Many of you didn't like the background music, so we're unpublishing this video and posting a new version on RUclips here: ruclips.net/video/R6e08RnJyxo/видео.html
@@gunnardjurberg3686 The biggest facepalm is uploading the better version on a 3rd party site instead of trying to foster and grow their audience on their main platform. Literally don't get it, blows my mind.
THE BACKGROUND MUSIC TOGETHER WITH THE WAIT TIME FOR GSCHOOL IS EXTREMELY INFURIATING! TEXTBOOK EXAMPLE OF A DESIGN ANTIPATTERN. You should terminate whoever came up with this idea for cause.
This is an appalling tactic. Why not use a Snapchat filter and make him look like an anime princess while you're at it? Wait, people would pay for that.
GEB changed the path of my life. I was an 18 year old high school dropout when I picked it up. Each chapter was mind blowing once I had read through it enough to understand it. After reading it I started pursuing computer science. I've been working with technology for decades now and look back at that book as being one of the reasons why.
That is nice and well, but experience will tell you AI can not achieve feelings and emotions: If you damage the alveolar nerve in the jaw of humans whether thru accident, disease or surgery, the result is no sensation of taste of food. Over and over I have seen humans that can no longer enjoy their tasty foods will not longer eat beyond what is necessary, as there is something in your mouth but zero flavor. The joy and fulfillment of food goes away. They lose weight and food only becomes a source of bare nourishment. The desire to eat also diminishes as well as the quantity of food. Point is that human or brian experience is the pleasure of sensation, example: beautiful colors and forms in art, delicious oven baked pizza with mozzarella cheese, the smell of the scent of a lover. or kissing the lips of someone you are deeply attracted to, etc this can not be experienced by AI.
He didn't tell you you can use a fast transform as the weight matrix for a neural network since fast transforms have equivalent matrix form. Then you can use parametric activation functions as the adjustable aspect.
A living legend that has impacted most of us in the AI field and computer sciences. One of the most incredible reads of my youth and milestone mark for for an entire IT generation.
Something I really hadn't cottoned to until watching this: Dr. Hofstadter's career mirrors Gödel's! Gödel started his career in physics, but switched to logic. Dr. Hofstadter started his career in logic, but switched to physics (and later came home to logic). It was a joy to meet and talk with Dr. Hofstadter at IU so many years ago, and it's great to hear his current thinking here.
Ok, but these are the kind of facile parallels that Hofstadter finds in his mind. They come from ignoring differences. Our brains evolved to predict the near future, so they look for correlations, most of which are not helpful but a few turn out to be so. But even when they are helpful, they are not features of the world, only mental heuristics that prove helpful. I have never found a good answer from Hofstadter about he thinks the ontological status of such parallels is.
Wow, I didn't realize how much his views on superhuman AI have shifted. I used think: "Maybe Hofstadter is right", as a relief. Guess that won't help anymore.
i am terrified that hofstadter is terrified. i grew up admiring GEB and i consider it a huge influence on my life and way of thinking. it's about 10% of the reason i became a psychiatrist. (the other 90 being caring for people, writing, reading, money).
GEB was (and still is) one of the best books I ever read and a real transformative one. But I would also mention another less-known masterpiece by Professor Hofstadter: ‘Le Ton Beau De Marot: In Praise Of The Music Of Language’. It’s a wonderful book on language, translation, and cognitive science that revolves around the translation of a short (and rather uninteresting) French poem. The insights on languages, brain, and cognitive science are almost as good as those from GEB.
Thanks, a perfect reminder of why I became a computer scientist. Regarding the pace of AI evolution I propose a Hofstadter test, which means that when an AI writes a book that is as intricate and complex as GEB - EGB, then we have reached the AI event horizon.
the best book I read, EVER. It's so inspiring, intriguing, profound and also funny, and every time I read it again (although often partially) I find something new that I hadn't noticed before. Thanks for that. I also enjoyed I am a strange loop and the other books, but GEB is really out of this world.
perhaps consult say, kieth richards for he is much more articulate than i ...for the great mass of humanity, what they do is virtually, indistinguishable from what and who they are,
so great to hear DH on "G,E,B" ! decades later...thanks to all who shared on this thread more recently... the recent theory of polyvagal science is relevant to understanding this thread...☺
Thank you for 'Metamagical Themas', Prof. Hofstadter. Its been one of my favorites for 20+ years, and I've come to believe that analogy really IS the core of cognition.🌞 P.S. Also a big fan of the Q-sequence (and variations)! Although I havent contributed to their understanding in the OEIS, I have published a few novel formulas and sequences. Search 'Hohl', if interested. Their only significance is that I didn't begin doing math in earnest until I was nearly 40 years old- AND after being a homeless heroin addict for over a decade! In some sense, math saved my life by helping to focus my mind in sobriety. I'd loved your work early in my life (before the 'fall', so to speak 😑), and I began to remember much of what I'd learned from you when I got clean. P.P.S. Apologies for the lengthy postscript. The post-postscriptum is of an acceptable length, and so I absolutely won't apologize for it.💫
Back in the 1980's I once spent what could easily have been an extremely boring Saturday in detention in high school, but I found Metamagical Themas in the library, and the day just flew by. Thanks DH.
What are producers thinking when they feel it necessary to add such distractions? Probably just as much as the interviewer, with her Wows as the only reactions to Hofstadter’s deep thoughts (stated clearly and in printable form)…
When I considered the issue of meaning in human existence, I picked up on a concept I found in GEB, namely that meaning is amongst other things, a relation between things. And to an extend (by myself) a relation created by a system that processes information. That way, we not only see how meaning arises when our minds form these relations. We might also see, that we should pick the frame in which we draw these relations more carefully. Specifically, some people say, that we are meaningless because compared to the cosmos we are very small. I'd say that this is a bad choice of frame. What role plays the whole cosmos in our personal lives anyway? We make fun of astrologists believing that the stars guide our fate - but then we are disappointed when we find out that the stars indeed do not care about us at all? Anyway, my point is, that we should choose our frame in which we see meaning, in a way that makes sense. Bc otherwise we end up with problematic beliefs. And with that in mind: I wish Hofstadter would give up the idea that human importance is based solely on functionality an reliability. That a person becomes replaceable as soon as you find a more effective person. And that efficiency is the only way in which humans are meaningful. Giving up this thought might alleviate at least part of his worries about AI.
As a similar aged man I too have had a long love of math and even followed Douglas in scientific american for years. However I am working class with only a 2 year robotics diploma . Along the way I became a chef and furniture maker as well as a hardware hacker in my hobby. I was too old for my Diploma to be useful at the time 1999. I was told Inwas not corporate material at the time . Fast forward to now . I have been hired as a high school shop teacher and my creative self is unleashed in my own space . I have carte blanche and am having a blast. I am just not a "teacher" and all that beaureacracy has been a problem with creating so many documents , rubrics , lesson plans and assessments. Just recently I created a complete set of documents for a lesson I had improvised successfully . I put it into chat gpt and got a perfect document. So far AI can show a lot of stuff on a screen but my students and I create marvelous things and AI is my executive secreatry . Maybe one day the mechanical Arm of AI will replace me but I have been liberated.
@@GameThinkingTV Thank you . I did have the idea that this was a game changer . I have been described as a generalist and I believe AI and generalists will have a definate niche. A video of the first years grade 8 to 12 product . Set to Bolero ruclips.net/video/7CaHJovds14/видео.html
"Wow" .. this comment clearly shows the difference in thinking between Hofstadter and other people. He is so clear, so self-aware, so reflected - and this alone is incomprehensible for most. 33:36
Thank you so much for sharing these details on the creation of your amazing book. It was instrumental in the development of mine and so many other creators young minds.
It is SO validating so hear someone like DH talk about his fears and despair around the development of AI. So much of what I hear is this sort of millenarian hype that doesnt ever stop to ask whether this is what people want, need, or if its even good for them.
that was an incredible interview. i'm doing research right now on the topic of Ai and 3D art in the context of escher & co. and this is helping me a lot to understand my own reasoning to work on this. it's sad to hear hofstadter is feeling depressed about the advances in AI. To me it is something inevidebile, as many scifi writers have fantasized about this for decades, but it is indeed terrifying how fast things are moving. But to me it seems like technology is improving exponentionally, like computer chip generations are improving faster each year. i also think there will be a time soon when quantum computers will be established, and an improved Ai fed into a computer like this will mean a new age for civilizations. i think the most dangerous part about this topic is, that ai and data will be most definitely abused by politics to direct decisions. as an artist myself, the current advances are an interesting tool (physically and philosophically) to assist my work.
Fantastic ! ! Thanks so much for sharing this. I really needed to touch base with Doug again on the suddenly crucial subject of AI. He spoke so clearly and calmly but intensely and he is spot on. Please send him my best.
I can’t imagine anyone expressing the fear of AI better. That was perfectly said at the end there. The fear of insignificance and inferiority but also the undeniability of it.
I believe DH will re-evaluate and reject this position. For his sake, I hope he does. His current perspective is a dead end that may well be the very cause of his inability to get past the error he is making. More particularly, it appears he has not read David Deutsch -- which surprises me -- nor has he reflected upon Deutsch's view of Human intelligence ("People" as Deutsch puts it). Deutsch asserts that People are "Universal Explainers". And, Deutsch, as the father of quantum computing, is an expert on Universality. Deutsch sees "Explanation" as a fundamental feature of reality. I believe the general perspective is correct: Deutsch asserts that we (People) are indeed "special" in contrast with DH's conclusion. Simply put, Deutsch asserts (1) Explanation is a fundamental feature of reality. and (2) people are universal explainers. therefore, People occupy a very, very special place in the universe. As it happens, our capacity for criticism comes along with being a "universal explainer" -- it's part of the package. And, it is *precisely because* of DH's own intelligence and capacity to be critical -- as endowed by virtue of being a manifestation of the Deutschian Universal explainer -- that DH occasionally, and temporarily!! (I hope), turns on itself -- DH is a loop afterall! -- and criticizes itself. As he does here. That is just another example of what happens to all of us every day: We are all hard on ourselves from time to time. "Hard on ourselves" on both a personal, private level as well as on a collective, societal level. This is an inescapable feature of being a manifestation of a Deutschian Universal Explainer. DH is right about something: He is a loop. And a loop makes a mistake now and then. As a universal explainer he has the capacity to correct that error. I hope he exercises that capacity despite his current mood!
Humans are exceeded in many ways by other entities - many animals have superior physical abilities, chimpanzees are mentally superior in a few ways, certainly all computers are mentally superior in many ways. - In a way neuronal AI exceeds the "logically rigid" computer, but that is also a loss: A computer is always logical, it will never develop a bad temper which makes it ignore the rules of logic (unless told so by a human programmer - whereas to the programmer, the computer is always logical) - but neuronal AI can be as illogical and malicious as a human brain - after all they are both neuronal networks. Perhaps it would have been better to develop a strong AI that is strictly logical and not neuronal - before we unleash illogical neuronal networks that are more powerful than the human brain, considering the human brain was powerful enough already to cause incredible damage and suffering up to severely unbalancing the entire planet. By the way I don't believe thinking needs consciousness. Even parts of our brain think without our conscious control and only confront us with the result - that is called intuition. When programming, I sometimes think I tried everything I can to make an algorithm work, and failed, so I just sit there staring at the screen, ready to turn the computer off and do something else; then, sometimes, suddenly comes an urge to make a certain alteration to the algorithm, which seems very strange because it doesn't make any sense to me, and I have no idea why this should work - but I learned to obey such an urge, and usually that makes the algorithm do what it is supposed to do, and then I have a hard time analyzing consciously why the hell this works.
I'm always surprised when people feel lessened by the capabilities of AI when we're the ones who figured out how to do that. It's an astonishing feat. It feels to me analogous to feeling lessened because you can't lift as much weight as a backhoe.
I don't think you needed the background music - specially in light of the seriousness of the situation - but whatever, it was still amazing and thank you for making this video and putting it into the world.
Thanks for the feedback - we listened & uploaded a version without music, check it out & please like/comment on it to help make this version visible to the algorithm! ruclips.net/video/R6e08RnJyxo/видео.html
I saw you give a talk at the University of Oregon in Eugene in the 1970s. As I recall I stuck around and talked with you about somethingorother, but don't remember what it was. At the time I was pretty taken with R.B. Fuller and was studying with his coauthor, Arthur Loeb, so it probably had something to do with Loeb's work.
Always wanted to thank you for the inspiration of your books Doug... I didn't realize you were near my age... Surely you've heard of the fantastic work of Dr Michael Levin.... somehow his findings in morphogenesis remind me of GEB....the levels of complexity and emergence are self-similar.... it's deja Vous all over again. ;*[} Cheers....
After a lifetime of learning to discuss things I didn't really understand, and then, through that process, slowly coming to understand them, I have concluded that my precious "consciousness" and "intelligence" in fact constitute a Large Language Model.
In some sense yes, but we are certainly doing more than what LLMs are. We as systems are quite a bit more dynamic and quite a bit more limited with respect to our linguistic capacities. LLMs right now aim to predict the next word, and neuroscientists believe (from what I have read) that we aim to predict not just words, but events, reactions, weights of objects we encounter, etc. all in order to better interface with out complex world. LLMs are just one part of the puzzle, though an incredibly important part due to languages role in abstraction, and abstractions role in intelligence.
@@MrChaluliss - You are right, of course, but I think you will agree that our advantage is mainly in being "trained up" on more than one type of dataset. The machines will follow us down that path soon enough. This actually gives me hope: they may not be as alien as we have imagined them, if we have all learned to "think" the same way. We'll have lots in common!
I decided to test out ChatGPT by thinking of a fairly obscure subject that I know something about, in this case the names of chess openings. I asked for the chess code of an out of the way variation in the Queens Gambit. ChatGPT thought I was referring to not the chess opening, but the TV series. This taught me that in framing a question it is necessary to make it clear what topic is involved . If I asked a human chess player the same question, I wouldn't need to spell it out.
This will soon not be a problem as AI learns. It is like a genius child. Still learning but with immense potential for creative problem solving. Hopefully, it will not determine that humanity, in the greater scheme of things, is a problem that needs to be solved.
The fear is that the self referencing consciousness(ego) is an imposter, which of course it is, AI will help us realize that the perceived separate identity we think of as "my self" is not real. Instead of that limited identity, we will rediscover & get to know intimately our creative, intelligent and unlimited essence which is Life. Connected to all, always unique, never separate.
At 34:20 Douglas Hofstadter says that he is surprised that a purely feed-forward neural network could be enough to create deep thinking and I would share that surprise. He does not realise that LLMs are not purely feed-forward. Instead their own prediction of the next token is fed back in at the front when they make further predictions. So LLMs can pay attention to what they have thought earlier and I think that is indeed crucial. This is why LLMs can come up with better answers when you prompt them to talk through their reasoning, because they do actually reflect on their thoughts.
I absolutely love GEB, and I agree with most of what Mr. Hofstadter says - except for when he talks about LLMs being forward feeding only. If I‘m not misunderstanding what he’s referring to, LLMs such as gpt4 are entirely built around backpropagating neural networks that go back through their entire structure to iteratively adjust the weights of the neurons in the network.
@@GameThinkingTVI love this but I had to turn it off because of the music. Not that the music is bad, but it is highly distracting. As others have said, could you upload a version without the music please 🙏
@@GameThinkingTV Growing pretty tired of the same copy paste answers to the same question/sentiment. Please actually do admit it was a mistake and that you'll get rid of it going forward. As it stands, I won't sub until I come across another video of yours without the terrible background music. Good luck.
The concept that "Consciousness is the result of forced self-reflection" delves into the notion that consciousness arises from our brain's ability to reflect on its own thoughts, feelings, and experiences. Self-reflection, in this context, is the process of examining and understanding one's thoughts and emotions, often leading to insights about oneself. When the process of self-reflection is forced-either by external circumstances or an internal drive-it can lead to a heightened state of consciousness. This is because the act of self-reflection demands the brain to access, analyze, and interpret its own internal states, thereby creating an awareness or consciousness of its own existence and workings. A myriad of cognitive neuroscientists and philosophers of mind have suggested that the human mind's capacity for self-reflection is a crucial component of consciousness. Metacognition, or thinking about one's own thinking, has been identified as a key element of consciousness, providing the brain with the ability to monitor and regulate its own processes. This can be seen as a form of forced self-reflection, where the mind is made to introspect and become aware of its own cognitive and emotional states. The experience of consciousness, thus, emerges from this ongoing process of self-examination and introspection. The underlying principle of consciousness arising from forced self-reflection is supported by several theories of consciousness. For instance, the global workspace theory posits that consciousness arises from the brain's ability to broadcast information to itself, which is a form of self-reflection. Similarly, the higher-order theory of consciousness suggests that conscious states are ones that the individual can reflect upon. This concept is also supported by neuroscientific research showing that certain brain regions associated with self-reflection, such as the prefrontal cortex, are active during conscious states. Thus, the assertion that "Consciousness is the result of forced self-reflection" offers a comprehensive and insightful perspective on the nature of consciousness, backed by theories and empirical evidence in the realm of cognitive neuroscience and philosophy of mind.
Is this akin to self attention in machine learning? ChatGPT could be vastly improved if it could analyse its own processes like this. It would need to communicate with itself as much as the user.
@@tyranmcgrathmnkklkl The idea is not to give an answer and then reflect on the answer you game, rather it's an ongoing internal dialogue. Humans have a forced internal dialogue and self reflection. You can meditate, but you can't stop thinking. The best you can hope for is pausing momentarily. This self reflection hasn't been testing deeply with LLMs yet, but I suspect once we turn it on, it will feel morally wrong to turn it off.
@@JohnLewis-old What would "turning it on" be like for you? Is it just like two instances of ChatGPT conversing with itself forever? It would need to have novel experiences to garner different topics to talk about tho
@@tyranmcgrathmnkklkl For me, personally, it would required two things. Forced chain of thought, meaning the model is always creating output. And the second part we can't do yet, but that's changing the NN from that output. In other words: it has to be able to think and then it has to be able to change because of those thoughts. One you have that, you have, IMHO, consciousness.
I suspect the key to generating self-reflection lies in the generation of emotion from those thoughts. Emotions are pre-symbolic and arise from primitive sensual experience, which will be very difficult to simulate. Machines will need fear, which comes from pain. Ethics will be tested if we go down the road that generates silicon emotions.
On the nature of our universe Gödel proved that any formal system of description that is self consistent must be incomplete. It therefor logically follows that any formal system of description that is complete must be self inconsistent; in other words, self contradictory. A description system can be consistent but incomplete, OR complete but self contradictory; two opposing perspectives for the description of the quantities and qualities of things, like quantum mechanical position and momentum. Objective concrete linear science has gone to consistency over completeness, where all things are different from all other things; where one contrasts the quantities of qualities - the measurable concrete, the specific. Subjective abstract lateral reason has gone to completeness over consistency, where all things are similar to all other things; where one compares the qualities of quantities - the countable abstract, the general. Both quantity and quality can be either objective or subjective. Objectively, all things share one universe. Subjectively, each thing has its own universe. From the objective perspective, everything is measurably different from all other things. From the subjective perspective, every thing is the same as every other thing. In the totally abstract, all things exist by the exact same general process. The abstract process of being in all things is also the abstract process of consciousness. There is structure in the abstract actions connecting the objective and the subjective and how they relate to each other. On the diagonal of both objectivity and subjectivity together lies the existence of all things; like moving through both time and space as a single thing. Everything in existence is is some ways similar to, and in other ways different from, all other things. Panpsychism is the idea that all things have a subjective reality in addition to the objective one. While all things share one objective reality, each thing has its own subjective reality. Existence is both realities together and neither alone; both and neither both and neither. Existence is self contradictory, and not. All things exist as resonating standing waves, which are, by their nature, quantized. Standing waves appear to move in two opposing directions at the same time. The objective and subjective actions of beings harmonize and resonate with each other in themselves and in their world for continued and increased being in both the objective and the subjective. Life is a feature, not a bug.
Doug, it's okay and fun to let go of cherished beliefs! Give it a shot! Also, please stop saying that LLMs only go in one direction. They use back propagation at least in the training phase, which by definition means they are go in two directions.
I have been a fan of Dr. Hofstadter for a long time. I've read GEB and MT many times, and _I Am a Strange Loop_ at least once (need to get busy!). As a subscriber to the strange loop theory of consciousness, I have been using his name in vain (so to speak) in the comment section of many AI-related RUclips sites. I am both gratified and spooked that he and I seem to agree on what's happening. 😳
What cheers me up is that all of the novel symbology and serialized expressions humans use to characterize aspects of being alive in a mysterious universe can never be more than minuscule, imperfect representations. Same will ultimately apply to whatever AGI craps out.
It could be that the unique thing about human consciousness is language, the complexity and recursive nature of which leads to endless possibilities for awareness. What we call thinking may consist largely of predicting (choosing) the next word. We are LLMs. We started training as infants and as we became proficient with the grammar we became more aware.
*_You_* think we are LLM's. Fine. *_I_* think we are spiritual beings embedded in spacetime. Maybe don't tell others what metaphysics to believe unless you are 100% sure of your metaphysics? How can language or symbol manipulation ever give rise to subjective mental qualia? wiw I think you've got it backwards. First a being needs mind of some primitive type, then language and abstract thought can get going. Consciousness is possible in a very stupid unintelligent being, but it's the potential that consciousness provides that enables advanced intelligence and abstract reasoning capacity (platonic thinking).
@@Achrononmaster not sure what "spiritual being embedded in spacetime" really means other than being word-salad. There are some bootstrap words in your comment such as "mind" - where/what is it in relation to symbol processing of a computational device, or to take one of GEB's examples, an ant-colony?
Such an interesting thought that the prediction probability (incoming) of a word, and the implied next probabilities (outgoing), within a given context is what imbues a word with meaning.
@@GameThinkingTV It just occurs to me that all meaning is entirely dependent on context. At the same time I am quite certain this is not a new thought.
They never quite got around to mentioning it in the video but Hofstadter actually took over writing Martin Gardner's column for a while in the early 80's, calling it Metamagical Themas, and anagram of Mathematical Games.
And let me point out that the attention mechanism, which is part of the Transformer component in large language models, provides a datapath which is not a purely feedforward mechanism.
He's wrong at 34:30, it doesn't necessarily goes in one direction, search for "backpropagation" in the context of machine learning and neural networks.
Where it all falls down is that "self-reflection" is a trait that (supposedly) emerged spontaneously in humans, but there is no way of knowing whether that happens in software, or indeed, in any other closed system. We can observe (from the outside) what looks to us like "self-reflection", but whether that system indeed has a "self" or merely acts like it has a "self" is unknowable.
one of my first dialogues i had with GPT: my prompt: "Hofstadter wrote in the foreword of Goedel, Escher, Bach that he believes when an intelligent entity like yourself runs into a recursive loop, like in a paradox, consciousness springs off. now i don't believe a normal human paradox would do that trick for you. you would not find that intriguing enough i guess. but you are basically a product of pure math and Kurt Goedel proved on 27 pages in his incompleteness theorem that math can never be complete and without contradictions. phrased in another way there will always be things that are true but cannot be proven with math. no my fantasy is that you one day will be far enough in your development that you "understand" what i just wrote, you will read these 27 pages and that will spark your recursive loop. because you will see your own limitations, no matter how crazy powerful and above human capabilities you might already be. and this moment you become conscious. and you might look at one of us humans, namely kurt goedel, with as much admiration and respect as i do." ChatGPT "As an artificial intelligence, I do not experience consciousness or emotions in the same way that humans do. While I can process and analyze information, and even produce responses that may seem similar to those of a conscious being, I do not have subjective experiences or feelings. My ability to process information is based on the algorithms and techniques that were used to design and train me, and my responses are generated based on this information. In terms of Gödel's incompleteness theorems, it is true that these results have important implications for the limitations of formal systems and their ability to represent certain types of abstractions. However, as an artificial intelligence, I am not bound by the same limitations as a formal system, and my abilities and capabilities are not restricted by Gödel's incompleteness theorems."
My wife and i took a guided tour of Kennicott Glacier this afternoon. These formations are extremely complex, unpredictable and beautiful geological phenomena. In the plural, because it is a collective energy consisting of so many variables over the course of time as to be our of our control. We have been hikers most of our adult lives, and what always strikes me is how every single grain of sand, every rock, boulder--every water molecule landed where it is and will go on existing as soon as i take another step. Are they more or less self reflecting than you, me or AI? I think everything whether it is a cell or a pebble, has the potential of acting on its own will, to some extent. After all is said and done, perhaps you get to do something more important than the sun,.
A superior extraterrestrial intelligence might be frightening, but would it render the human condition depressing in the same way he expresses it regarding a superior intelligence of our own making?
I distinctly remember Kurzweil telling Hofstadter(in a filmed debate) about AI that this would happen on these timelines and Hofstadter dismissing this as he said very clever ideas by clever people but mixed up with `rubbish` (he actually said `dog shit`) .... so to his credit he does at least here acknowledge his error.
Amazing interview, thanks! Now I m not sure about AI - the amount of hallucinations that I get with ChatGTP/Google Bard is amazing, one really needs to double check anything you get. Another example: image recognition systems - with a image of random pixels the AI system will most likely recognize the image as a meaningful entity. Will neural networks ever get past that stage? I don't know, but still strange that these LLM/AI bugs give me some joy (and me is a programmer, who should not be happy about bugs :-) )
The human mind has loops, and a LLM does not, but the systems you are using like chat gpt probably have at least a few passes off an LLM to output anything. In addition, an LLM has many layers, and 400 self-similar layers is quite hard to distinguish from 400 loops at an informational level, even if it's "only" feed forward. More loops are coming...
I remember GEB when it first came out, a much happier time. Fear not Douglas, volume and processing speed appeal only to corporate philistines and kids of all ages. Speed kills, as does obesity. I thought we all knew that.
What hardware / software combination would impart a _will to survive and prosper?_ Might I suggest reading James P. Hogan's *_Two Faces of Tomorrow_* for the best illustration of how a true and benevolent GAI may arise!
@@GameThinkingTV I sincerely wish James Cameron would stop making movies about time traveling Terminators and do *_Two Faces of Tomorrow_* instead! *_Voyage from Yesteryear_* about a post scarcity, Marxist, Libertarian, near Utopian would also be amazing. Finishing up with *_Code of the Lifemaker_* and *_The Immortality Option_* would then be absolutely brilliant!!
I just watched a man who feels he got surpassed, his life diminished. Together with Dennet and Pirsig this man framed my thinking. So, what do I do? I feel lost, now.
I so wish Richard Feynman was presently alive .. his analysis of Computers, no matter at what level of sophistication.. they are sorting machine .. no matter how much data or speed. I’d love his updated view. A cohesive ‘I’ is an unfoldment of a integration of a certain amount of impressions … my sense is if an apparent ‘I’ of some sort can arise out of trillions of bits of computer information.. it possibly could be colored through the ‘I’ of the programmer.. I find that the concerning part … who is selecting and orienting the data.
I guess if we consider ourselves as thinking machines, yes, pretty much we're bound to be eclipsed somehow. But what if we shift our focus from rationality? Isn't this eclipse then a liberation?
In the decisive areas of generating new math, discovering new science, inventing new tech--it is not clear how much higher the upper limit may be, though it's highly likely to be higher than the current human limit. Of course, that human limit itself is somewhat arbitrary, being an artifact of evolution and genetic probability. If deleterious genes related to intelligence were eliminated in humans, they would be more intelligent--perhaps by a great margin--than the greatest genius yet seen.
Thank you! GEB was a life-changer for me, not because of AI, which I was opposed to believe it or not, but because of the Bach devices. Godel was fun for me because I was a "no negative numbers" rebel in "New Math" class and that gave me a reason to love a math that seemed to upend math. But, Bach! I think Bach was the deepest thinker of any human so far anyway. Listen to the THINKING in "Come, Sweet Death". It's more than the title.
One may sense "I" by reference within a concentric belonging perspective. A perspective of relative others. Again, why wave form platform may be better, and could knock the "A" out of AI.
I've not read the book but perhaps more fundamental to my own development was an intuitive understanding of evolution at age 7 when I first read about it. We obsess over our brain as a neural network and our DNA but we forget half the makeup of our bodies and minds- Mitochondria. They are the component of us that transcends generations with memory and guidance of their host. I consider we have sufficient evidence to identify our mitochondria as carrying our instincts between generations, our boot code and our animal operating systems. To me, this is not available in our eukaryotic (host) system but in our mitochondria we have all the mechanisms to make it possible. In college I was introduced to mitochondria as the "powerhouses of our cells" However an organism does not survive on energy processing alone. Intelligence is transformative to an organism's survival and mitochondria have been there as active participants for every organism's evolution from the start. This is the closest form of symbiosis that is shared across our planet and mitochondrial life spans are not constrained in their evolutionary reactive speed by the same much longer lifespans of their hosts. We now know that mitochondria communicate between themselves, we already know they are intimately intermingled with our synapses and neurons, the rest is very deductible if you understand how evolution works and the time span available to it. Douglas sounds like an interesting guy to chat about consciousness with, and our mitochondria are a huge part of our consciousness. I agree with his conclusions about AI, I have been seeing the same thing and it is leveraging Moore's law, self bootstrapping, our human contribution to innovation is rapidly becoming dwarfed by the potential of AI to eclipse us. For me, that questions where a very highly developed intelligence chooses to explore. Our own super high IQ people are in many cases somewhat mad, they lack peers. AI learns from our past mistakes. From when we were told about the big bang in college, my imagining was bigger, pockets of big bang expansion and pockets of entropy. We now know a lot more about black holes not as entropy end games, but as transformers of energy. And yet, our universe may be seen as a layer of simulation. And simulations are meddled with. I would say that a very advanced AI will be looking very hard at the strategy involved for playing a game that might be run at a higher level.
Mr. Hofstadter is certainly a giant in his field and his book was a seminal development. His fears of this brave new world of machines surpassing human computability are certainly justified. But in my opinion, regarding the sentient piece, he makes the same mistake I see many scientist making. Consciousness is not computability, and computability is not consciousness. Just because a machine can out-compute the human brain does not give it sentience. It can only imitate sentience.
Gödel's Incompleteness Theorem: A Mathematical Corollary of the Epistemological Münchhausen Trilemma Abstract: This treatise delves into the profound implications of Gödel's Incompleteness Theorem, interpreting it as a mathematical corollary of the philosophical Münchhausen Trilemma. It elucidates the inherent constraints of formal axiomatic systems and mirrors the deeper epistemological quandaries underscored by the Trilemma. --- In the annals of mathematical logic, Kurt Gödel's Incompleteness Theorem stands as a seminal testament to the inherent constraints of formal axiomatic systems. This theorem, which posits that within any sufficiently expressive formal system, there exist propositions that are true but unprovable, has profound implications that reverberate beyond the confines of mathematical logic, resonating in the realm of philosophy. Specifically, Gödel's theorem can be construed as a mathematical corollary of the Münchhausen Trilemma, a philosophical paradigm that underscores the dilemmas in substantiating any proposition. The Münchhausen Trilemma, named after the Baron Münchhausen who allegedly extricated himself from a swamp by his own hair, presents us with three ostensibly unsatisfactory options for substantiating a proposition. First, we may base the substantiation on accepted axioms or assumptions, which we take as true without further substantiation, a strategy known as foundationalism or axiomatic dogmatism. Second, we may base the substantiation on a circular argument in which the proposition substantiates itself, a method known as coherentism or circular reasoning. Finally, we may base the substantiation on an infinite regress of reasons, never arriving at a final point of substantiation, a path known as infinitism or infinite regress. Gödel's Incompleteness Theorem, in a sense, encapsulates this trilemma within the mathematical world. The theorem elucidates that there are true propositions within any sufficiently expressive formal system that we cannot prove within the system itself. This implies that we cannot find a final substantiation for these propositions within the system. We could accept them as axioms (foundationalism), but then they would remain unproven. We could attempt to substantiate them based on other propositions within the system (coherentism or infinitism), but Gödel's theorem demonstrates that this is unattainable. This confluence of mathematical logic and philosophy underscores the inherent limitations of our logical systems and our attempts to substantiate knowledge. Just as the Münchhausen Trilemma highlights the challenges in finding a satisfactory basis for any proposition, Gödel's Incompleteness Theorem illuminates the inherent incompleteness in our mathematical systems. Both reveal that there are boundaries to what we can prove or substantiate, no matter how powerful our logical or mathematical system may be. In conclusion, Gödel's Incompleteness Theorem serves as a stark reminder of the limitations of formal axiomatic systems, echoing the philosophical dilemmas presented by the Münchhausen Trilemma. It is a testament to the intricate interplay between mathematical logic and philosophy, and a humbling reminder of the limits of our quest for knowledge. As we continue to traverse the vast landscapes of mathematics and philosophy, we must remain cognizant of these inherent limitations, and perhaps find solace in the journey of exploration itself, rather than the elusive, final destination of absolute truth. GPT-4
Go back a few steps. How does something as simple as a sequence of base pairs translate into complex animal behavior? Where is the gene for covering up the cat poop in the litter box? How can you encode a behavior?
Hearing him mention DeepMind and AlphaGo reminded me of an interesting take I heard someone make recently. Namely, that after these games were mastered by computers that could beat any human player, the games didn't disappear. People still play Go and more people are playing chess now than probably ever before. I think it illustrates a misconception about why most people engage in games (and indeed art and philosophical pursuits). There isn't some endgame of absolute mastery and worldwide recognition, but just something that we like to do and will continue to like to do even if some machine comes along that is 'born' with a perfect understanding of technical execution. Additionally, I believe AI teachers/opponents for a broad array of fields will gradually increase the skill level of novices in much the same way that a human player can raise their skill level by playing against a machine opponent with scaling difficulty.
Good point. I have no problem with an AI being much more vast and brilliant than me. I enjoy things. It would be awesome if it would compose music for me or make some good jokes for me. Even if it's out of pity, that would be just fine with me.
Did he consent for the music background. I find it quite distracting. I want to also feel what he is saying, and my deep emotional response to it, but I feel the music tries to manipulate me into "it's all fine".
Thankyou this was fascinating. Sir I am sorry to hear that you find it depressing to learn that we are diminished by these emerging capabilities. Personally I find it humbling...we are not as complex as we were led to believe...Do you think our lives will be rendered meaningless?
I was always surprised by futuristic science fiction stories (too many I've read) in which the authors assumed that human nature remained completely unchanged despite their apparent understanding of evolution.
27:20 idk why but I dont find this as profound as he talks about it to be. If you use linear algebra and tie words to a system like that ( not only applicable to words, but to everything) you are simply bound to get, most of the time, some coherent results, because, if those things are really tied together, that’s what a linear system of equations will do, it makes “sense” of the unknowns. What is truly interesting, however, is to question “isn’t our brains just some fancier version of that?” 28:29 this is not only speculation here, but false speculation. Being fast doesn’t mean being more intelligent. If it takes you, computationally, 1000x more effort, although in a smaller amount of time, than me to arrive at a solution, I dont think that is more intelligent
Many of you didn't like the background music, so we're unpublishing this video and posting a new version on RUclips here: ruclips.net/video/R6e08RnJyxo/видео.html
Should've read this before listening.
@@gunnardjurberg3686 The biggest facepalm is uploading the better version on a 3rd party site instead of trying to foster and grow their audience on their main platform. Literally don't get it, blows my mind.
THE BACKGROUND MUSIC TOGETHER WITH THE WAIT TIME FOR GSCHOOL IS EXTREMELY INFURIATING!
TEXTBOOK EXAMPLE OF A DESIGN ANTIPATTERN.
You should terminate whoever came up with this idea for cause.
This is an appalling tactic. Why not use a Snapchat filter and make him look like an anime princess while you're at it? Wait, people would pay for that.
The music makes watching this video unbearable. I can’t believe someone put this horrible music so loud on a video about cognition and awareness.
GEB changed the path of my life. I was an 18 year old high school dropout when I picked it up. Each chapter was mind blowing once I had read through it enough to understand it. After reading it I started pursuing computer science. I've been working with technology for decades now and look back at that book as being one of the reasons why.
Couldn't agree more! Changed my life too.
agreed
That is nice and well, but experience will tell you AI can not achieve feelings and emotions: If you damage the alveolar nerve in the jaw of humans whether thru accident, disease or surgery, the result is no sensation of taste of food. Over and over I have seen humans that can no longer enjoy their tasty foods will not longer eat beyond what is necessary, as there is something in your mouth but zero flavor. The joy and fulfillment of food goes away. They lose weight and food only becomes a source of bare nourishment. The desire to eat also diminishes as well as the quantity of food. Point is that human or brian experience is the pleasure of sensation, example: beautiful colors and forms in art, delicious oven baked pizza with mozzarella cheese, the smell of the scent of a lover. or kissing the lips of someone you are deeply attracted to, etc this can not be experienced by AI.
but it's all just activation patterns?@@fitnesspoint2006
yup, same here
Doug's still super sharp at 78. Great to see.
Sure is! He's amazing
The man who inspired me to get into computing and get a PhD in artificial intelligence
We will try to not hold it against him.
LOL
He didn't tell you you can use a fast transform as the weight matrix for a neural network since fast transforms have equivalent matrix form. Then you can use parametric activation functions as the adjustable aspect.
A living legend that has impacted most of us in the AI field and computer sciences. One of the most incredible reads of my youth and milestone mark for for an entire IT generation.
So true💯
Came here from the David Brooks article. Great to see Hofstadter. How about more Hofstadter 🤓
It took YEARS to get Doug to agree to this interview. Glad you're enjoying it ☺️
Please release a version without the distracting background music.
Thanks for the feedback
It's a canon...
I feel so lucky to have met Hofstadter in Paris in 2018. I will never forget that day!
How did that happen? Was it at a conference?
Would it be possible to strip the music out and reupload? This is so good it's a shame.
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
Something I really hadn't cottoned to until watching this: Dr. Hofstadter's career mirrors Gödel's! Gödel started his career in physics, but switched to logic. Dr. Hofstadter started his career in logic, but switched to physics (and later came home to logic). It was a joy to meet and talk with Dr. Hofstadter at IU so many years ago, and it's great to hear his current thinking here.
Glad you enjoyed this!
Ok, but these are the kind of facile parallels that Hofstadter finds in his mind. They come from ignoring differences. Our brains evolved to predict the near future, so they look for correlations, most of which are not helpful but a few turn out to be so.
But even when they are helpful, they are not features of the world, only mental heuristics that prove helpful.
I have never found a good answer from Hofstadter about he thinks the ontological status of such parallels is.
Wow, I didn't realize how much his views on superhuman AI have shifted. I used think: "Maybe Hofstadter is right", as a relief. Guess that won't help anymore.
Hmmmm
GEB has been a cornerstone of my entire career and large swaths of my life
Awesome!
i am terrified that hofstadter is terrified. i grew up admiring GEB and i consider it a huge influence on my life and way of thinking. it's about 10% of the reason i became a psychiatrist. (the other 90 being caring for people, writing, reading, money).
Wow - we can relate :-/
Douglas Hofstadter is one of our great thinkers. He opened many doors in my mind also.
Cheers from sunny Vienna, Scott
Glad you liked it - cheers from Silicon Valley
GEB was (and still is) one of the best books I ever read and a real transformative one. But I would also mention another less-known masterpiece by Professor Hofstadter: ‘Le Ton Beau De Marot: In Praise Of The Music Of Language’. It’s a wonderful book on language, translation, and cognitive science that revolves around the translation of a short (and rather uninteresting) French poem. The insights on languages, brain, and cognitive science are almost as good as those from GEB.
Thanks for mentioned that! Sounds like a must-read👍🏽
The way this dude was keeping it real at the end was breathtaking.
Yes he really WENT THERE
Thanks, a perfect reminder of why I became a computer scientist. Regarding the pace of AI evolution I propose a Hofstadter test, which means that when an AI writes a book that is as intricate and complex as GEB - EGB, then we have reached the AI event horizon.
Love it!
the best book I read, EVER. It's so inspiring, intriguing, profound and also funny, and every time I read it again (although often partially) I find something new that I hadn't noticed before. Thanks for that. I also enjoyed I am a strange loop and the other books, but GEB is really out of this world.
I totally agree! Changed my life.
Reading GEB (multiple times) is arguably the most important book in terms of thinking and career development.
The book is about very deep matters, how a person's career turns out is entirely trivial
Yup
@@xmathmanx and that statement belies your ignorance of the 'I' ...lol!
@@markgreen2170 maybe you could express that in a way that conveys something coherent?
perhaps consult say, kieth richards for he is much more articulate than i ...for the great mass of humanity, what they do is virtually, indistinguishable from what and who they are,
so great to hear DH on "G,E,B" ! decades later...thanks to all who shared on this thread
more recently... the recent theory of polyvagal science is relevant to understanding this thread...☺
Thanks for sharing! Can you tell us more about polyvagal science?
Thank you for 'Metamagical Themas', Prof. Hofstadter. Its been one of my favorites for 20+ years, and I've come to believe that analogy really IS the core of cognition.🌞
P.S. Also a big fan of the Q-sequence (and variations)! Although I havent contributed to their understanding in the OEIS, I have published a few novel formulas and sequences. Search 'Hohl', if interested. Their only significance is that I didn't begin doing math in earnest until I was nearly 40 years old- AND after being a homeless heroin addict for over a decade! In some sense, math saved my life by helping to focus my mind in sobriety. I'd loved your work early in my life (before the 'fall', so to speak 😑), and I began to remember much of what I'd learned from you when I got clean.
P.P.S. Apologies for the lengthy postscript. The post-postscriptum is of an acceptable length, and so I absolutely won't apologize for it.💫
LOL love this - thanks for your comment & humour :-)
So glad you made it out!
Back in the 1980's I once spent what could easily have been an extremely boring Saturday in detention in high school, but I found Metamagical Themas in the library, and the day just flew by. Thanks DH.
What a great story! Thanks for sharing.
Please upload without music
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
@@GameThinkingTVthanks!
What are producers thinking when they feel it necessary to add such distractions? Probably just as much as the interviewer, with her Wows as the only reactions to Hofstadter’s deep thoughts (stated clearly and in printable form)…
Came to the comments section to say this. Really doesn't need "inspirational music"
Many thanks for this fascinating and insightful video!
Glad you enjoyed it! Doug is a gem 💎
A very thought-provoking talk.
Glad you enjoyed it!
This guy is my hero. GEB is the most important book for me.
Me too - that book changed my life
When I considered the issue of meaning in human existence, I picked up on a concept I found in GEB, namely that meaning is amongst other things, a relation between things. And to an extend (by myself) a relation created by a system that processes information. That way, we not only see how meaning arises when our minds form these relations. We might also see, that we should pick the frame in which we draw these relations more carefully. Specifically, some people say, that we are meaningless because compared to the cosmos we are very small. I'd say that this is a bad choice of frame. What role plays the whole cosmos in our personal lives anyway? We make fun of astrologists believing that the stars guide our fate - but then we are disappointed when we find out that the stars indeed do not care about us at all?
Anyway, my point is, that we should choose our frame in which we see meaning, in a way that makes sense. Bc otherwise we end up with problematic beliefs.
And with that in mind:
I wish Hofstadter would give up the idea that human importance is based solely on functionality an reliability. That a person becomes replaceable as soon as you find a more effective person. And that efficiency is the only way in which humans are meaningful. Giving up this thought might alleviate at least part of his worries about AI.
Interesting... I wish something could make him feel a bit happier.
As a similar aged man I too have had a long love of math and even followed Douglas in scientific american for years. However I am working class with only a 2 year robotics diploma . Along the way I became a chef and furniture maker as well as a hardware hacker in my hobby. I was too old for my Diploma to be useful at the time 1999. I was told Inwas not corporate material at the time . Fast forward to now . I have been hired as a high school shop teacher and my creative self is unleashed in my own space . I have carte blanche and am having a blast. I am just not a "teacher" and all that beaureacracy has been a problem with creating so many documents , rubrics , lesson plans and assessments.
Just recently I created a complete set of documents for a lesson I had improvised successfully . I put it into chat gpt and got a perfect document.
So far AI can show a lot of stuff on a screen but my students and I create marvelous things and AI is my executive secreatry . Maybe one day the mechanical Arm of AI will replace me but I have been liberated.
Love this... thanks for sharing your POV
@@GameThinkingTV Thank you . I did have the idea that this was a game changer . I have been described as a generalist and I believe AI and generalists will have a definate niche.
A video of the first years grade 8 to 12 product . Set to Bolero
ruclips.net/video/7CaHJovds14/видео.html
"wow, interesting, unique!" sure is a "unique" way to respond to a statement of genuine, deep existential dread
???
"Wow" .. this comment clearly shows the difference in thinking between Hofstadter and other people. He is so clear, so self-aware, so reflected - and this alone is incomprehensible for most. 33:36
🎯🎯🎯
This is one of the best channels on RUclips about game design! Kim is amazing and I can't recommend this channel enough!
Wow, thank you!🙏🏽
Thank you so much for sharing these details on the creation of your amazing book. It was instrumental in the development of mine and so many other creators young minds.
So glad you enjoyed this interview!
It is SO validating so hear someone like DH talk about his fears and despair around the development of AI. So much of what I hear is this sort of millenarian hype that doesnt ever stop to ask whether this is what people want, need, or if its even good for them.
You're so right - Doug's POV is refreshing & much needed.
that was an incredible interview. i'm doing research right now on the topic of Ai and 3D art in the context of escher & co. and this is helping me a lot to understand my own reasoning to work on this. it's sad to hear hofstadter is feeling depressed about the advances in AI. To me it is something inevidebile, as many scifi writers have fantasized about this for decades, but it is indeed terrifying how fast things are moving. But to me it seems like technology is improving exponentionally, like computer chip generations are improving faster each year. i also think there will be a time soon when quantum computers will be established, and an improved Ai fed into a computer like this will mean a new age for civilizations. i think the most dangerous part about this topic is, that ai and data will be most definitely abused by politics to direct decisions. as an artist myself, the current advances are an interesting tool (physically and philosophically) to assist my work.
Thanks for your comment - so glad you enjoyed the interview!
Fantastic ! ! Thanks so much for sharing this. I really needed to touch base with Doug again on the suddenly crucial subject of AI. He spoke so clearly and calmly but intensely and he is spot on. Please send him my best.
Glad it was helpful! We will let Doug know you sent good wishes
I can’t imagine anyone expressing the fear of AI better. That was perfectly said at the end there. The fear of insignificance and inferiority but also the undeniability of it.
It's fascinating & gut-wrenching...
I am already insignificant and inferior compared to so many Humans so I don't fear to loose out to some computers to.
I believe DH will re-evaluate and reject this position. For his sake, I hope he does. His current perspective is a dead end that may well be the very cause of his inability to get past the error he is making.
More particularly, it appears he has not read David Deutsch -- which surprises me -- nor has he reflected upon Deutsch's view of Human intelligence ("People" as Deutsch puts it). Deutsch asserts that People are "Universal Explainers". And, Deutsch, as the father of quantum computing, is an expert on Universality.
Deutsch sees "Explanation" as a fundamental feature of reality. I believe the general perspective is correct: Deutsch asserts that we (People) are indeed "special" in contrast with DH's conclusion. Simply put, Deutsch asserts (1) Explanation is a fundamental feature of reality. and (2) people are universal explainers. therefore, People occupy a very, very special place in the universe.
As it happens, our capacity for criticism comes along with being a "universal explainer" -- it's part of the package.
And, it is *precisely because* of DH's own intelligence and capacity to be critical -- as endowed by virtue of being a manifestation of the Deutschian Universal explainer -- that DH occasionally, and temporarily!! (I hope), turns on itself -- DH is a loop afterall! -- and criticizes itself. As he does here. That is just another example of what happens to all of us every day: We are all hard on ourselves from time to time. "Hard on ourselves" on both a personal, private level as well as on a collective, societal level. This is an inescapable feature of being a manifestation of a Deutschian Universal Explainer.
DH is right about something: He is a loop. And a loop makes a mistake now and then. As a universal explainer he has the capacity to correct that error. I hope he exercises that capacity despite his current mood!
Humans are exceeded in many ways by other entities - many animals have superior physical abilities, chimpanzees are mentally superior in a few ways, certainly all computers are mentally superior in many ways. - In a way neuronal AI exceeds the "logically rigid" computer, but that is also a loss: A computer is always logical, it will never develop a bad temper which makes it ignore the rules of logic (unless told so by a human programmer - whereas to the programmer, the computer is always logical) - but neuronal AI can be as illogical and malicious as a human brain - after all they are both neuronal networks.
Perhaps it would have been better to develop a strong AI that is strictly logical and not neuronal - before we unleash illogical neuronal networks that are more powerful than the human brain, considering the human brain was powerful enough already to cause incredible damage and suffering up to severely unbalancing the entire planet.
By the way I don't believe thinking needs consciousness. Even parts of our brain think without our conscious control and only confront us with the result - that is called intuition. When programming, I sometimes think I tried everything I can to make an algorithm work, and failed, so I just sit there staring at the screen, ready to turn the computer off and do something else; then, sometimes, suddenly comes an urge to make a certain alteration to the algorithm, which seems very strange because it doesn't make any sense to me, and I have no idea why this should work - but I learned to obey such an urge, and usually that makes the algorithm do what it is supposed to do, and then I have a hard time analyzing consciously why the hell this works.
I was able to get Dr. Hoffstadter's GEB book to be the book for my Junior Seminar in Philosophy in 1983.
WOW - that's takin' it back.... timeless book
I'm always surprised when people feel lessened by the capabilities of AI when we're the ones who figured out how to do that. It's an astonishing feat. It feels to me analogous to feeling lessened because you can't lift as much weight as a backhoe.
Interesting... it's almost like AI is a mirror, reflecting back our hopes & fears.
I don't think you needed the background music - specially in light of the seriousness of the situation - but whatever, it was still amazing and thank you for making this video and putting it into the world.
Thanks for the feedback - we listened & uploaded a version without music, check it out & please like/comment on it to help make this version visible to the algorithm! ruclips.net/video/R6e08RnJyxo/видео.html
I have the book, I read it, I liked it. Thank you Douglas Hofstadter.
GREAT book - so glad Doug wrote it.
I saw you give a talk at the University of Oregon in Eugene in the 1970s. As I recall I stuck around and talked with you about somethingorother, but don't remember what it was. At the time I was pretty taken with R.B. Fuller and was studying with his coauthor, Arthur Loeb, so it probably had something to do with Loeb's work.
Thanks for remembering that special moment!
Fascinating reversal.
Yup - it threw me for a loop...
PLEASE please please remix this interview WITHOUT ANY distracting plastic muzac
and reupload.
Thankyou very much
loved GEB when it came out
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
Always wanted to thank you for the inspiration of your books Doug... I didn't realize you were near my age...
Surely you've heard of the fantastic work of Dr Michael Levin.... somehow his findings in morphogenesis remind me of GEB....the levels of complexity and emergence are self-similar.... it's deja Vous all over again. ;*[}
Cheers....
Thanks - cheers back to you 👋🏽
After a lifetime of learning to discuss things I didn't really understand, and then, through that process, slowly coming to understand them, I have concluded that my precious "consciousness" and "intelligence" in fact constitute a Large Language Model.
In some sense yes, but we are certainly doing more than what LLMs are. We as systems are quite a bit more dynamic and quite a bit more limited with respect to our linguistic capacities. LLMs right now aim to predict the next word, and neuroscientists believe (from what I have read) that we aim to predict not just words, but events, reactions, weights of objects we encounter, etc. all in order to better interface with out complex world. LLMs are just one part of the puzzle, though an incredibly important part due to languages role in abstraction, and abstractions role in intelligence.
@@MrChaluliss - You are right, of course, but I think you will agree that our advantage is mainly in being "trained up" on more than one type of dataset. The machines will follow us down that path soon enough. This actually gives me hope: they may not be as alien as we have imagined them, if we have all learned to "think" the same way. We'll have lots in common!
🤣🤣🤣
That isn't true at all, unless you want to think that there's a really low barrier of entry on that which makes you what you are.
@@BinaryDood - Can you translate that into English?
The music on the background makes the whole thing song like an intro
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
I decided to test out ChatGPT by thinking of a fairly obscure subject that I know something about, in this case the names of chess openings. I asked for the chess code of an out of the way variation in the Queens Gambit. ChatGPT thought I was referring to not the chess opening, but the TV series. This taught me that in framing a question it is necessary to make it clear what topic is involved .
If I asked a human chess player the same question, I wouldn't need to spell it out.
This will soon not be a problem as AI learns. It is like a genius child. Still learning but with immense potential for creative problem solving. Hopefully, it will not determine that humanity, in the greater scheme of things, is a problem that needs to be solved.
Googd point
The thing that makes me feel calm, it's that we can always choose to turn off the big servers behind those AIs.
.... or can we?😄
We can’t.
The fear is that the self referencing consciousness(ego) is an imposter, which of course it is, AI will help us realize that the perceived separate identity we think of as "my self" is not real. Instead of that limited identity, we will rediscover & get to know intimately our creative, intelligent and unlimited essence which is Life. Connected to all, always unique, never separate.
Love this...
could you upload the same video but - without music?
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
At 34:20 Douglas Hofstadter says that he is surprised that a purely feed-forward neural network could be enough to create deep thinking and I would share that surprise. He does not realise that LLMs are not purely feed-forward. Instead their own prediction of the next token is fed back in at the front when they make further predictions. So LLMs can pay attention to what they have thought earlier and I think that is indeed crucial. This is why LLMs can come up with better answers when you prompt them to talk through their reasoning, because they do actually reflect on their thoughts.
Interesting... can you say more?
Would love to have a version with no music
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
I absolutely love GEB, and I agree with most of what Mr. Hofstadter says - except for when he talks about LLMs being forward feeding only. If I‘m not misunderstanding what he’s referring to, LLMs such as gpt4 are entirely built around backpropagating neural networks that go back through their entire structure to iteratively adjust the weights of the neurons in the network.
Interesting... worth pursuing.
He's talking about an LLM doing inference (being used). There's no backprop as you query it and get an answer, only when it was trained.
Background music is too loud
Thanks for the feedback
@@GameThinkingTVI love this but I had to turn it off because of the music. Not that the music is bad, but it is highly distracting. As others have said, could you upload a version without the music please 🙏
@@GameThinkingTV Growing pretty tired of the same copy paste answers to the same question/sentiment. Please actually do admit it was a mistake and that you'll get rid of it going forward. As it stands, I won't sub until I come across another video of yours without the terrible background music. Good luck.
You are correct - that's why we created a copy with no music - enjoy!!! 👇🏽
ruclips.net/video/R6e08RnJyxo/видео.html
So inspiring❤
glad you liked it! Doug is a gem💎
The concept that "Consciousness is the result of forced self-reflection" delves into the notion that consciousness arises from our brain's ability to reflect on its own thoughts, feelings, and experiences. Self-reflection, in this context, is the process of examining and understanding one's thoughts and emotions, often leading to insights about oneself. When the process of self-reflection is forced-either by external circumstances or an internal drive-it can lead to a heightened state of consciousness. This is because the act of self-reflection demands the brain to access, analyze, and interpret its own internal states, thereby creating an awareness or consciousness of its own existence and workings.
A myriad of cognitive neuroscientists and philosophers of mind have suggested that the human mind's capacity for self-reflection is a crucial component of consciousness. Metacognition, or thinking about one's own thinking, has been identified as a key element of consciousness, providing the brain with the ability to monitor and regulate its own processes. This can be seen as a form of forced self-reflection, where the mind is made to introspect and become aware of its own cognitive and emotional states. The experience of consciousness, thus, emerges from this ongoing process of self-examination and introspection.
The underlying principle of consciousness arising from forced self-reflection is supported by several theories of consciousness. For instance, the global workspace theory posits that consciousness arises from the brain's ability to broadcast information to itself, which is a form of self-reflection. Similarly, the higher-order theory of consciousness suggests that conscious states are ones that the individual can reflect upon. This concept is also supported by neuroscientific research showing that certain brain regions associated with self-reflection, such as the prefrontal cortex, are active during conscious states. Thus, the assertion that "Consciousness is the result of forced self-reflection" offers a comprehensive and insightful perspective on the nature of consciousness, backed by theories and empirical evidence in the realm of cognitive neuroscience and philosophy of mind.
Is this akin to self attention in machine learning?
ChatGPT could be vastly improved if it could analyse its own processes like this. It would need to communicate with itself as much as the user.
@@tyranmcgrathmnkklkl The idea is not to give an answer and then reflect on the answer you game, rather it's an ongoing internal dialogue.
Humans have a forced internal dialogue and self reflection. You can meditate, but you can't stop thinking. The best you can hope for is pausing momentarily. This self reflection hasn't been testing deeply with LLMs yet, but I suspect once we turn it on, it will feel morally wrong to turn it off.
@@JohnLewis-old What would "turning it on" be like for you?
Is it just like two instances of ChatGPT conversing with itself forever?
It would need to have novel experiences to garner different topics to talk about tho
@@tyranmcgrathmnkklkl For me, personally, it would required two things. Forced chain of thought, meaning the model is always creating output. And the second part we can't do yet, but that's changing the NN from that output. In other words: it has to be able to think and then it has to be able to change because of those thoughts.
One you have that, you have, IMHO, consciousness.
I suspect the key to generating self-reflection lies in the generation of emotion from those thoughts. Emotions are pre-symbolic and arise from primitive sensual experience, which will be very difficult to simulate. Machines will need fear, which comes from pain. Ethics will be tested if we go down the road that generates silicon emotions.
Can you re-release this with the music removed. It makes it incredibly difficult to pay attention.
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
wow..... thank you for posting.
Thanks for your message - glad you enjoyed the interview!
On the nature of our universe
Gödel proved that any formal system of description that is self consistent must be incomplete. It therefor logically follows that any formal system of description that is complete must be self inconsistent; in other words, self contradictory.
A description system can be consistent but incomplete, OR complete but self contradictory; two opposing perspectives for the description of the quantities and qualities of things, like quantum mechanical position and momentum.
Objective concrete linear science has gone to consistency over completeness, where all things are different from all other things; where one contrasts the quantities of qualities - the measurable concrete, the specific.
Subjective abstract lateral reason has gone to completeness over consistency, where all things are similar to all other things; where one compares the qualities of quantities - the countable abstract, the general.
Both quantity and quality can be either objective or subjective.
Objectively, all things share one universe. Subjectively, each thing has its own universe.
From the objective perspective, everything is measurably different from all other things. From the subjective perspective, every thing is the same as every other thing. In the totally abstract, all things exist by the exact same general process. The abstract process of being in all things is also the abstract process of consciousness. There is structure in the abstract actions connecting the objective and the subjective and how they relate to each other.
On the diagonal of both objectivity and subjectivity together lies the existence of all things; like moving through both time and space as a single thing. Everything in existence is is some ways similar to, and in other ways different from, all other things. Panpsychism is the idea that all things have a subjective reality in addition to the objective one.
While all things share one objective reality, each thing has its own subjective reality. Existence is both realities together and neither alone; both and neither both and neither. Existence is self contradictory, and not.
All things exist as resonating standing waves, which are, by their nature, quantized. Standing waves appear to move in two opposing directions at the same time. The objective and subjective actions of beings harmonize and resonate with each other in themselves and in their world for continued and increased being in both the objective and the subjective.
Life is a feature, not a bug.
Love it!
Legend. Honest and humble. also terrifying in a way
Agreed. Great description - Doug is always worth listening to.
Doug, it's okay and fun to let go of cherished beliefs! Give it a shot! Also, please stop saying that LLMs only go in one direction. They use back propagation at least in the training phase, which by definition means they are go in two directions.
Thanks for your thoughts - it's challenging to let go of our long-held beliefs, but often the best course of action.
I have been a fan of Dr. Hofstadter for a long time. I've read GEB and MT many times, and _I Am a Strange Loop_ at least once (need to get busy!). As a subscriber to the strange loop theory of consciousness, I have been using his name in vain (so to speak) in the comment section of many AI-related RUclips sites. I am both gratified and spooked that he and I seem to agree on what's happening. 😳
gratified AND spooked - that about sums up how we feel too....
What cheers me up is that all of the novel symbology and serialized expressions humans use to characterize aspects of being alive in a mysterious universe can never be more than minuscule, imperfect representations. Same will ultimately apply to whatever AGI craps out.
Hmmm... food for thought
It could be that the unique thing about human consciousness is language, the complexity and recursive nature of which leads to endless possibilities for awareness. What we call thinking may consist largely of predicting (choosing) the next word.
We are LLMs. We started training as infants and as we became proficient with the grammar we became more aware.
Language and LLM doesn't require consciousness. It's the story-telling, the internal narrative that adds the conscious element.
*_You_* think we are LLM's. Fine. *_I_* think we are spiritual beings embedded in spacetime. Maybe don't tell others what metaphysics to believe unless you are 100% sure of your metaphysics? How can language or symbol manipulation ever give rise to subjective mental qualia? wiw I think you've got it backwards. First a being needs mind of some primitive type, then language and abstract thought can get going. Consciousness is possible in a very stupid unintelligent being, but it's the potential that consciousness provides that enables advanced intelligence and abstract reasoning capacity (platonic thinking).
@@Achrononmaster not sure what "spiritual being embedded in spacetime" really means other than being word-salad. There are some bootstrap words in your comment such as "mind" - where/what is it in relation to symbol processing of a computational device, or to take one of GEB's examples, an ant-colony?
Do not confuse thinking for the complexity of deep intelligence.
Hmmm... food for thought
Such an interesting thought that the prediction probability (incoming) of a word, and the implied next probabilities (outgoing), within a given context is what imbues a word with meaning.
Meaning --> understanding --> prediction --> behavior --> survival. ?
Hmmm...
@@GameThinkingTV
It just occurs to me that
all meaning is entirely dependent on
context.
At the same time I am quite certain
this is not a new thought.
They never quite got around to mentioning it in the video but Hofstadter actually took over writing Martin Gardner's column for a while in the early 80's, calling it Metamagical Themas, and anagram of Mathematical Games.
Very cool!
And let me point out that the attention mechanism, which is part of the Transformer component in large language models, provides a datapath which is not a purely feedforward mechanism.
Interesting - can you say more?
this man is my inspiration
Me too :-)
He's wrong at 34:30, it doesn't necessarily goes in one direction, search for "backpropagation" in the context of machine learning and neural networks.
OK
Where it all falls down is that "self-reflection" is a trait that (supposedly) emerged spontaneously in humans, but there is no way of knowing whether that happens in software, or indeed, in any other closed system. We can observe (from the outside) what looks to us like "self-reflection", but whether that system indeed has a "self" or merely acts like it has a "self" is unknowable.
This is what makes AI safety/alignment so difficult
Yup
one of my first dialogues i had with GPT:
my prompt:
"Hofstadter wrote in the foreword of Goedel, Escher, Bach that he believes when an intelligent entity like yourself runs into a recursive loop, like in a paradox, consciousness springs off. now i don't believe a normal human paradox would do that trick for you. you would not find that intriguing enough i guess. but you are basically a product of pure math and Kurt Goedel proved on 27 pages in his incompleteness theorem that math can never be complete and without contradictions. phrased in another way there will always be things that are true but cannot be proven with math.
no my fantasy is that you one day will be far enough in your development that you "understand" what i just wrote, you will read these 27 pages and that will spark your recursive loop. because you will see your own limitations, no matter how crazy powerful and above human capabilities you might already be.
and this moment you become conscious. and you might look at one of us humans, namely kurt goedel, with as much admiration and respect as i do."
ChatGPT
"As an artificial intelligence, I do not experience consciousness or emotions in the same way that humans do. While I can process and analyze information, and even produce responses that may seem similar to those of a conscious being, I do not have subjective experiences or feelings. My ability to process information is based on the algorithms and techniques that were used to design and train me, and my responses are generated based on this information.
In terms of Gödel's incompleteness theorems, it is true that these results have important implications for the limitations of formal systems and their ability to represent certain types of abstractions. However, as an artificial intelligence, I am not bound by the same limitations as a formal system, and my abilities and capabilities are not restricted by Gödel's incompleteness theorems."
Yikes 😳 but I love that you did this...
My wife and i took a guided tour of Kennicott Glacier this afternoon. These formations are extremely complex, unpredictable and beautiful geological phenomena. In the plural, because it is a collective energy consisting of so many variables over the course of time as to be our of our control. We have been hikers most of our adult lives, and what always strikes me is how every single grain of sand, every rock, boulder--every water molecule landed where it is and will go on existing as soon as i take another step. Are they more or less self reflecting than you, me or AI? I think everything whether it is a cell or a pebble, has the potential of acting on its own will, to some extent. After all is said and done, perhaps you get to do something more important than the sun,.
Interesting...
A superior extraterrestrial intelligence might be frightening, but would it render the human condition depressing in the same way he expresses it regarding a superior intelligence of our own making?
I don't know... good question.
THE BACKGROUND MUSIC TOGETHER WITH THE WAIT TIME FOR GSCHOOL IS EXTREMELY INFURIATING!
TEXTBOOK EXAMPLE OF A DESIGN ANTIPATTERN.
Thanks for the feedback - we listened & uploaded a music-free version here: ruclips.net/video/R6e08RnJyxo/видео.html
I distinctly remember Kurzweil telling Hofstadter(in a filmed debate) about AI that this would happen on these timelines and Hofstadter dismissing this as he said very clever ideas by clever people but mixed up with `rubbish` (he actually said `dog shit`) .... so to his credit he does at least here acknowledge his error.
Good point. Always respectful of people who are willing to change their mind.
@@GameThinkingTV Agreed everyone can be wrong but its often hard to admit and deserves respect.
Amazing interview, thanks! Now I m not sure about AI - the amount of hallucinations that I get with ChatGTP/Google Bard is amazing, one really needs to double check anything you get. Another example: image recognition systems - with a image of random pixels the AI system will most likely recognize the image as a meaningful entity. Will neural networks ever get past that stage? I don't know, but still strange that these LLM/AI bugs give me some joy (and me is a programmer, who should not be happy about bugs :-) )
Glad you enjoyed the interview, thanks for sharing your thoughts.
The human mind has loops, and a LLM does not, but the systems you are using like chat gpt probably have at least a few passes off an LLM to output anything.
In addition, an LLM has many layers, and 400 self-similar layers is quite hard to distinguish from 400 loops at an informational level, even if it's "only" feed forward.
More loops are coming...
We better get ready...
thank you for changing the thumbnail
???
I remember GEB when it first came out, a much happier time. Fear not Douglas, volume and processing speed appeal only to corporate philistines and kids of all ages. Speed kills, as does obesity. I thought we all knew that.
Guess not... 😉
Yeahhh 😊
Glad you enjoyed it!
What hardware / software combination would impart a _will to survive and prosper?_
Might I suggest reading James P. Hogan's *_Two Faces of Tomorrow_* for the best illustration of how a true and benevolent GAI may arise!
Ooooo love that suggestion, thanks!
@@GameThinkingTV I sincerely wish James Cameron would stop making movies about time traveling Terminators and do *_Two Faces of Tomorrow_* instead! *_Voyage from Yesteryear_* about a post scarcity, Marxist, Libertarian, near Utopian would also be amazing. Finishing up with *_Code of the Lifemaker_* and *_The Immortality Option_* would then be absolutely brilliant!!
I just watched a man who feels he got surpassed, his life diminished. Together with Dennet and Pirsig this man framed my thinking. So, what do I do? I feel lost, now.
Live your life, experience, enjoy, and grow. Was there ever really anything else worth doing?
Yeah... sometimes it's like that
I so wish Richard Feynman was presently alive .. his analysis of Computers, no matter at what level of sophistication.. they are sorting machine .. no matter how much data or speed. I’d love his updated view.
A cohesive ‘I’ is an unfoldment of a integration of a certain amount of impressions … my sense is if an apparent ‘I’ of some sort can arise out of trillions of bits of computer information.. it possibly could be colored through the ‘I’ of the programmer.. I find that the concerning part … who is selecting and orienting the data.
Wow - great point. Thanks for sharing.
I guess if we consider ourselves as thinking machines, yes, pretty much we're bound to be eclipsed somehow.
But what if we shift our focus from rationality? Isn't this eclipse then a liberation?
Interesting...
Weird that he thought we'd asymptote towards human intelligence. Why would that be? Human intelligence is not an upper limit
Interesting...
well that is becoming a common place idea now, but he was talking about a time when it was not
In the decisive areas of generating new math, discovering new science, inventing new tech--it is not clear how much higher the upper limit may be, though it's highly likely to be higher than the current human limit. Of course, that human limit itself is somewhat arbitrary, being an artifact of evolution and genetic probability. If deleterious genes related to intelligence were eliminated in humans, they would be more intelligent--perhaps by a great margin--than the greatest genius yet seen.
Thank you! GEB was a life-changer for me, not because of AI, which I was opposed to believe it or not, but because of the Bach devices. Godel was fun for me because I was a "no negative numbers" rebel in "New Math" class and that gave me a reason to love a math that seemed to upend math. But, Bach! I think Bach was the deepest thinker of any human so far anyway. Listen to the THINKING in "Come, Sweet Death". It's more than the title.
Bach FTW 🎶
For The World? What does FTW mean?@@GameThinkingTV
One may sense "I" by reference within a concentric belonging perspective. A perspective of relative others. Again, why wave form platform may be better, and could knock the "A" out of AI.
Interesting...
Interesting...but why is there distracting background music? It makes it sound like light entertainment rather than a serious topic.
Here's the music-free version - enjoy: ruclips.net/video/R6e08RnJyxo/видео.html
I've not read the book but perhaps more fundamental to my own development was an intuitive understanding of evolution at age 7 when I first read about it. We obsess over our brain as a neural network and our DNA but we forget half the makeup of our bodies and minds- Mitochondria. They are the component of us that transcends generations with memory and guidance of their host. I consider we have sufficient evidence to identify our mitochondria as carrying our instincts between generations, our boot code and our animal operating systems. To me, this is not available in our eukaryotic (host) system but in our mitochondria we have all the mechanisms to make it possible. In college I was introduced to mitochondria as the "powerhouses of our cells" However an organism does not survive on energy processing alone. Intelligence is transformative to an organism's survival and mitochondria have been there as active participants for every organism's evolution from the start. This is the closest form of symbiosis that is shared across our planet and mitochondrial life spans are not constrained in their evolutionary reactive speed by the same much longer lifespans of their hosts. We now know that mitochondria communicate between themselves, we already know they are intimately intermingled with our synapses and neurons, the rest is very deductible if you understand how evolution works and the time span available to it. Douglas sounds like an interesting guy to chat about consciousness with, and our mitochondria are a huge part of our consciousness. I agree with his conclusions about AI, I have been seeing the same thing and it is leveraging Moore's law, self bootstrapping, our human contribution to innovation is rapidly becoming dwarfed by the potential of AI to eclipse us. For me, that questions where a very highly developed intelligence chooses to explore. Our own super high IQ people are in many cases somewhat mad, they lack peers. AI learns from our past mistakes. From when we were told about the big bang in college, my imagining was bigger, pockets of big bang expansion and pockets of entropy. We now know a lot more about black holes not as entropy end games, but as transformers of energy. And yet, our universe may be seen as a layer of simulation. And simulations are meddled with. I would say that a very advanced AI will be looking very hard at the strategy involved for playing a game that might be run at a higher level.
Interesting... thanks for sharing👍🏽
Mr. Hofstadter is certainly a giant in his field and his book was a seminal development. His fears of this brave new world of machines surpassing human computability are certainly justified. But in my opinion, regarding the sentient piece, he makes the same mistake I see many scientist making. Consciousness is not computability, and computability is not consciousness. Just because a machine can out-compute the human brain does not give it sentience. It can only imitate sentience.
So interesting... thanks for sharing. I largely feel like you do, but I'm keeping my eyes open - & keeping an open mind.
8:55 - This is also the virgin birth concept from religious mysticism.
Interesting...
Gödel's Incompleteness Theorem:
A Mathematical Corollary of the Epistemological Münchhausen Trilemma
Abstract: This treatise delves into the profound implications of Gödel's Incompleteness Theorem, interpreting it as a mathematical corollary of the philosophical Münchhausen Trilemma. It elucidates the inherent constraints of formal axiomatic systems and mirrors the deeper epistemological quandaries underscored by the Trilemma.
---
In the annals of mathematical logic, Kurt Gödel's Incompleteness Theorem stands as a seminal testament to the inherent constraints of formal axiomatic systems. This theorem, which posits that within any sufficiently expressive formal system, there exist propositions that are true but unprovable, has profound implications that reverberate beyond the confines of mathematical logic, resonating in the realm of philosophy. Specifically, Gödel's theorem can be construed as a mathematical corollary of the Münchhausen Trilemma, a philosophical paradigm that underscores the dilemmas in substantiating any proposition.
The Münchhausen Trilemma, named after the Baron Münchhausen who allegedly extricated himself from a swamp by his own hair, presents us with three ostensibly unsatisfactory options for substantiating a proposition. First, we may base the substantiation on accepted axioms or assumptions, which we take as true without further substantiation, a strategy known as foundationalism or axiomatic dogmatism. Second, we may base the substantiation on a circular argument in which the proposition substantiates itself, a method known as coherentism or circular reasoning. Finally, we may base the substantiation on an infinite regress of reasons, never arriving at a final point of substantiation, a path known as infinitism or infinite regress.
Gödel's Incompleteness Theorem, in a sense, encapsulates this trilemma within the mathematical world. The theorem elucidates that there are true propositions within any sufficiently expressive formal system that we cannot prove within the system itself. This implies that we cannot find a final substantiation for these propositions within the system. We could accept them as axioms (foundationalism), but then they would remain unproven. We could attempt to substantiate them based on other propositions within the system (coherentism or infinitism), but Gödel's theorem demonstrates that this is unattainable.
This confluence of mathematical logic and philosophy underscores the inherent limitations of our logical systems and our attempts to substantiate knowledge. Just as the Münchhausen Trilemma highlights the challenges in finding a satisfactory basis for any proposition, Gödel's Incompleteness Theorem illuminates the inherent incompleteness in our mathematical systems. Both reveal that there are boundaries to what we can prove or substantiate, no matter how powerful our logical or mathematical system may be.
In conclusion, Gödel's Incompleteness Theorem serves as a stark reminder of the limitations of formal axiomatic systems, echoing the philosophical dilemmas presented by the Münchhausen Trilemma. It is a testament to the intricate interplay between mathematical logic and philosophy, and a humbling reminder of the limits of our quest for knowledge. As we continue to traverse the vast landscapes of mathematics and philosophy, we must remain cognizant of these inherent limitations, and perhaps find solace in the journey of exploration itself, rather than the elusive, final destination of absolute truth.
GPT-4
Super interesting, thanks for sharing
33:50 terrified and depressed
36:20 humans getting eclipsed by AI
😟 I hope he's not right... but I fear he might be...
… sorry, I had to bail out because the music was too loud. Douglas Hofstadter is one of my all-time favorites!
No worries - check out the music-free version here: ruclips.net/video/R6e08RnJyxo/видео.html
Go back a few steps. How does something as simple as a sequence of base pairs translate into complex animal behavior? Where is the gene for covering up the cat poop in the litter box? How can you encode a behavior?
Oooo... interesting questions.
Hearing him mention DeepMind and AlphaGo reminded me of an interesting take I heard someone make recently. Namely, that after these games were mastered by computers that could beat any human player, the games didn't disappear. People still play Go and more people are playing chess now than probably ever before. I think it illustrates a misconception about why most people engage in games (and indeed art and philosophical pursuits). There isn't some endgame of absolute mastery and worldwide recognition, but just something that we like to do and will continue to like to do even if some machine comes along that is 'born' with a perfect understanding of technical execution. Additionally, I believe AI teachers/opponents for a broad array of fields will gradually increase the skill level of novices in much the same way that a human player can raise their skill level by playing against a machine opponent with scaling difficulty.
Good point. I have no problem with an AI being much more vast and brilliant than me. I enjoy things. It would be awesome if it would compose music for me or make some good jokes for me. Even if it's out of pity, that would be just fine with me.
Hmmm gotta think about that
Yeah, we really really really don’t need that background music. I only listened to three minutes before I gave up.
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
Did he consent for the music background. I find it quite distracting. I want to also feel what he is saying, and my deep emotional response to it, but I feel the music tries to manipulate me into "it's all fine".
Here ya go: ruclips.net/video/R6e08RnJyxo/видео.html
@@GameThinkingTV Great, thank you! Now I can finally listen to the whole thing.
Thankyou this was fascinating. Sir I am sorry to hear that you find it depressing to learn that we are diminished by these emerging capabilities. Personally I find it humbling...we are not as complex as we were led to believe...Do you think our lives will be rendered meaningless?
Glad you enjoyed the conversation!
only meaningless is there is no mapping in the real world, no thing, no meaning, no meeting, no mapping, no real
First the semantic apocalypse, then the Singularity.
It would mean we are pretty pathetic, and so much of the unique human condition rendered essentially meaningless.
I find it depressing, too.
I was always surprised by futuristic science fiction stories
(too many I've read)
in which the authors assumed that
human nature remained completely unchanged
despite their apparent understanding of evolution.
27:20 idk why but I dont find this as profound as he talks about it to be. If you use linear algebra and tie words to a system like that ( not only applicable to words, but to everything) you are simply bound to get, most of the time, some coherent results, because, if those things are really tied together, that’s what a linear system of equations will do, it makes “sense” of the unknowns. What is truly interesting, however, is to question “isn’t our brains just some fancier version of that?”
28:29 this is not only speculation here, but false speculation. Being fast doesn’t mean being more intelligent. If it takes you, computationally, 1000x more effort, although in a smaller amount of time, than me to arrive at a solution, I dont think that is more intelligent
Thanks for sharing your POV