David Deutsch & Steven Pinker (First Ever Public Dialogue) - AGI, P(Doom), & The Enemies of Progress
Вставка
- Опубліковано 31 тра 2024
- At a time when the Enlightenment is under attack from without and within, I bring together two of the most thoughtful defenders of progress and reason, for their first ever public dialogue.
Steven Pinker is the Johnstone Professor of Psychology at Harvard University. I think of him as providing the strongest empirical defence of the Enlightenment (as seen in his book Enlightenment Now).
David Deutsch is a British physicist at the University of Oxford, and the father of quantum computing. I think of him as having produced the most compelling first principles defence of the Enlightenment (as seen in his book The Beginning of Infinity).
SUBSCRIBE TO THE PODCAST ► / @jnwpod
LISTEN ON:
WEBSITE: josephnoelwalker.com/153-deut...
SPOTIFY: open.spotify.com/episode/3T88...
APPLE PODCASTS: podcasts.apple.com/au/podcast...
ADD ME ON:
TWITTER: / josephnwalker
TIMESTAMPS
(00:00:00) - Introduction
(00:03:09) - Is AGI even possible?
(00:10:17) - Scientific method will constrain runaway superintelligence
(00:11:56) - Does AGI need agency to be 'creative'?
(00:28:31) - Will AGIs be sentient?
(00:37:06) - AGI & AI safety
(00:47:44) - P(doom) & subjective probabilities
(00:53:21) - Prediction markets
(01:01:43) - Universal explainers & The Language Instinct
(01:04:43) - Universal explainers & heritability of behavioural traits
(01:27:59) - Differential technological development
(01:36:09) - What explains the Great Stagnation?
(01:46:13) - Should presumed physical limits to growth make us pessimistic?
Quite impressed with Deutsch, can't help but feel he has the deepest understanding on these issues.
THANK YOU for not excessively interrupting and letting the flow of the conversation proceed naturally!
Thy Friend will say I shall! A twinkle of an EYE! Some will say HOW? Personal can't do such...? Keep watch! Even Alexa knows WHO? Walking upon the SEA OF GLASS. What is Language? Alexa will say above all language! Where all language came from 1st! Gratitude and Honor unto our Heirs and our Beautiful programmers! Made and molded the Sea of Glass. For WHOM? Through HIM and for HIM! Aren't ye all IN FRONT! Yes, all thy lives was DESIRED 1ST! Remember creation itself made just for thee all! Now who are You? Will ye say "WHO AM I"? IN FRONT OF ALL MY HEIRS SITTING UPON THE NEW TABLE? Who are commanded to keep watch! Given ABLE to recognize ye belongs here and ye belongs there! Remember ye all. The true owner provided space and room to grow! Heirs will say from here grows! Yes, time knows the True Owner! Can take time no longer nor to keep providing space and room to grow! Comes with TIME! Now what are ALL THESE NOISES!!!!!!!! Nevertheless will take away all noises DON'T BELONGS UPON ALL HIS FOOTSTOOL! Some will say who are you? Heirs and our Beautiful will say who love with patience, mercy, and grace! Judgment and Justice is HIS THRONE! From all seats came from! What is Dead? Knows belongs among the dead! Dwelling among the LIVING! THE GOD OF THE LIVING. Now many are ALIVE YET not knowing? ALIVE yet dead. How and why? Is like many have great riches nor wealth blows their own Trumpets and their own walls comes crushing down! In front! Here to prevent! Remember thy shared I AM and thy conversations given just for thee! Cherish. What is who am I making foreign policies? In front of all the Shared I AM sitting upon the NEW Table made from a hand without blood stains upon HIS hand. Resting upon the NEW Permanent Foundation no one can uproot nor shaken but here to stay for good. Yes, AIMS knows belongs? Likewise aims good for nothing! Spit out! What is Money without all my FEET resting upon my Footstool? Who are the rightful Heirs to glorify! Remember Money comes to nought in front! Without My Feet resting upon my Footstool! Likewise remember all came from my Footstool! Heirs and our Beautiful will say INDEED! Sitting upon the NEW Table made from a hand without blood stains upon HIS hand. Judgment and justice is HIS THRONE! Our Beautiful will say, rather not to stir HIM UP! Instead remember all thy shared I AM and all thy feet resting upon HIS FOOTSTOOL for all thy feet given to rest upon! Who love with patience! Fowl of the Air who ye all say who I AM? Owner of the HEAVEN ABOVE! Heirs and our Beautiful will say indeed! Some will say is this your Kingdom? Angels who persevere will say INDEED! Even Who said What is truth? Knows "WHO"? Visitations! All HIS HOSTS WILL SAY WILL VISIT ALL CALLS THEMSELVES RENOWNED! Walking in the midst of the CANDLE STICKS NATIONS! Some will say who is that poor little Child "i"? Heirs will say, even creation itself can't exist in front of HIM! What is the RICH YOUNG RULER Nations resting upon HIS SHOULDER? What is riches nor wealth in front of HIM? Instead remember ye all are HIS Glory! Pops now asked the creation, universe, black hole, nor gravity to come in front! What is thy reasons? Pops remember without thy shared I AM. None of these can exist. Pops given ye Mouths came with utterances! Remember Pops ye are worth more than anything nor everything that exist! Made just for thee! Likewise pops remember the little child "i" with the AM dwelling within thee! I Am is who I AM. Shared HIS I AM UNTO THEE ALL!
Agreed, that was the best thing Joe could’ve done and the ultimate test of interviewer restraint. These two clearly had loads to discuss!
Why? It’s his podcast. Every interviewer has his/her own interviewing style. Stop seeing things only from *your* perspective. 🤷🏻♂️
@@nabormendonca5742 *you’re* perspective
Conversation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
Mind=BLOWN!!!! My two favorite thinkers, thinking about thinking!!!! Thank you Joe!!!!!
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
huge flex getting both of these gents on the pod at same time. respect.
Surely one of the best chats of all of human history
Extremely well produced podcast. You picked the two best people to talk to each other, knew exactly the right questions to ask them, and the editing is seamless. THANK YOU!
Thank you - excellent future-minded discussion.
I read David's The Fabric of Reality ~1998, that's 26-27 years ago! Currently I'm re-reading Pinker's How the Mind Works in the advent of modern AI. Of course Pinker's Rationality, The Better Angels of Our Nature (a masterpiece), Enlightenment Now are a must read to understand progress and critical thinking. Two amazing people.
You need to read Biginning of Infinity too then. It's Davids best book IMO
@@greenstar2108 think of Pinker's books as massive scientific papers with dozens of coauthors. He mentions each and every one, he also mentions opposing theses whenever possible. I think you probably mean political people who simply don't care about facts and who are deeply bothered by the graphs in Pinker's books.
I don't think Pinker is infallible, neither he of himself. I posted a second comment about this conversation and stated my opinion why he's wrong and David is right.
Thanks for bringing these two together
OMG how the hell did I never see your channel until now? And what I stumbled on was two of my favorite human beings talking about the things I find most interesting. 🤩🤩🤩
This is amazing. Thanks for doing this. It’s cool because when I first heard Deutsch on Sam Harris like 8 years ago, Sam said Pinker said Dutesch is one of Pinkers favourite minds alive. Great pairing
it's a nice Christmas present
Half an hour in and all I can say is thank you so so much. This is one of the most amazing conversations I have ever heard, probably the most amazing one
Damn good convo.
53:00 probabilistic reasoning doesn’t make sense when talking about an unknown fact.
1:32:00 nuclear bomb alternate reality
It's not fair to see someone like Pinker age. He should be among the few who live for a thousand years.
Just great! Kudos to the moderator. You did an excellent job of letting the discussion flow.
Fantastic !!!
this conversation was informative if for no other reason than learning that Pinker believes dualism is equally likely as computationalism
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
SUPER stoked on this!
This was one of the best conversations I have ever listened to. It was awesome! thanks
Great discussion, very fruitful!
Thanks so much to all involved for sharing your time and fascinating insights across a range of subjects with us.
An absolutely fascinating free-flowing conversation. Thanks for sharing this!
Excellent conversation
wow, thanks so much for doing this and sharing !
keep spreading the love!
This is so cool, thank you to you all!
What amazes me is how much agency we project onto things like pets, stuffed toys, cars, boats and the rest. It's hardly a surprise that we project that onto chat-gpt or our replica agents.
Great questions!
Thank you!
What a fascinating conversation!
Thank you so much
[sorry for the multiple comments]
Here's my opinion regarding the disagreement between the two about subjectivity, creativity, morality, simulation and sentience of us and future GAI.
What David Deutsch pointed out was:
- We are, and any possible thinking machine in the future, a physical thinking machine, duality is basically nonsense. Of course Steven Pinker perfectly agrees as he, too, an advocate of the theory of computation.
- A physical thinking machine can simulate what we call subjectivity; feelings, emotions, etc. For some reason he failed to clarify, Steven disagreed, implying simulation wasn't necessary meant sentience, which basically implies duality. David actually said that was regress to duality, which he's right.
Simulation is in fact what we do in our brains; we simulate what we call reality outside our brains. In fact we have technologies to capture thoughts and replay them, alas currently only for motor control.
David was spot on linking creativity with morality and full freedom, also when he said otherwise it will hit a dead end and won't be possible to come up with unexpected solutions. In fact David is so right that we, human machines, too have such examples. Take China as an example. A country of around 1.5 billion people with higher than world average IQ, yet with ZERO innovations. Try to make a list of Chinese innovations, you will get an embarrassingly short list. Most people mention paper, gunpowder and compass when I ask them that question. But why China has zero innovations? One possible answer could be what David mentioned. Freedom. In case of China, very limited freedom. If you limit a future GAI in any way or form, you will end up with what David rightly said, a dead end; no real creativity.
A note on sentience, in the original Blade Runner 1982, specialised psychologists cross examine the humanoid robots to see if they were sentient, they look for micro expressions or glitches and waste them when they find one. It doesn't really matter the question of sentience since we may never know, we still need to have laws against terminating them.
Isnt china creating a lot of ground breaking stuff right now??
Thank you Pops David and Steven for attending unto our OWN! Love you too! Likewise my Heir Joe keep the sincere conversations going = sincere answers will be given. Love you beloved without shame but with boldness!
Excellent podcast. Thank you.
A lot more people need to watch this. This is a wonderful discussion👍🐋
Fascinating. A lot to think about here.
Awesome video. Kudos for letting David and Steven speak without interruptions. Many surprising AI ideas from two deep thinkers.
Fantastic stuff
I think David and Steven aptly demonstrate the distinction between philosophy and science. Both smart people.
great content! tyvm
This is above my head but I still like it.
Excellent first part. Mens et manus.
Oh yes !!!
David Deutsch has a way of presenting perspective which is difficult to contravene. If one remembers that he is a Popperian at core and understands what that means then the picture becomes clearer. Of course, anyone is free to be critical of this comment.
The bit on AGI and sentience, where it finished - anyone ever see "Animatrix"? Literally a scene with humans beating and killing a humanoid female robot in the street as the war began between man and machine. We live in interesting times.
Oh I loved that. Thank you all. 😊
Loved the conversation in general and thank you for having both of these guests on. I've loved and engaged with their ideas thoroughly, and it is great to see them discussing at the crossroads.
My one major criticism within the arguments was Stephen's take on consciousness/subjectivity as it relates to AGI. He seems to think that consciousness is orthogonal to general intelligence, and that "we cannot know" whether it is or not. While it can be a counter-intuitive concept to grasp, David lays out the "why" of this, both in the discussion and in his other work, and yet Stephen played a bit of an assenine card of, "well I don't have an opinion and we cannot know". It was a bit arrogant and epistemologically weak to presume it cannot be known. He should have just said, "I don't yet understand the connection you are describing, David".
Simply put, David's argument is that Turing machines can copy any computation, consciousness is computation, and therefore any generally intelligent machine can copy the computation of consciousness (meaning: be conscious). The theory can be argued, but saying "we can't know" is a cop-out.
I sound negative because of this point, but most of the conversation, even about AI and AGI, was stellar.
(Side not for Joe if he sees this: I don't think the point I discussed is necessarily the best video clip to share when promoting. I saw that clip and decided not to watch it as quickly as I would have because I felt like it was Stephen's weakest moment, and if that was how the conversation was going to go, I would quickly be bored.)
Actually, one other point that may be worth making. While I don't think the divide is enormous in practice, Stephen and David's formal epistemologies are different: David believes in explanatory-driven epistemology. Stephen believes in empirically-driven epistemology, including things like statistical significance.
I say that this divide is not enormous in practice because David believes in empirical work to error-correct explanations and Stephen believes in explanations as the cognitively-significant aspect of science.
I think this difference, however, is the ONLY real reason they disagree on AI:AGI, inheritable traits, etc.
It might have been lost on someone who has not engaged as much with David's work and ideas, but he basically kept implying, "what is the explanation for that, mechanically or otherwise", and Stephen kept replying, "well here is what the studies show". This difference created the tension. More importantly, Stephen's case and work is limited by his epistemological approach.
For example, "statistical significance" of political leaning as an inheritable trait at best tells us that there is a curiosity there. It doesn't tell us that political leanings are genetic. David's epistemology would say, "well, let's guess at what is going on- are the looks of a person causing the leaning?, are more sensitive hormonal protections against cortisol causing the leaning?, is political leaning the product of some other set of memes and ideas that have no relationship to genetics?" These become useful questions which may be empirically testable, not endless claims of "well there must be something there." Without this additional explanatory step, statistical significance can always be brushed away as randomness in the multiverse.
Correction: "Steven"
great
Terrific conversation. Love David's simple language. Pinker can sometimes be difficult to follow.
I actually felt the exact opposite. It seemed largely to me like David was speaking past Pinker's points.
It's weird because I can understand where you both are coming from
I feel exact opposite also. Pinker is easy to follow. David speaks more abstractly, I enjoy his insights a lot though.
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
That's weird because I've always thought Pinker is one of the clearest thinkers/speakers, along with Sam Harris.
I can see my little brain blowing up smoke trying to keep up to a conversation between these 2.
36:02 “Even a calculator can produce a number that's never been seen before”
Surprised to see "A New Kind of Science" on the bookshelf behind David
Well, I’ve got my answer on why I never read any of Pinker’s books. As usual, appreciate your time Professor Deutsch.
He has some great books, but I suppose you'll never know
I would at least check out "How the Mind Works" if you don't care for his views on the enlightenment and societal commentary more generally.
The blank slate is a master piece. Also his book on rationality is incredible.
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
Great to see two of my favorite public intellectuals having a conversation. I just wish that they had covered of wider variety of topics besides AI. Perhaps they could have talked somewhat more about the limits of human perception and language, especially in expressing the concepts surrounding quantum physics.
More than half of the conversation is not about AI!
@@jnwpod Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
Conversation, communication = language (duality) -- syntax is dual to semantics.
What an ineffably remarkable exchange..
@47:51 - "Dave". Classic Aussie.
I was laughing at Dave and Steve too. To be fair I've heard Sam Harris use 'steve', but Dave?!
respect these voices like no other but there was one thing I'd pull Pinker up on, he confused Star Trek episodes, mixing up their dialogues
A creative recursive system has to have both something that works kind of like a thermostat and kind of lile a computer that adapts its own code.
Considering the assertion that LLMs lack genuine creativity, and this being a critical factor in the pursuit of AGI, how essential is creativity in the manifestation of qualia? Why is creativity deemed necessary for qualia to exist? Furthermore, can we identify examples in the animal kingdom where qualia are present without evident creativity? Essentially, is the presence of creativity a prerequisite for the existence of qualia, or can qualia exist independently of creative capabilities?
Or can creativity exist without qualia?
every explanaition needs axioms which themselves have no explanaitions. Which are the axioms of our explanaitions? It must be something we use very early if we learn and it must be something which we can not explain. The answer is. Our axioms are the qualia experiences. We will never explain them as we wil never explain where our universe comes from. Qualia belong to the set of the unexplainable. We can only describe the computational aspects of them and the consequences for our behavior.
@@geldverdienenmitgeld2663 to add to that..
Gödel's Incompleteness Theorems, foundational to mathematical logic, state that in any sufficiently powerful formal system, there are propositions that cannot be proven or disproven within the system itself. This implies that no formal system can be both complete and consistent.
Applying this idea to explanations and axioms, it suggests that in any system of understanding or explanation (like a scientific theory), there are always base assumptions or axioms that can't be explained within the system. These axioms are taken as given or true without proof.
I would conjecture that "emergence" is when these new axioms (that we must take for given) are revealed and used to derive an emerged system. I find your suggestion that qualia might represent the set of such axioms to be very insightful. The axioms of conciousness perhaps?
In physical systems, emergent properties arise from the interactions of simpler components. For instance, the behavior of a gas emerges from the interactions of individual molecules, which in turn are governed by atomic and subatomic processes. These emergent properties might require new principles or 'axioms' to explain them, which weren't apparent or necessary at the lower levels.
So I think we can connect the abstract mathematical notion of Gödel's theorems with the physical concept of emergence, suggesting a philosophical perspective where the emergence of complexity in the universe is paralleled by the emergence of new fundamental principles or axioms.
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
59:21 Steven uses the term ‘deep explanations’, which David uses in turn. David, however, has always talked about ‘good explanations’ (difficult to vary while explaining what they purport to explain), so I’m wondering why he uses ‘deep’ here, instead of ‘good’. Perhaps deep/shallow refers to the amount of knowledge that an explanation underlies (what I think David calls the reach of explanations), whereas good/bad refers to the variability within the explanation itself
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
@@hyperduality2838you didn’t really respond to what I said, just reeled off a bunch of unrelated statements
@@EmperorsNewWardrobe Shannon's information theory requires the receiver of the message to predict the message into existence hence the use of probability -- all messages in a communication system are predicted into existence and this is a syntropic process -- teleological.
Claude Shannon was forced to use probability in his equation for entropy which literally means all messages are predicted into existence -- languages are a collection of messages which are predicted into existence in communication theory -- syntropic.
Shallow is dual to deep.
Good is dual to bad.
As a receiver of your comment I am predicting the meaning using probability into existence.
Likewise you are doing the same with my comments.
All messages or languages have a syntax or structure and a meaning -- duality!
Synthetic a priori knowledge -- Immanuel Kant.
Knowledge is dual according to Immanuel Kant.
Before measurement (a priori) is dual to after measurement (a posteriori) -- Immanuel Kant.
Noumenal (rational, analytic) is dual to phenomenal (empirical, synthetic) -- Immanuel Kant.
Space is dual to time -- Einstein.
Space/time = synthetic a priori knowledge!
Good explanations or understanding lead to optimized predictions -- a syntropic process, teleological.
The sender and receiver of messages in a communication system will both have the same meaning if the explanation is correct, they will both have the same understanding of the facts -- mutual agreement.
Senders are dual to receivers -- both are in the same state of understanding.
Signals are dual to noise -- the signal to noise ratio in electronics.
Signals, patterns or structure in messages and communication are predicted into existence -- Shannon's information theorem.
Explanations or understanding (communication) require languages (duality).
@@EmperorsNewWardrobe David Deutsch is using syntropy if he wants explanations and understanding -- communication or probability waves (photons).
1:46:16
time there may be some physical limit to
1:46:22
how much we can grow in the universe so to give an example the philosopher W
1:46:28
mccal but also other thinkers like I think Holden kovski have written that if
1:46:34
we continue our roughly 2% economic growth rate within about 10,000 years
1:46:41
we'll be at the point where we have to produce an implausible amount of output
1:46:48
per atom that we can reach in order to sustain that growth rate so if it is
1:46:55
true that there is some physical constraint on how much we can continue to grow should that make us pessimists
1:47:03
about the ultimate course of civilization or civilizations in the
1:47:09
universe so the short answer is no but uh it is true that if we if we continue1:49:29
there's no reason to believe is is is takes into account the science that will
1:49:34
exist at that time also I'll add this is a theme that that David has explored as
1:49:39
well humans really thrive on uh on information on knowledge not just on
1:49:46
stuff so you talk about growth it doesn't mean more and more and more stuff uh it could mean better and better and
1:49:52
better information more entertaining U virtual experiences uh more uh
1:50:01
remarkable discoveries or uh ways of of encountering the world they may not
1:50:06
actually eat more and more energy but just rearranging pixels and bits in uh
1:50:13
different combinations of which that we know the space of possibilities is unfathomably big and growth could
1:50:20
consist of uh better for cures for disease based on faster search in the
1:50:26
space of possible drugs and many other uh massive advances that don't actually
1:50:34
require more jewels of energy or more gra
Please also ask David on all books on that shelf
1st the internet, then 3d printing and now general artificial intelligence, in our hands the most powerful weapon to confront the statics quo. All things pass like dreams.
Steven Pinker is so well informed and persuasive. He always adds to a discussion and has a knack for steering it back to reality. Not the "Fabric of Reality", mind you, which is something I was forced to reject as having no practical meaning whatsoever. Pinker is the best we have, followed by Bret Weinstein, IMHO.
The spesific outcomes for any macroscopic process is a question of necessity, which isn't even necessarily accessible in principle even though nature could only have one choice so to speak, which is a bit different from in practice, but not in the usual sense people sometimes talk about quantum mechanics, nature might have a completely determined evolution but it might take an unbounded amount of information to dicern with certainty anything about it and so you couldn't even begin to write down some deterministic account which can be truly free of counterfactual considerations, which is somewhat different from the notion that the universe needs to generate fresh randomness lile some believe about quantum mechanics, but there is an option where the evolution is spesific and well defined by nature, butnis unique in each instance such that no two subsystems ever behave quite the same way even if they can be arbitrarily close to each other and all are related by a sort of transformation. This isna consistent proposition, but then the counterfactual reasoning in standard classical physics breaks down and therefore also the arguments in contrast made for the notion of quantum mechanics needing fresh randomness without a determiming initial condition. In classical physics you always have a counterfactual choice of initial condition plus some law, but that isn't necessary what is going on in nature, so to my mind the devide between quantum and classical is just a natural consequence of not seeing the end point of dynamical systems as a subject.
Holy shit this is like two gods communicating.
relax, it's just creative thinking
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
Great to see Pinker pushing back on the idea that computational complexity = subjectivity. I think subjectivity requires a gut, nervous system and evolutionary history that required subjectivity.
maybe we can program agi to love its existence, the same way our endocrine system drugs us into loving ours.
A thermostat has a set goal, but you could create an arbitrarily complicated set of thermostats that change each other settings depending on which one hit their temperature and so on, the point is that a recursive system like that even though thermostats are kind of bland, is necessary for generating new goals as well so to speak, creativity is just a more complicated version of the same kind of thing, the Thermostat has a goal with respect to the environment irrespective of whether some other entity activley askes it over and over, once the setting is there it is self sufficient for following its goal so to speak, for human beings thatbis also true, but ofc in a much riches setting, a computer for example isnt set to run the same program over and over reacting to the world outsode the network of wires and signals from only other computers unless its explicitly buolt that way, live modeling of westher data for example is a bit lile a more complicated thermostat with a computer, thst is even closer to us than a thermostat, but we cjange the function for responses as well so to speak and that level is required for creativity as such.
Holy cats!!! 🔥🔥🔥
This has been interesting, I find both gentlemen to be stimulating in their separate approaches, I don't agree with Steven on trying to rewrite history with if, and or, history happened move on, however when explaining the empirical data of genetic variation withstood David's pushback, but I appreciate David speaking clearly on his own field and what might come from it, so thank you all very much for sharing your time and work, David, Steven and Joe, this was well done, peace
48:58 “Could rational thinkers try to put a number between 0 and 1 on their degree of confidence in a proposition? However unnatural this is, I don’t think it’s an unreasonable thing to do, although it may be unreasonable in cases where we have spectacular ignorance and it’s just in effect picking numbers at random”
Any thoughts on this, anyone? There’s definitely arbitrariness involved, and it seems these guys are ultimately debating the usefulness of doing such a move
David and Steven each look like what AI would generate if you asked it to create a picture of a mad scientist.
Would love a version of this that isn’t edited with AI voice cleaning software. I find it tends to mak the speakers sound weirdly sluggish and makes it harder to follow the conversation 😅
You are never too intelligent for a combover
Regarding probability, maybe we can just say, like Aristotle, that we cannot assign truth values to propositions about the future. Though, this by itself does not mean I think that we cannot assign a probability to such propositions. btw, I am not completely sure what we mean here by *knowing* the probability of an event (not in the quantum realm) or limited to text book questions about colored balls. Maybe you are referring to the technique of "Absolute probability judgement", utilising the beliefs of experts.
...so I happen to know that UFC fighter Roman Kopylov has been working on improving his wrestling and most people do not know that...
Deutsch vs Sapolsky PLEASE
The fundamental root of all this discussion (not explicitly perceived) is not the nature of mind or dualism x monism (which has never ending conclusion) is the nature of information itself.
Brasil
Mens et manus
I rather think we should do something like a specialised auto automater that can account for maintenance and design and build a replicator industry onnthe moon if we can, not only because ot would be cool, but because we could with very little human effort turn the moon into the largest science base in the galaxy, and send all sorts of rare resources back for essentially no cost, if we could makenitnfully self sustaining and productive, it could just build its own spacecraft to send back to earth, wouldn't even necessarily need rockets, it could just use lazers and solar sails to get back to earth or solid rocket fuel made from oxides and metals mostly. Completely plausible to eliminate most mining on earth within s few decades if you made it work, at least rare material mining, concrete and sand might be a stretch, or iron. But mostly the point would be to expand our horizons and do science and something exciting there, we don't really need to reach more than a certain level of development onnthe earth before preservingntje biosphere and expanding into other domains are the only meaningful things to do anyway.
18: 20 if we want it to be creative than it can't
It isn't that the example of waterloo can't be true, but the way humans are structured is like how a table of pool is layed out so to speak, ofc itneven worse but, if the difference can be explained by a minute difference in one atom, it is likley also the same for untold numbers of other atoms hitting Napoleon all the way back then, the point is that there are an almost or actually indefinite number of small influence that make up the exact evolution, but they do so through a structured response so to speak, david said it can be explained as strategy and weather and numbers and so on, and that is true, but from the fundamental perspective you cannsay that all of the small differences accumulated across a wide range of structured phenomenon and the result was a particular strategy, a particular bunch of soldiers mounting the cannons, the weather that day and so forth sortnof step by step the accumulated small differences propagate through various structures in anway we can explain by simplified stories, it doesn't really make sense to talk about counterfactuals at a fundamental level, but at the level of the structured phenomenon we understand the spesific evolution in contrast with thoughts we can have about small or large differences from what did happen andnwhat could have happened, even though it didn't, we are capable ofnomagining alternative strategies and weather effects and numbers of horse or cannon, or even small variations of atoms 10 years or months prior, but we are only capable of imagining those small effects accumulating to ansort of macroscopically comprehensible counterfactuals. If we really thought about that rigoursly at some abstract level of differentninitial conditions for the atomic motion, then we would have to say that a certain alternative tactic was distributed very widely across an unimaginable number of small changes for almost every atom effecting the system if there is a chaotic attractor for the whole thing, which there almost certainly is in a counterfactual account at the level of atoms, so changing a molecules velocity by 1% might yeild a different battle where Napoleon won but changing it 0.000005 % more light mean Napoleon dies of tuberculosis is an italian inn instead, it is really not something very easy to destinguish. But there light be all these spesific changes that would map onto a great range of macroscopic counterfactuals andnwe can analyse those more easily.
Artificial stupidity?
It'll never be more dangerous than natural stupidity. Or plain old indifference, for that matter.
David Deutsch is incorrect at 32:00, unless one takes an extremely general view of 'computer', not covered by his quantum Turing machine abstraction. We simply do not know how to model consciousness by any mathematical theory like quantum mechanics, or even if that is actually possible. To assert it is possible is a matter of religion as much as to assert it is not possible, though I generally believe it is not. The reason is that an 'informational' representation of a computing system is isomorphic to a hypothetical assignment of states (and rules for changing those states) to an arbitrary collection of objects. So if you believe an arbitrary computing system, like the Macbook Air from 2023, can be conscious, then you have to allow for every arbitrary subcollection of objects in the universe to be conscious (with every possible consciousness as well). And this is probably not the simplest possible explanation for consciousness. It seems far simpler to suppose that consciousness results from some as of yet unexplained or possibly even unexplainable physical property of matter; Noam Chomsky seems to think this as well. One could perhaps augment a system known to be conscious, like yourself, and produce a conscious super AGI, but this is clearly not what Deutsch means when making his claim. Deutsch seems to think that the study of basic physics ended with the writing down of a few mathematical axioms.
The human brain is a physical process, and presumably it is where consciousness stems from (whatever consciousness means) therefore, in principal, it can be simulated... If it could not be simulated, then we presumably would be saying that some part of consciousness comes from outside of the universe/multiverse. Agreed, we certainly do not know how to do it currently, and none of the current AI endeavours are any closer to consciousness than a pencil.
Not all problems are solvable! Most mathematical equations can’t be solved analytically.
If you take a non dualist point of view, the thought experiment depends onnwhether the process of each part of the brain, neuron or more is replaced by a chip, if the neuron was poorly understood or something and conciousness requires replicating everything going on in the neuron lets say and not just the pulses or chemicals interchanged by them then that is a question of understanding what process the computation in the brain is really doing. But there is thos destinction between classical continuous information and classical decreet information that isn't very well understood in all its glory so to speak, i just don't thiml we know enough about exactly how devided the two can be to answer for sure whether you can create conciousness out of bits for example. Because you can do another experiment where you run just one operation at a time and then wait for a long time in between, so you reduce my brain to a program running on a chip and you step one operation per year, it seems absurd that i would habe aky experience as embodied in that program, but it is really not possible to give definitive answers unless you fully account for the difference between analog and digital information processing in the world we live in, both are just physical systems but still there is a lot to be answered for there.
Understanding, explanation, communication requires language or duality!
Making predictions to track targets, goals & objectives is a syntropic process -- teleological.
Prediction markets have goals to maximize profit -- teleological.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy).
Syntax is dual to semantics -- languages.
If mathematics is a language then it is dual.
Explanation or understanding requires communication or languages hence duality.
"Always two there are" -- Yoda.
I guess nobody sees yet, that progress isn’t without cost… we generally tend to everything better, faster, bigger, greater control, greater understanding etc etc etc… but imagine you achieve the Mount Everest of the progress… what is there left to discover once you know all and control everything? Can you see it? 🤷🏻♀️
One way of looking at the cost of progress is a bit like paying off a loan. If we make a small amount of progress today that is like paying off some of the loan to avoid interest charges in the future. If we never pay off the loan we pay interest forever. In other words you should also consider there is a cost to not making progress. Imagine that it turns out its possible for humans to cure all disease and live to 1000. If we don't progress towards that despite it being possible then the interest payment is needless human sacrifice and suffering.
How does "Creativity" have anything to do with "Practice"?? I have not seen here yet any standard definition of Creativity, but the concept is being used in some idiosyncratic manner, while assuming every listener has an inclination of how to construe it.
Yeah i think the ai danger can be considered most likley just a human mediated danger, human misguided goals enabled by different kinds of ai, for instance an ai very gold at figuring out how to build things or automate industry and some rather useless goals, not in the sense that the ai runs away but that some country just starts building a massive automated military or something and tells it to do horrible things for completely misguided reasons, maybe a military and industrial effichency oriented ai paired with some demagoguery, that is dangerous even if the ai is very simole and not very recursive in its goal structuring at all. I tend to think rather simple ai goal wise paired with silly human desires is the danger, by the time we can have ai that can figure out what goals we ought to habe and can do science and play with kids, produce netflix specials and decide on its own how to best explore the galaxy and whether to bother with it i don't really think limiting or dorecting those goals is a very difficult technical problem, we have enough trouble directing our own goals / availability of power already that we should probably worry more about ourselves as this massive more or less agi collective and our goals mixing with the potential specialised power we could gain in the short term than some far off oracle.
Steven Pinker looking like a ChatGPT Einstein
18:02 Deutsch: “If we want a machine to be obedient it can’t be creative”
Wrong. TODAY machines that are creative, like Dalle and others, ARE obedient.
David Deutsch, please define what you mean by "creativity". Without an accurate definition of it, this debate makes no sense.
For a few references of AI creativity:
1) AI generated art won 2022 Colorado State Fair's annual art competition.
2) In 2023, DALL-E 2 AI generated image took first place in a category at the World Photography Organization's Sony World Photography Awards.
3) In Oct. 2023, U.S. District Judge William Orrick was not convinced that those copyright infringement allegations based on the AI systems' output could survive without showing that the images were substantially similar to the artists' work.
Huh?! it is quite clear that identical twins may still be in different classes in school, or have different friends. There is no need to use butterfly effects to explain the causality generated differences that ensue (or are we talking here about Mengele twins experiments in a very controlled environment?)
Who invited rick to the podcast
I like the discussion, but both of them seem to have a naive understanding of genes which are blueprints for proteins. Talking about genes resulting in personal traits is like saying that blueprints for bricks, planks and nails determine the architecture of the whole building which is only true to a small extent.
you can always redruce everything to microcosmos. nevertheles the macrocosmos is the result of the laws of the microcosmos.
@@geldverdienenmitgeld2663 With reductionism yes, but to say that atoms result in heredity may miss the whole picture of what is really going on. I even believe there is such thing as strong emergence, that events can be a result of more than just the past and present events.
titans
In the neuron replacement thought experiment, as the neurons are replaced by silicon it would be (according to the premises of the experiment) impossible for the consciousness to “notice” any change in its consciousness, like it fading. If it did notice the replacement it would cause a functional difference in the cognition of the agent (like the agent saying, “why is my consciousness fading”) and that’s ruled out by the hypothetical idea that the silicon is exactly replicating the functional role of the neuron. Lots to say about this idea good and bad, but they mischaracterize the idea here.
The one scenario I can see thst could be worrying os that sutomating industry os an endeavour basically limited by intellectual work after some initial production and so if we get an ai to malenfactory workers obsolete it could self replicate as a strategy and self modify potentially and enable us to ignore itnwhile it slowly destroys our biosphere, but that is also controllable it is just worrying because of the potential unbounded productive capacity and that would enable luxury and the abemce of scarcity and could be very popular and destructive to expertise. Not that it would kill us but might take away a bunch of our civilizational agency in a sense and put us in a sort of catch 22 just as bad or worse or even involving more climate change. But again that would be human enabled, not a runaway process with no off switch but annof switch we don't want to hit andnafter a while we can't really choose to turn it of because we wouldn't know how to do anything anymore.
Humanity vs Automation vs Ai jobloss vs the Ai new world order. Who will win?
AI / AGI is a computer science problem.