech, ""Don't worry about it". And "we use computers so much these days, maybe it's a sign". Then still goes onto speak with such humility, humour, and a weird humbleness. Weird because he obviously knows so much but believes he doesn't.
The talk's from a book promotion tour, for his "Love and Math". If I remember correctly, he's got a few more of those Alan Watts like comments. ua-cam.com/video/YnqQ-BWMHrE/v-deo.html (if you're interested).
I don't get it. What does it mean to define an axiom?
That sounds like a vague description to me, which is fine for people like watts, but no good at all if you want clarity
@@Leksa135axioms are statements, so yeah, what does it mean to 'define' a statement? Seems like it doesn't mean anything honestly
@@Leksa135 At around 4:46, you'll see them show Euclid's five axioms (also called postulates). You'll see it says, "first axiom", "second axiom", etc. Underneath each of those is what is technically called a definition. What this is, is a definition for each axiom.
To return back to @FrigginTommyNoble's comment, the system of mathematics being used is called Euclidean Geometry. But the Euclidean Geometry system can't define it's axioms or itself, it needed Euclid (i.e. a person) to define them. Hence, no system (i.e. no mathematical system) can define all of it's own axioms. Which is what Kurt Gödel proved mathematical, or he proved that all mathematical systems will be incomplete, hence Gödel's incompleteness theorems.
This guy might be the best guest you have had on Lex I love this dude
agreed!
When I saw this dude on Numberphile a long time ago I knew he was the real deal. Such a delight to see him again here!
It's great to hear about Godel's Incompleteness Theorems. What he didn't mention is the motivation behind how Godel ended up with the Incompleteness Theorems. So at the end of 20th century, a genius named Georg Cantor (founder of set theory) wanted to understand God (just like Einstein wanted to understand God). For him God was represented by infinity. Therefore, he wanted to understand infinity, which we failed to understand for thousands of years (and still to this present day). So he asked himself the simple question, we can add numbers, subtract them, etc, but what about infinity? To this end, he constructed a theory that is now known as cantor's set theory. Unfortunately his basis of reasoning (called axioms) contained a contradiction that is now known as Russell's paradox. Nevertheless, he proved that the set of natural numbers is smaller than the set of real numbers. And he wanted to prove whether there is a set in between the naturals and the reals (that is now known as the continuum hypothesis), but failed to prove it. Because of these, David Hilbert (widely considered to be the greatest mathematician at that time and of the 20th century) came up with what's known as the Hilbert's program. In this program, he posed a number of problems, among which is the quest for a basis of reasoning that is both complete (that is to say, if a mathematical statement is true, then it must be provable from this basis of reasoning) and sound (that is, the basis of reasoning only proves true statements). Note that these two properties (completeness and soundness) are the two fundamental properties of all algorithms. Now, comes Godel. So originally godel wanted to prove that such a theory exists, but ended up with the incompleteness theorems in the end. But, the results are so earth shattering that it completely destroyed Hilbert's and mathematicians' dream of having a sound and complete theory for mathematics. On top of that, the mathematics he used to prove the incompleteness theorems were so new that only a handful of mathematicians understood it. Amongst these mathematicians that understood Godel's results were Alan Turing. So what Alan Turing said is that what Godel really means is that there is no machine/algorithm that is able to prove/determine whether an arbitrary program on an arbitrary input will stop or run forever. This is the Turing halting problem. It is important to note that all of these led to the birth of computer science and eventually led to the famous P vs NP problem. Historically, Godel was the first to informally pose the P vs NP problem in a letter to Jon Von Neumann. He also was the first to prove that Einstein's theory of relativity allowed time travel. And he gave the proof to Einstein as a birthday present.
@@dudeshiya Thank you for this write up! Is there any book you could recommend to read about this in detail?
@@blueskies3336 If you are capable of reading German, Dirk W. Hoffmann's "Grenzen der Mathematik" is a must. It is mathematically rigorous enough, but as comprehensible as possible for this hard topic. It gives a detailed historical account of the developments.
Maybe one could try reading it with the advancing translation technology...
I saw that in September 2023 a book called "Foundations of Logic" by Westerståhl will be published, it seems similar from its summary.
This is the first time I've been introduced to this guy. I like how he seems to be more of a "unification of knowledge" type of person, rather than just a mathematician. He draws from examples everything from math, to pop-culture, to eastern and western philosophy, and so on. Thanks again Lex!
Here are brief statements of the theorems for those interested:
Gödel's First Incompleteness Theorem states that "Any effectively generated theory capable of expressing elementary arithmetic cannot be both consistent and complete. In particular, for any consistent, effectively generated formal theory that proves certain basic arithmetic truths, there is an arithmetical statement that is true, but not provable within that theory."
Gödel's Second Incompleteness Theorem states that "For any effectively generated formal theory T including basic arithmetical truths and certain truths about formal provability, T includes a statement of its own consistency if and only if T is inconsistent."
Just prior to publication of his incompleteness results in 1931, Gödel already had proved the completeness of the First Order logical calculus; but a number-theoretic system consists of both logic plus number-theoretic axioms, so the completeness of PM and the goal of Hilbert's Programme (Die Grundlagen der Mathematik) remained open questions. Gödel proved (1) If the logic is complete, but the whole is incomplete, then the number-theoretic axioms must be incomplete; and (2) It is impossible to prove the consistency of any number-theoretic system within that system. In the context of Mr. Dean's discussion, Gödel's Incompleteness results show that any formal system obtained by combining Peano's axioms for the natural numbers with the logic of PM is incomplete, and that no consistent system so constructed can prove its own consistency.
What led Gödel to his Incompleteness theorems is fascinating. Gödel was a mathematical realist (Platonist) who regarded the axioms of set theory as obvious in that they "force themselves upon us as being true." During his study of Hilbert's problem to prove the consistency of Analysis by finitist means, Gödel attempted to "divide the difficulties" by proving the consistency of Number Theory using finitist means, and to then prove the consistency of Analysis by Number Theory, assuming not only the consistency but also the truth of Number Theory.
According to Wang (1981):
"[Gödel] represented real numbers by formulas...of number theory and found he had to use the concept of truth for sentences in number theory in order to verify the comprehension axiom for analysis. He quickly ran into the paradoxes (in particular, the Liar and Richard's) connected with truth and definability. He realized that truth in number theory cannot be defined in number theory, and therefore his plan...did not work."
As a mathematical realist, Gödel already doubted the underlying premise of Hilbert's Formalism, and after discovering that truth could not be defined within number theory using finitist means, Gödel realized the existence of undecidable propositions within sufficiently strong systems. Thereafter, he took great pains to remove the concept of truth from his 1931 results in order to expose the flaw in the Formalist project using only methods to which the Formalist could not object.
Gödel writes:
“I may add that my objectivist conception of mathematics and metamathematics in general, and of transfinite reasoning in particular, was fundamental also to my work in logic. How indeed could one think of expressing metamathematics in the mathematical systems themselves, if the latter are considered to consist of meaningless symbols which acquire some substitute of meaning only through metamathematics...It should be noted that the heuristic principle of my construction of undecidable number theoretical propositions in the formal systems of mathematics is the highly transfinite concept of 'objective mathematical truth' as opposed to that of demonstrability...” Wang (1974)
In an unpublished letter to a graduate student, Gödel writes:
“However, in consequence of the philosophical prejudices of our times, 1. nobody was looking for a relative consistency proof because [it] was considered that a consistency proof must be finitary in order to make sense, 2. a concept of objective mathematical truth as opposed to demonstrability was viewed with greatest suspicion and widely rejected as meaningless.”
Clearly, despite Gödel's ontological commitment to mathematical truth, he justifiably feared rejection by the formalist establishment dominated by Hilbert's perspective of any results that assumed foundationalist concepts. In so doing, he was led to a result even he did not anticipate - his second Incompleteness theorem -- which established that no sufficiently strong formal system can demonstrate its own consistency.
See also,
Gödel, Kurt "On Formally Undecidable Propositions of Principia Mathematica and Related Systems I" Jean van Heijenoort (trans.), From Frege to Gödel: A Sourcebook in Mathematical Logic, 1879-1931 (Harvard 1931)
Thank you!
This was one of my favourite shows from Lex. Edward is a truly remarkable human being, and it's always beautiful to see so much love and compassion in one's heart.
Great show. I really love Frenkel. He is so clear and his enthusiasm and sense of wonder is infectious!
Edward Frenkel is so extremely lucid, just extraordinary.
This is your brain on logic.
My ex-wife actually had Edward Frankel as an instructor at Berkeley nearly 10 years ago.
‘We’re still feeling the tremors today.” Wow.
I think it's valuable to also go a bit into the whole "completeness & consistency" thing.
One could start with definitions and explaining why they're important things we want from formal systems.
Then one could proceed to a little history of how "cracks" in set-theory based formal systems began to be discovered by Frege and Russel almost as soon as those systems arose.
The story continues with a quick overview of the various approaches to these issues, like ZF(C), NBG, "New Foundations" and type theory (with Russell for a while, then dormant for a long time, then getting a big comeback with Per-Martin Löf and lots more interest recently with Homotopy Type Theory).
This brings us to a classification and analysis of the underlying issue - that of predicativity and impredicativity - one might briefly explain what that is and why it's problematic - using various examples of paradoxa of (direct or indirect) self-referentiality.
We can then explain how these developments and the predominant research institutions in Germany and Eastern Europe lead to Gentzen's proof of the consistency of (Peano) arithmetic - and how that was a process of formalization which took us around 2.5 millennia from basic arithmetic and logic to Gentzen's proof. The importance could hardly be overstated.
The "victory march" of formalization and the power of formal systems seemed assured.
... and then came Gödel.
"and then came Gödel"
No, Gödel was first (1931) and Rosser's extension of Gödel's first incompleteness theorem (which is often falsely attributed to Gödel) was in the same year as Gentzen's proof, 1936.
Gentzen evaded the incompleteness theorems: the system he used to show that PA is consistent is not stronger than PA, but miraculously manages to capture the structure of proofs in PA (in particular, it is not weaker than PA).
I think Gentzen's result is even more breathtaking and few people (if any) understand it.
Godel vs Gentzen? Epic math battles in history? Please tell me this is a UA-cam series 🫡
What a great an inspiring guy Mr. Frankel is. It’s simply great learning from his talks.
Ed Frenkel is one of my favourite people. His book is fantastic.
A full 14 minute in depth explanation of goedel's impossibility theorems, and then lex goes "so every why has a definite answer"
??? every why still can have a definite answer but not on the same system
To be fair, as far as I understand Goedel's incompletness theorem (I think thats what you mean, 'incompletenes' rather than 'impossiblity'?) the theorem only states that there is at lease one state that is not within the universal set. And that by the virtue of at least one, that makes the whole 'completeness' incomplete.
It does not mean 'there are questions that can not be anwsered', it just means, 'Mathematically, there is at least one question that can not be anwsered'.
I choose to interpret that to mean "Do we get Pizza or Fish and Chips." being the question.
@@jaywulfnot necessarily a question being asked but rather and assumption being posed
@@jaywulfbecause a question would imply that if an answer was derived, the answer would be used to start the system of logic, but within the logic the axiom was already assumed without validation (except if you go meta)
9:28 No, no it doesn’t. It “proves” these things if you aren’t a shrewd enough logician to understand that a self-contradicting and semantically empty statement is simply “incoherent” within any coherent formal system (that is coherent with the Logos).
Also the halting problem is solved by the Logos and by translating programs into a number of versions until a halting version is found (if the original does not halt by the time that other halting version is found).
It’s all easy and these people that think LOGIC is imperfect or can’t prove everything are the fools, even if they’re “big names” and did some cool things.
When the Math doesn’t “add up” it’s YOU/US (the fallible mathematician and fallible humans) that screwed up, it’s not Math and Universal Logic (Logos) that were actually imperfect at their Root and in their Essence.
Perfect Logic is Perfect Logic and that is tautological and 100% certain.
There is no “truth” that is not encompassed by the Universal Logic or (objectively) true without being so according to the Standard of Universal Logic.
It is illogical and logically contradictory nonsense to speak of a “truth beyond Logos” or any aspect of Logic that cannot be symbolically represented with Formal Logic.
In another paper, Gödel developed an axiomatic system containing the self-referential statement, “This statement is false.” He then proved - within the same system -that “This statement is false” is true. All he needed was the countable numbers (the set N) and a few very simple rules.
On “Emergence:” Taking simple rules then applying them to a simple structure to produce complex “behaviour,” is also a subjective process. In what axiomatic system can you consistently define both “simple” and “complex,” then show that there are no self-referential contradictions?
The Chomsky hierarchy defines classes of complexity of behaviour. Any Turing complete system can capture the fullness of complexity, and they are all well defined computational systems. Also, the simple structure of these systems can produce the most complex behaviour possible: irreducibly complex behaviour.
In "another paper"? Which would that be?
"All he needed was the countable numbers (the set N) and a few very simple rules." This comic reference to Peano arithmetic and gödelisation really hurts!
I love his comment on the findings of Godel and Turing being "life affirming". Very well said.
I love this stuff. This channel never seems to disappoint.
Wow!! Great explanation of incompleteness! The best I have seen so far!
GAYDOH
would you say the explanation was complete?
Ha ha good one.. As complete as an be at this level I guess...
Gödel's theorems challenged the notion of completeness and consistency within formal systems and had a profound impact on the philosophy of mathematics. They demonstrate inherent limitations of formal systems and suggest that there will always be truths that lie beyond the reach of any particular system. These theorems have also influenced the field of computer science, particularly in the areas of artificial intelligence and algorithmic complexity theory.
It’s so obvious, you can’t put all your eggs in one basket. We need competing ideas, we need theories that “contradict” others, but because of this they work (meaning multi-valued logics that immediately eliminate Gödel’s theorems), etc.
Thanks for your clear exposition of the core ideas embraced in Godel's amazing theorem...
Great video, people take calculus and algebra classes for years and no one explains to them the fundations of what you are studying as clear as this guy does
Do people?
It takes a deep and complete understanding to make it as simple as it can possible be explained.
@@unkokusaiwa HAHAAA
this is a Thankyou to Lex and his team for committing to in-person interviews with high quality audio equipment.
It makes all the difference to the audience experience.
Reminded me about a study of the mona lisa smile, where they discovered that the smile appears to be more obvious if we use our peripheral view, so that its way she sometimes appear to smile but then again, she does not. Maybe some of those trick images may have an explanation that is more complex than subjective.
Greatly appreciated broadcast. Thanks for sharing! Ex-Maths teacher. Have a GOOD weekend!
Excellent interview
I wish my father had teached me mathematics like him. So calm and collected makes it easy😢😅
Don't forget that he has spent decades of his life dedicated to learning and communicating mathematics. It take a lot of hard work to be able to do what he does.
Is your dad a mathematician? If not the case, pls respect your dad.
@@linchenpal yes
Some people are natural at math but cannot teach or explain it. Others work harder to learn it, but do better conveying it.
But most people are bad at what they do and don't care to try to learn everything they are capable of.
Editing a foundational part of the whole interview✨🙏
Thanks for coming up with this interesting topic with this brilliant guy..❤🎉
Great explanation!
Regarding the perception problem at 14:52 , the top down perception in the brain can provide a trivial explanation. Just like Lex mentioned for neural networks, the bottom up sensory features leads two activated outputs: 0.5 Rabbit and 0.5 Duck. However, the top down awareness in the human brain can only attend to one output at time. So if you attend to the duck output, the duck neuron will be activated. Now because the information comes from top to bottom, all the related neurons to Duck will activate (none will activate for the Rabbit). And that is why you suddenly perceive it as 100% Duck or 100% Rabbit if your top down awareness attend to the Duck or vice versa.
simply giving it the name "topdown" is neither much of an explanation nor interesting. and who is this "you" picking what to attend to? that's the interesting stuff
@@iranjackheelson The brain is a reactor and converter not a creator. So this whole story goes into lala land categories.
@@s.muller8688 you're being sarcastic right? brain is exactly not just a converter or reactor. it is a creator indeed. part of the reason why you can't predict what you want to eat for lunch 100%.
@@iranjackheelson what you going to eat at lunch is already stored in the memory in the form of known data, which than randomly get's chosen by thought. Nice try.
@@s.muller8688 decision to what to eat for lunch is influenced by your prior beliefs and states, but it's not just stored as "known data". nice try
Frenkel is probably my favorite guest of all time I’ve rewatched the interview many times
There are some nice ideas about emergence of complexity. As nothing is the same, there is an incremental effect by repetition, similar to what our memory in the brain does with the episodic memory, each time we see a cat for example, the cat experience adds meaning to our definition of cat, even if it is the same cat at the same place. A bit like Peircean semiotics thirdness, when we interpret a sign it can generate a new one, even more if instead of just one triadic relationship there is a whole network of it, by aggregation and interconnection, at some point it generates more complexity of evolving meaning, because it cannot be the same, different than in mathematics. In mathematics if we add 1 + 1 it is always 2, but in reality that is impossible, and the 2 will be always slightly different each time we add 1+1. In short, complexity has to emerge because repetition is impossible.
I've been waiting for this!
Gödel should have trolled us and left his incompletence theorem incomplete.
What is the name of the knot displayed in the title?
I really liked when he jst softly said the key to genius . " To have open end process " to let yr conscious intelligence lead . Rather than deciding one thing and another
Wonderful guest! Obviously Brilliant, yet modest! 👏
Brilliant discussion
beautiful story so far really, guys...
I don't know why, but Edward Frenkel is my favorite guest so far. Invite him again pls!
Love your podcast, always a hit. But this was a grand slam.
Fantastic stuff one of my favorite clips.
I read that 3-SAT is Turing complete. And that's just basic Boolean algebra! And Gödel's incompleteness theorems don't apply to Boolean algebra. However, for example the set of natural numbers is infinite, so one has to use something like induction to define that in Boolean algebra. Or define a Turing machine that runs forever and outputs the natural numbers.
🎯 Key Takeaways for quick navigation:
00:02 🧠 *Gödel's Incompleteness Theorem Overview*
- Gödel's Incompleteness Theorem explores inherent limitations in mathematical reasoning.
- Mathematics relies on axioms, fundamental statements taken without proof.
- Euclidean geometry serves as an example of a formal system based on axioms.
07:18 🤖 *Formal Systems and Axioms*
- Mathematics is structured by formal systems using axioms.
- Different choices of axioms lead to distinct branches of mathematics.
- Rules of inference guide the logical derivation of theorems within these systems.
09:25 🔍 *Gödel's First Incompleteness Theorem*
- Gödel's First Incompleteness Theorem refutes the idea that all of mathematics can be algorithmically derived.
- It asserts that within a consistent formal system, there exists a true statement unprovable by syntactic processes.
- Challenges the notion that all truths can be computationally generated.
11:45 🧩 *Gödel's Impact and Open-Ended Exploration*
- Gödel's theorem sparked a revolution, challenging the completeness of mathematical systems.
- Analogous to Turing's halting problem, revealing inherent limitations in algorithmic comprehension.
- Encourages an open-ended perspective on computation, leaving room for future discoveries.
13:30 🌌 *Emergence, Complexity, and Neural Networks*
- Explores the emergence of complexity from simple rules, analogous to Game of Life.
- Discusses the challenge of training neural networks on ambiguous perspectives.
- Raises questions about the subjective nature of AI interpretations and biases.
17:00 🔄 *Niels Bohr's Complementarity*
- Introduces Niels Bohr's complementarity principle to explain subjective perspectives.
- Compares complementarity to ambiguous visual phenomena, emphasizing our limitations.
- Suggests embracing the mystery and open-ended nature of understanding complex systems.
Made with HARPA AI
Your summary is riddled with errors.
@@kurtgodel5236 You'll have to be more specific
@@TK_Prod Would be happy to oblige, if youtube algorithms were less prone to delete my comments.
Loops are similar to self referential statement that are connected to some paradoxes in naive set theory. But they are completelly resolved in modern set theory ZFC. Also there are no connection between these paradoxes and Godel results. Godel theorem are valid for every formal system strong enought to interpet Peano arithmetic.
I shit my pants when he tied everything to complementarity. The golden thread of Platonism. Would love to hear this guy talk with Joscha Bach, John Vervaeke, Max Tegmark, Penrose, Graham Priest, or any other modern great who understands and promotes this principle.
I've been watching a lot of physics videos lately, mostly from Drs. Sabine Hossenfelder (Science without the Gobbledygook) and Matt O'Dowd (PBS Spacetime). Both channels take dips into the Quantum and explore some of the weirdnesses within.
I've been thinking a lot about the ideas of complementarity and how they might relate to quantum superposition and the measurement problem. See, I'm starting to think that maybe the wave function doesn't collapse at all, but rather we are just observing one "version" of the particle. Its unobserved complement might still be just as real as the one we measure, but can't be seen at the same time. In a sense, maybe particles really are in two places at once, but we can only observe one at a time.
Thoughts?
@@sabinrawr yea I have a lot of thoughts about this. Safe to say there is a lot of confusion surrounding QM as it is popularly understood.
I tend to go with interpretations akin to objective collapse or Quantum Bayesianism.
That's why this is my favourite channel on platform
Wow. Bravo , that was thrilling
Lex on Godel's incompleteness and Turing's undecidability: _"It's very depressing."_
Frenkel: _"Or life affirming!"_
Edward is ahead of Lex in spiritual development. When we embrace that we will never attain a _Theory of Everything,_ it opens up greater possibilities!
I think Lex was joking about programming. There being no means to figure out the perfect program. :)
@@evertoaster Agreed, Lex was not being 100% serious in his answer
In cellular automata the complexity emerges not completely from the rules, but from action of the rules on the initial setup. The rules themselves are just the computational engine. And as any good computer it has the same halting problem. So add a good initial setup and you have complexity - which is a manifestation of the halting problem.
Would love to hear Douglas Hofstadter interview 🖤
I know the theorem and the proof and all the theory and i can say that i like the way he summariezed it for wide audiance
GAYDOH
Lex is such a good interviewer.
Be smart. Ask good questions. STFU. 😮
Postmoderns have substituted emergence for manifestation, as if "because, because" had any explanatory value.
He is a very thoughtful man.
Great talk as always, it is very humbling to remind ourselves of the ground where we're standing. Cheers from Brasil!
❤ from Brasil
This was fun to listen to.
The guest is amazing, and his eyes are so bright
I think the duality like it is mentioned towards the end of the video and also the problem, that identity can't work like in formal logic and math (where something is identical to itself and if it is different, it can't be identical, but in reality everything is always changing, like the ship of Theseus, where every part is replaced one after the other until no part of the original ship is still there) can both be solved if we assume, that nature is really dialectic. And dialectics also allows contradictions, so we don't have the a problem if we prove something and also it's contradiction. So maybe we could advance our understanding of the world if we assume that the natural world is dialectic and only obey formal logic under special constrained circumstances. We would probably need a new kind of logic then, which includes dialectics.
"I used the formal system to destroy the formal system." - Thodel
In mathematics and IT, sometimes, evaluating whether the a particular process will complete or not is just as complex as actually doing the process.
In 2015, a group from London proved that many-body quantum systems are analogs of Turing machines, essentially computing for the rules of quantum mechanics. Because the system essentially has to reference itself in optimization of electron distribution, within a limited number of excitations, they then went on to demonstrate that some properties, like spectral gap prediction, are undecidable. Reductionism has a limit and there are things in life not only that we can't predict, but neither can nature!
Эдуард как всегда на высоте - чертов гений!
Your talk with Edward took a most interesting path from subjective axiomatic foundations, creation of mathematics thru logical inference & syntactic process, to Goedele’s & Turing’s Theorems opening
space for new things.
You then jumped to cellular automata & emergent complex behavior, leading Alex to bridge to neural nets via figure-ground human perception. At this point I flashed on the interesting possibility that we are all (inescapably) training AI systems to be human in the most fundamental way possible.
My thought was that the methods of our perception are at their core, the syntactic evolutionary algorithms that created life and us. Also, that we are incapable of recapitulating ourselves in our dogs, children or AI, in any other way.
The intimacy of the evolutionary syntactic process that created us is the heart and soul of our perceptual engine. As we discover this syntax in recreating ourselves, we give this soul to our syntactic silicon selves, our AI.
Everything is layers of complexity. My research leads me to conclude that we are close to an inflection point in our understanding of the second law of thermodynamics. When that happens, this next layer of complexity may provide a foundation for a true science of life… and AI is our ashlar test vehicle.
I've heard it explained in terms of games. If you have chessboard that is halfway through a game say, there is no way to derive using the rules (axioms) what the board looked like say 10 moves before. That's about as far as I get.
This clip was amazig. I love it.
What this guy is explaining is at the heart of existence itself.
Is there a quantum generalization of logic in which statements can be in a superposition of true and false states, and is there an analog of Godel's theorem in the framework of quantum logic? (More basically, what is a "quantum proof"?)
Top guest 👍
This interviewee is a very smart man.
I think Euclidean geometry is a good example for the difference between physics and math. In math, the 5th postulate is just an axiom. in classical physics, it's derived from observation and in general relativity, which was necessary because classical physics didn't agree anymore with some observations (like the orbit of Mercury), it is only valid in the special case of flat space, which is a good approximation in some cases, but strictly only exists in an empty universe or in single points which have a curvature of 0.
Wonderful interview. If Penrose is right in that consciouness is not an algorithmic computation, as per Godel, then we are not wholly deterministic, a comforting thought!
Or that even if we are wholly deterministic, it is not in a way that can be replicated by any formal / computational approach known today. Penrose has basically said something similar.
@Mattje8 Very true. Do you think if we are not wholly deterministic, that Penrose's idea, if true, would explain the phenomenon? Could it be that superposition and entanglement provide human's with the ability to think outside the box, to get the "god's eye view" ?
13-42 Сложность возникает из одного из свойств Информации - которое говорит, Информация имеет тенденцию к накапливанию и усложнению своей внутренней структуры и содержания.
Complexity arises from one of the properties of Information (with a capital “I”) - which says, Information tends to accumulate and complicate its internal structure and content.
the dress. I saw both versions, depending on I don't know what. Sometimes I can see both within 5 minutes, but I have to look away and forget before I take on a new look.
That was great!
11-40 The presenter raised the question of calculation, but he understands this term only from one side. In fact, the word calculation can be understood as a process when images of information (not the information itself, but its stored state, impression, correlation) are transferred to the state of Active Information. For example, the image of the word “fox” is transmitted to the brain. The paper acts as a carrier of the image of the word “fox”. Note that there is no information about the fox itself in the symbols written on the paper, but there is a certain correlation. The observer, in the form of a light stream, acts on the paper (“reads”) the image of the word “fox” and stores it in the form of a frequency modulated signal with its spectrum. Note that nothing remained of the letters “fox”; the letter was considered an observer (light flux), produced in the form of active information and stored in its format in the signal spectrum. The human eye acts as an observer and reads the image of letters from the medium (light flux) and writes new correlations in its own format, for example, into energy signals traveling along nerve endings to the brain. Etc. And only at the last stage, the neural network of the brain, reading the correlations that came to it and using memory and reference sets, creates (generates) an active information flow and generates information about the fox (thereby reproducing reliable information from the word image stored on paper). So calculation is an analogue of the active process of the Observer, it is a mechanism as a result of which “real” Information is born within the context of this Observer.
Incredible, as a mathematician and AI scientist I loved what he had to say.
I practice magick for the exact reason he spoke of in the end. It's worth noting that Carl Jung also practiced magick and wrote and illustrated his dreams and other deep psych work in his Red Book.
People think, "oh you're crazy you think Harry Potter is real," but it couldn't be further from the truth. I believe in targeted rituals that help expand my own understanding of what lies below the tip of the iceberg in my subconscious. As well as to harvest results from it. I think it's worth looking into if you find it intriguing.
However, perhaps the counterexamples produced by Gödel and Turing come down to loops. A clear disjunction between formal systems and meaning happens in music. The works of Bach all follow rules which can be expressed as a mathematical formal system. The mathematical validity is a framework which helps support the emotional meaning (different for everyone), but the emotional meaning is outside the formal system.
The eternal golden braid
this is like listening to a theologian saying the practice of theology itself is nonsense when it takes itself to seriously. refreshing.
An innocent question: if the incompleteness theorem refers to the existence of ''valid'' propositions whose truth or falsity is impossible to prove within a formal system of axioms, then, on what basis is it assumed that the proposition was considered ''valid'' to start with (if it was)? Or, is ''valid'' above = merely ''formally constructable''?
I worked for IBM in the 1980's on Artificial Neural Network systems and encountered the dog-cat problem the guest describes here (only they were printed numbers with voids in the images such that 0 and 8 could not be distinguished reliably). The ANN would classify those as patterns as rejects, i.e. unreadable or unrecognizable, based on the difference between the highest class score and the next highest against a confidence level, which was derived statistically across all patterns. But rather than being attributed to perception as the guest suggests, it really was the result of two factors: too low an optical resolution and an insufficiently precise definition what would constitute an "8" and a "0" in the case where the "8" was degraded and the training patterns were manually tagged based on those low resolution images. What needed to be done was the training patterns should have been tagged by the printer program that generated them. As to cats vs dogs, it goes back to precisely defining what a cat is vs what a dog is ontologically. For that you'd have to look at every possible variation of what each is, their similarities and differences in order to derive a precise definition, which is what the ANN is attempting to do. So it's a chicken-egg problem, but as mis-tagged images are discovered in the training set that cause errors in the test set, then those mis-tagged images can be corrected, further refining the definitions for each. This kind of corrective (and adaptive) feedback is what makes these systems work.
You’ve missed the point entirely.
I'm surprised Frenkel is not familiar with the framework of PAC learning, it seems like the perfect answer to his question about uncertainity regarding labels.
Mr. Lex, for every question you have that bothers me, you have ten that I find fascinating.
Great clip, but left the Göedel incompleteness theorem explanation Incomplete!
13:00 LEX: I think your comments here really convey (in a wonderful way) how you view such matters. I approach such topics with more caution than yourself, but this really does convey your Intent regarding AI and related matters.
I like the late Dr. Van Till's view of how that its necessary to start with "the very first principle" of our creator's existence. THEN we can make sense of everything in virtue of that divine ultimacy of reality. Top down rather than bottom up. It acknowledges the need for divine revelation, not only in order for us to know that the ultimate nature of reality is divine, but so we can have intelligibility for facts generally, regarding things we can see and touch.
Except it's been understood for a while now that there doesn't need to be a creator in the universe we inhabit. It's an established part of physics that structure can emerge from "nothing" due to spontaneous symmetry breaking.
"Structure" referring the laws of the universe that give rise to what we see as material reality. It's understood that as energy levels lower, the many symmetries of the universe break down in ways that create more and more differentiated structure as opposed to the extremely high symmetry of nothingness (everything being the same under all possible transformations.
This isn't just speculation. It's proven that the most fundamental laws of physics, the conservation laws, are all just emergent expressions of symmetries in the universe. They did not have to be decided on or written down by a creator. They just emerge due to the "shape" of reality.
Even the most fundamental "things" in reality, quantum fields, emerge spontaneously as energy decreases. The four fundamental forces emerge spontaneously from one as the universe cools. At the beginning of the big bang, there was a single fundamental force, and as the universe cooled, broke into 4 different forces. More symmetry to less symmetry. More similarity in the universe to less, which is to say more structure emerging from less. No creator needed. The big bang itself is very likely the breaking of time symmetry.
The existence of things like "charge" is due to spontaneous symmetry breaks too.
Like liquid water being the same/symmetric from all sides spontaneously turning into a snow flake as energy lowers and suddenly being symmetric from only 6 sides. That's a rough analogy of a symmetry break. The structure of the snowflake emerged from the less structured water completely passively due to the symmetries of liquid water breaking as energy falls.
None of this proves there is no God, but it DOES mean that God isn't necessary or inevitable to explain the universe existing. Structure can emerge from where there was no structure completely passively as things go from high energy to lower energy, which is just to say time passes.
Great conversation. An interesting detail: Mr. Fridman wears a jacket and tie, while Mr. Frenkel wears neither jacket nor tie. I like Mr. Frenkel's look.
I've been saying Goedel's name wrong my whole life
Axioms are the dials you tweak to reveal new math. In the case of Euclid, it's clear now that the deformation of the space/plane can lead to new latent spaces of differing, but equally valid math. What's more interesting to me, is the meta-view of why some of these axioms can unveil new useful math, but others are absurdities or yield no value. What system or model manages the distribution of the valid and useful axioms, and can that system be tapped to learn _what we should learn about?_
4:00 Did he mean to say Alan Moore? Or did Alan Watts also say who watches the watcher?
The weirdest thing is the proof isn’t even that complicated once you develop all the foundational theory of logical deduction. Since there are a finite number of rules for deduction you can assign an integer code to every rule. I don’t remember exactly, but you take a self-referential statement like “this statement is false” and turn it into integer code. It becomes a “theorem” about integers, but the self referential nature can be used to show that if a proof of the special theorem constructed exists, then the statement must be false which is a contradiction. The conclusion must be that no proof of the constructed theorem exists. A proof that the theorem is false also doesn’t exist following the same method, which is the part thats really hard to wrap your head around.
The key was indeed the Godel numbering system...
Nagel and Newman have a lot to answer for.
I freaking love this guy
Euclidian geometry refers to the uniqueness of "STRAIGHT line" and not just a line as you have stated in the video.
“No system can fully understand its own deficiencies"
Math is intriguing...once you know how it's used to understand the universe, our world, and ultimately...ourselves.
Do logical inferences require axioms? If not it seems it is the most foundational subject. What are arguments against a logic platonic idealism, if they underlie math?
Do logically valid inferences "require axioms"? No. Inferences are not proofs. Do proofs "need axioms"? No. Gentzen-style calculi don't.
complexity emerges from simplicity due to entropy. Simple things have simple structures of information, which have a higher degree of freedom of arrangement, which creates complexity in arrangement. From complexity emerges (systemic) simplicity, because complex systems have a lower degree of freedom of arrangement.
Due to decoherence…
Damn, that's a good explanation.
14:25 I’ve seen the exact dress in person. It’s absolutely blue & black. My theory is that women often sift through dark closets, where a dim lit white & gold dress appears like a sunlit blue & black dress.
He describes a line as an example of an axiom. He should have started with "point" as that is the ultimate axiom.
Einstein did think highly of Godel but Einstein was at the Institute of Advanced Studies years before Godel did.
This guy is a god. Love his appearances on Numberphile
Makes sense
Yes, but try to redo planar geometry without the fifth postulate. The fifth postulate is obviously not true for spherical surfaces. So it obviously shouldn't be postulated for spherical surfaces.