 Our next talk is going to be about AI and it's going to be about proper AI It's not going to be about deep learning or buzzword bingo. It's going to be about actual psychology It's going to be about computational meta psychology and now please welcome Yosha Thank you I'm interested in understanding how the mind works and I believe that the most fruitful perspective at looking at of looking at minds Is to understand that we are systems that you throw patterns at them You find meaning and we find meaning in those in very particular ways and this is what makes us who we are So the way to study and understand who we are in my Understanding is to build models of the information processing that constitutes our minds Last year about the same time I answered the four big questions of philosophy. What's the nature of reality? What can we know who are we what should we do? So now how can I top this? I'm going to give you the drama that divided a planet some of the very very big events that happened in the course of last year So I couldn't tell you about it before What color is the dress I mean If you are have do not have any mental defects you can clearly see it's white and gold, right? Turns out Most people seem to have mental defects and say that it's blue and black. I have no idea why Well, okay. I have an idea why that is the case I guess that you got to it has to do with color renormalization and color renormalization Happens differently apparently in different people. So we have different wiring to renormalize the white balance And it seems to work in real-world situations in pretty much the same way But not necessarily for photographs which have only a very small Fringe around them which gives you hint about the lighting situation and that's why you get these huge divergences, which is amazing So what we see that our minds cannot know objective truths in any way outside of mathematics They can generate meaning though How does this work I? Did a robotic soccer for a while and there you have the situation that you have a bunch of robots that are Situated on the playing field and they have a model of what goes on in the playing field physics generates data for their sensors they read the bits of the sensors and then they use them to update the world model and Sometimes we didn't want to take the whole playing field along and the physical robots because they are expensive and heavy and so on Instead if you just want to improve the learning and the gameplay of the robots you can use a simulation So we wrote a computer simulation of the playing field in the physics and so on that generates pretty much the same data and put The robot mind into a simulated robot body and it works just as well That is if you are the robot because you cannot know the difference if you are the robot You cannot know what's out there the only thing that you get to see is What is the structure of the data at your systemic interface? And then you can derive a model from this and this is pretty much the situation that we are in that is VR minds that are somehow Computational that are able to find regularity in patterns and they seem to have access to something that is full of Regularity so we can make sense of it Now if you discover that you are in the same situation as these robots Basically, you discover you are some kind of apparently biological robot that doesn't have direct access to the world of concepts That doesn't has never actually seen matter and energy and other people all it got to see was little bits of Information that were transmitted through nerves and the brain had to make sense of them by counting them in elaborate ways What's the best model of the world that you cannot can up is what's really the state of affairs? What's the system that you are in and What's the best algorithm that you should be using to fix your world model and This question is pretty old and I think there has been answers for the first time by Ray Solomon of in the 1960s He discovered an algorithm that you can apply when you discover that you are robot and all you got is data What is the build like and his algorithm is basically a combination of Bayesian reasoning induction and Occam's razor and We can mathematically prove that we cannot do better than Solomon of induction Unfortunately, Solomon of induction is not quite computable But everything that you're going to do is some going to be some approximation of Solomon of induction So our concepts cannot really refer to facts in the world out there We do not get the truth by referring to stuff out there in the world We get meaning by suitably encoding the patterns that our systemic interface and AI has recently made huge progress in encoding data at perceptual interfaces Deep learning is about using a stacked hierarchy of feature detectors That is we use pattern detectors and we built them into networks that are arranged and hundreds of layers and then we adjust the links between these layers Usually some kind of using some kind of gradient descent and you can use this to classify for instance images and parts of speech So we get two features that are more and more complex They started with very very simple patterns and then get more and more complex until we get to object categories And now the systems are able in image recognition tasks to approach performance. There's very similar to human performance Also, what is nice is that this seems to be somewhat similar to what the brain seems to be doing in visual processing and if you take the activation in different levels of these networks and you improve this or enhance this activation a little bit what you get is stuff that looks very psychedelic which might be similar to what happens if you put certain illegal substances into people and Enhance the activity on certain layers of their visual processing Pose that she has the lighting the dress that she has on her facial expression and so on and then be good only to this thing That is left after we removed all the news and stator but what if we want to get to Something else for instance if you want to understand poses could be for instance that we have several dances and we want to understand What they have in common so our best bet is not just have a single classification with filtering But instead what we want to have we take the low-level input and get a whole universe of features that is interrelated So we have different levels of interrelations at the lowest level We have percepts on a slightly higher level that we have simulations and on the even higher level We have a concept landscape How does this representation by simulation work now imagine you want to understand sound? If you are a brain and you want to understand sound you need to model it Unfortunately, we cannot really model sound with neurons because sound goes up to 20 kilohertz or if you are all like me maybe to 12 kilohertz 20 kilohertz is what babies do and Neurons do not want to do 20 kilohertz. That's way too fast for them. They're like that's something like 20 Hertz So what do you do you need to make a Fourier transform before we a transform measures the amount of energy at different frequencies and Because you cannot do this with neurons You need to do it in hardware and turns out this is exactly what we are doing We have this cochlear which is this snake a snail like thing and our ears and what this does it transforms energy of sound Different frequency intervals into energy measurements and then gives you something like what you see here And this is something that the brain can model So we can get a neural simulator that tries to recreate these patterns and then can predict the next input from the cochlear That it understands the sound of Course if we want to understand music we have to go beyond understanding sound We have to understand the transformations that sound can have if you played a different pitch We have to arrange this sound in the sequencer that gives you reasons and so on and then we want to identify some kind of musical grammar That we can use to again control the sequencer so we have stacked structures that simulate the world and once you've learned this model of music once you've learned the Musical grammar the sequencer and the sounds you can get This to the structure of the individual piece of music so if you want to model a world of music you need to have the lowest level of the percepts then we have a higher level of mental simulations and which give the sequences of the music and the grammars of music and Beyond this you have a conceptual landscape that you can use to describe the different styles of music and If you go up in a hierarchy you get to more and more abstract models more and more conceptual models And more and more analytic models and these are causal models at some point these causal models can be weakly Deterministic basically associative models, which tell you if this state happens. It's quite probable that this one comes afterwards Or you can get to a strongly determined model a strongly determined model is one which tells you if you are in this state And this condition is met you're going to go exactly in this state If this condition is not met or a different condition is met you go into this state and this is what we call an algorithm It's now you're in the domain of computation Computation is likely different for mathematics. It's important to understand this For a long time people have thought that the universe is written in mathematics or that Mines are mathematical or anything is mathematical. In fact, nothing is mathematical mathematics is just the domain of formal languages It doesn't exist Mathematics starts with a void You throw in a few axioms and if you chose a nice axioms then you get infinite complexity Most of which is not computable a Mathematics you can express arbitrary statements because it's all about formal languages many of these statements will not make sense Many of these statements will make sense in some way But you cannot test whether they make sense because they're not computable Computation is different Computation can exist it starts with an initial state and then you have a transition function You do the work you apply the transition function you get into the next state Computation is always finite Mathematics is the kingdom of specification and computation is the kingdom of implementation It's very important to understand this difference All our access to mathematics, of course is because we Do computation we can understand mathematics because our brain can compute some part of mathematics very very little of it and to a Very constrained complexity but enough so we can map some of the infinite complexity non-computability of mathematics Into computational patterns that we can explore So computation is about doing the work. It's about executing a transition function Now we saw that matter representations is about percepts mental simulations Conceptual representations and these conceptual representations give us concept spaces and the nice thing about these concept spaces is that they give Us an interface to our mental representations We can use to address and manipulate them and we can share them in cultures and These concepts are compositional you can put them together to create new concepts and they can be described using higher-dimensional vector spaces They don't do simulation and prediction and so on but we can capture regularity in our concepts with them with these vector space You can do amazing things for instance If you take the vector from king to queen pretty much the same vector as to between men and women and because of these properties Because it's really a high-dimensional manifold these concept spaces We can do interesting things like machine translation without understanding what it means that is without doing any proper mental representation That predicts the world So this is type of mental representation that is somewhat incomplete, but it captures the landscape that we share in a culture and Then there is another type of mental representation that is linguistic protocols Which is basically a formal grammar and a vocabulary and we need these linguistic protocols to transfer mental representations between people And we do this basically by scanning our mental representations disassembling them in some way or disambiguating them And then we use a discrete string of symbols to get this to somebody else and he trains an assembler That reverses this process and builds something that is pretty similar to what we intended to convey And if you look at the progression of AI models it pretty much went the opposite direction so AI started with linguistic protocols and which were expressed in formal grammars and then it got to concept spaces and now it's about to address presets and At some point in near future. It's going to get better at mental simulations and At some point after that we get to attention directed and motivationally connected systems that make sense of the world that are In some sense able to address meaning. This is the hardware that we have can do What kind of hardware do we have? That's a very interesting question It would start out with the question how difficult is it to define a brain? We know that the brain must be somewhere hidden in the genome the genome fits on a CD wrong. It's not that complicated It's easier than Microsoft Windows and we also know that about 2% of the genome is Coding for proteins and maybe about 10% of the genome has some kind of stuff that tells you when to expect per switch protein And the remainder is mostly garbage It's all viruses that are left over and has never been properly deleted and so on because there are no real code revisions in the genome so How much of these 10% that is 75 megabytes code for the brain? We don't really know what we do know is we share almost all of this with mice Genetically speaking a human is a pretty big mouse with a few bits changed so to fix some of the genetic expressions and That is most of the stuff there is going to code for cells and metabolism and what your body looks like and so on But if you look at how much is expressed in the brain and only in the brain in terms of proteins and so on we find it's about well of the 2% It's about 5% that is only the 5% of the 2% that is only in the brain Another 5% of the 2% is predominant in the brain that is more in the brain than anywhere else Which gives you sometimes thing like a lower bound which means to encode a brain Genetically based on the hardware that we are using we need something like at least 500 kilobytes of code Actually this V a very conservative lower bound is going to be a little more I guess But it sounds surprisingly little right, but in terms of scientific theories, this is a lot I mean the universe according to the core theory of quantum mechanics and so on is like so much of code It's like half a page of code. That's it. That's all you need to generate the universe And if you want to understand evolution, it's like a paragraph It's couple lines really to understand an evolutionary process And there's a lot lots of details that you get afterwards because this process itself doesn't define what all the animals are going to look Like in a similar way is the code of the universe doesn't tell you What this planet is going to look like and what you guys are going to look like it just defining the rulebook and In the same sense the genome defines the rulebook by which our brain is built The brain Boots itself in a developmental process and this booting takes some time So some initial learning in which initial connections of watch and basic models are built of the world so we can operate in it and How long does this booting take? I think it's about 80 mega seconds That's the time that a child is awake until it's three and a half years old by this age You understand Star Wars and I think everything after understanding Star Wars is cosmetics You're going to be online if you get to a ripe old age for about 1.5 giga seconds And in this time, I think you are going to get not too much more than five million concepts Why I don't know if you look at this child if a child would be able to form a concept Let's say every five minutes then by the time it's about four years old it's going to have something like 250,000 concepts and So a quarter million and if we extrapolate this into our lifetime at some point It slows down because we have enough concepts to describe describe the world. Maybe it's something. It's I think it's less than five million How much storage capacity does the brain has and I think that the estimates are pretty divergent The lower bound is something like a hundred gigabytes and the upper bound is something like two point five petabytes There is even even some higher Outliers this if you for instance think that we need all the synaptic vesicles to store information Maybe even more fits into this but the two point five petabytes is usually Based on what you need to code the information that is in all the neurons But maybe the neurons do not really matter so much because if a neuron dies It's not like your world is changing dramatically. The brain is very resilient against individual neurons failing So the hundred gigabyte capacity is much more what you actually Store in the neurons if you look at all the redundancy that you need and I think this is much closer to the actual Bell Park figure Also, if you want to store five hundred five million concepts and maybe ten times Or hundred times the number of percepts on top of this. This is roughly the ball park figure that you're going to need so our brain is a prediction machine it what it does is it reduces entropy of the environment to solve whatever Problems you are encountering if you don't have an other feedback loop to fix them So normally when something happens we have some kind of feedback loop that regulates our temperature or that makes problems go away And only when this is not working the employee cognition and then we start this arbitrary Computational process that is facilitated by the neocortex and this neocortex can really do arbitrary programs But it can do so only with a very limited complexity because really you just saw it's not that complex It the modeling of the world is very slow It's also something that we see in our a more eye models to learn the basic structure of the world takes a very long time to learn Basically that we are moving in 3d and our objects are moving and what they look like Once you have this basic model we can get to very very quick understanding Visit this model basically encoding based on the structure of the world that we've learned and This is some kind of data compression that we are doing we use this model this grammar of the world These simulation structures that we've learned to encode the world very very efficiently How much data compression do we get? Well, if you look at the retina the retina gets data in the order of about 10 gigabits per second and The retina already compresses these data and puts them into optic nerve at the rate of about one megabits per second This is what you get fed into the visual cortex and the visual cortex does some additional compression and By the time it gets to layer four of the first layer of vision to a V1 We are down to something like one kilo bit per second so if you extrapolate this and you get lift to the age of 80 years and You are awake for two-thirds of your lifetime That is you have your eyes open for two-thirds of your lifetime the stuff that you get into your brain By your visual perception is going to be only two terabytes Only two terabytes of visual data throughout all your lifetime. That's all you're going to get ever to see Isn't this depressing? So I would really like to Tell you choose wisely what you're going to look at Okay, let's look at this problem of neural compositionality Brains have this amazing thing that they can put meta representations together very very quickly For instance, you read a page of code you compile it in your mind into some kind of program It tells you what this page of code is going to do isn't that amazing and then you can forget about this disassemble it All and use the building blocks for something else. It's like Legos. How can you do this with neurons? Legos can do this because they have a well-defined interface. They have all these lots You know that fit together in well-defined ways. How can you rinse do this? Well, and you rinse could maybe learn the interface of other neurons But that's difficult because every new one looks slightly different after all this is some kind of biologically grown natural stuff So what you want to do is you want to encapsulate this Diversity of the neurons to make them predictable to give them a well-defined interface And I think that nature's solution to this is cortical columns Cortical column is in circuit of between a hundred and four hundred neurons and this circuit has some kind of neural network that can learn stuff and after it's learned a particular function and in between it's able to link up this other cortical columns and We have about a hundred million of those depending on how many neurons you assume are in there We guess it's something at least 20 million and maybe something like a hundred million And these cortical columns what they can do is they can link up like Lego bricks and then perform by Transmitting information between them pretty much arbitrary computation. What kind of computation? Well Solomon of induction and They have some short-range links to their neighbors which come almost for free because Well, they're connected to them then direct neighborhood and they have some long range connectivity So you can combine everything in your cortex with everything So you need to have some kind of global switchboard some grid like architecture of long range connections They're going to be more expensive. They're going to be slower, but they're going to be there So how can we optimize what these guys are doing in some sense? It's like an economy It's not an entry-based system as we often use in machine learning It's really an economy you have the question is you have a fixed number of elements How can you do the most valuable stuff with that fixed resources most valuable stuff the problem is economy So you have an economy of information brokers every one of these guys of these little cortical columns It's a very simplistic information broker and they trade rewards against Nick entropy against reducing entropy in them in the world and to do this we just saw that they need some kind of standardized interface and Internally or to have used this interface they're going to have some kind of state machine and then they are going to pass messages between each other and What are these messages? Well, it's going to be hard to discover these messages by looking at brains Because it's very difficult to see in brains what they're actually doing You just need all these neurons and if he would be waiting for newer science would discover anything We wouldn't even have gradient descent learning or anything else We wouldn't have newer learning. We wouldn't have all these advances in AI You can submit who was that that the biggest or the last contribution of newer science to artificial intelligence was about 50 years ago That's depressing and it might be over Emphasizing the unimportance of newer science because newer science is very important once you know what you're looking for you Can actually often find this and see whether you're on the right track But it's very difficult to take newer science to understand how the brain is working because it's really like understanding Flight by looking at birds through a microscope So what are these messages? You're going to need messages that tell these cortical columns to join themselves into a structure and to unlink again once they're done You need ways that they can request each other to perform computations for them You need ways they can inhibit each other when they're linked up so they don't do conflicting computations Then they need to tell you whether the computation or the result of the computation that they're asked to do is probably false Or whether it's probably true But you still have to wait for others to tell you whether the details work out or whether it's confirmed true But that the concept that they stand for is actually the case and then you want to have learning to tell you how well this worked So you will have to announce a bounty that tells him to link up and kind of reward signal That makes them do computation in the first place and then you want to have some kind of reward signal when you cut a result as an Organism when you reached your goal if you made the disturbance go away or whatever if you consume the cake And then you will have some kind of reward signal that you give everybody That was involved in this and this reward signal facilitates learning So the difference between the announced reward and the consum assumed reward is the learning signal for these guys So they can learn how to play together and how to do the Solomonov induction Now I told you that Solomonov induction is not computable And that's mostly because of two things first of all it needs infinite resources to compare all the possible models And the other one is that we do not know the prior probability for our Bayesian model We do not know how likely unknown stuff is in the world So what we do instead is we set some kind of hyperparameter some kind of default prior probability for concepts that are encoded by the cortical columns and If we set this parameter very low, then we are going to end up with inferences that are quite probable For unknown things and then we can test for those if we set this parameter higher We are going to be very very creative But we end up with many many theories that are difficult to test because maybe there are many too many theories to test Basically every of these cortical columns will now tell you when you ask them if they are true Yes, I'm probably true But I still have to ask others to work on the details So these others going to be get active and they're being asked by this asking element Are you going to be true and they say yeah, probably yes I just have to work on the details and they're going to ask even more So your brain is going to light up like a Christmas tree and do that's all these amazing computations And you see connections everywhere. Most of them are wrong You're basically in a psychotic state if your hyperparameter is too high Your brain invents more theories that it can disprove Would it actually be sometimes be good and to be in this state you bet So I think every night our brain goes in that state We turn up this hyperparameter we dream we get all kinds of weird connections And we get to see connections that otherwise we couldn't be seeing even though that because they're highly improbable But sometimes they hold and we see oh my god The DNA is organized in a double helix Wow, and this is what we remember in the morning All the other stuff is deleted So we usually don't form long-term memories and dreams if everything goes well if you accidentally trip this up your modulators for instance by consuming illicit substances or Because you just go randomly psychotic. You both basically enter a dreaming state. I guess you get to a state where the brain Starts inventing more concepts that we can disprove so you want to have a state that is well balanced and The difference between a highly creative people and very religious people is probably a different setting of this hyperparameter So I suspect that people that are genius like people like Einstein and so on Do not simply have better new ones than others what they mostly have is a slightly hyperparameter that is very finely tuned So they can get better balance at than other people in finding theories that Might be true, but can still be disproven So inventiveness it could be a hyperparameter in the brain If you want to measure the quality of the belief that we have We are going to have to have some kind of cost function which is based on the motivational system and To identify if the belief is good or not We can have structural criteria for instance how well does it predict the world or how well does it reduce uncertainty in the world? Or is it consistency in sparse and then of course utility how well does it help me to satisfy my needs and the motivational system is going to Evaluate all these things by giving a signal and the first signal kind of signal is There are possible rewards if we are able to compute a task and this is probably done by dopamine So we have a very small area in the brain a substancia niga and then the ventral tegmental area And they produce dopamine and this gets fed into The dorsal lateral cortex and the frontal lobe which control attention and tell you what things to do and If we have successfully done what you wanted to do We consume the rewards And we do this with another signal, which is serotonin It's also announced the motivational systems was very small area the raffa nuclear and it feeds into all the areas of the brain we're learning as necessary connections as strengths once you get to result and These two substances are emitted by the motivational system. The motivational system is a bunch of needs They're centrally regulated below the cortex. They're not part of your mental representations They are part of something that is more primary than this. This is what makes us go. This is what makes us human This is not our rationality. This is what we want and the needs Physiological they are social and they are cognitive and we are pretty much born with them They cannot be totally adaptive because if you were adaptive, you wouldn't be doing anything. The needs are resistive They are pushing us against the world if you wouldn't have all these needs if you wouldn't have this motivational system You would just be doing what's best for you, which means collapse on the ground be a vegetable brought given to gravity Instead you do all these unpleasant things you get up in the morning you eat you have sex you do all these crazy things It's only because the motivational system forces you to The motivational system takes this bunch of matter and makes us do all these strange things just so genomes get replicated and so on and So to do this it is going to build resistance against the world And the motivational system is in this sense forcing us to do all these things by giving us needs And the need has some kind of target value and current value if we have a differential between the target value and the current value We would perceive some urgency to do something about the need and when the target value approaches the current value Get a pleasure signal which is a learning signal if it gets away from it We get a displeasure signal Which is also a learning signal and we can use this to structure our understanding of the world to understand What goals are and so on and goals are learned needs are not To learn we need success and failure in the world, but to do things we need anticipated reward So it's dopamine that makes the brain go round dopamine makes you do things But in order to do this in the right way you have to make sure that the cells cannot produce dopamine themselves If they do this they can start to bribe others to work for them You're going to have something like a bureaucracy in your neocortex Where different bosses try to subdue others to do their own bidding and pitch against other groups in the neocortex It's going to be horrible You want to have some kind of central authority that makes sure that the brains don't produce the dopamine themselves It's only being produced in a very small area and then given out and pass through the system And after you're done with it. It's going to be gone. So there is no rewarding of the dopamine And in our society the role of dopamine is played by money Money is not rewarding itself. It's in some sense a way that you can trade against the reward You cannot eat money You can take it later and get an arbitrary reward for it And in some sense money is the dopamine that makes Organizations and society companies and many individuals do things they do stuff because of money But money and if you compare to dopamine is pretty broken because you can hoard it So you're going to have these cortical columns in the real world which are individual people or individual corporations They're hoarding the dopamine. They sit on this very big pile of dopamine. They're starving the rest of the Society of the dopamine They don't give it away and they can make it do its bidding. So for instance, they can pitch a substantial part of society against understanding global warming because their profit of global warming or a lot of technology that Reads to global warming, which is very bad for all of us So our societies have a nervous system that lies to itself How can we overcome this? Actually, we don't know for to do this We would need to have some kind of centralized top-down reward motivational system We have this for instance in the military. You have the system of Military rewards that you get and these are completely controlled from the top also within working organizations You have this in corporations. You have centralized rewards. It's not like Rewards flow bottom up. They always flow top down And there was an attempt to model society in such a way that was in chile in the early 1970s The alianna government had the idea to Redesign society or economy in a society using cybernetics So alianna invited a bunch of cyber neticians to redesign The chilean economy and this was meant to be the control room where alianna and his chief economists would be sitting To look at what the economy is doing We don't know how this would have worked out because we know how it ended in 1973 There was this big push in chile and this experiment ended among other things Maybe it would have worked. Who knows nobody tried it So, uh, there's something else that is going on in people Beyond the motivational system. That is, um, we have social criteria for learning We also check if our ideas are normatively acceptable And this is actually a good thing because individuals may shortcut the learning through communication Other people have learned stuff that we don't need to learn ourselves and we can build on this So we can accelerate learning by many orders of magnitude which makes cultures possible and which makes anything possible because If you are on your own, you're not going to find out very much in your lifetime You know how they say Everything you do you do by standing on the shoulders of giants Or in a big pile of dwarfs. It works either way Social learning usually outperforms individual learning. You can test this but in the case of conflict Uh, between different social truths, you need to have some way to decide who to believe So you have some kind of reputation estimate for different authorities and you use this to check whom you believe And the problem of course is there's an existing society in real society This reputation system is going to reflect power structures which might distort your beliefs systematically Social learning therefore leads groups to synchronize their opinions And the opinions become get another role. They become an important part of signaling which group you belong to So opinions start to signal group loyalty in societies And people in this in that such a world they should optimize not for getting the best possible opinion in terms of truth They should guess uh, they should optimize for doing having the best possible opinion with respect to agreement with their peers If you have the same opinion as your peers, you can signal them your part of the in-group They're going to like you if you don't do this Chances are you're not they're not going to like you. There's rarely any benefit in life to be in disagreement with your boss right so If you evolve an opinion forming system in these circumstances You should be ending up with an opinion forming system that leaves you with the most useful opinion Which is the opinion in your environment and it turns out most people are able to do this effortlessly They have an instinct that makes them adopt the dominant opinion in their social environment It's amazing right and uh, if you are a nerd like me, you don't get this So in the world out there explanations piggyback on your group allegiance For instance, you will find there's a substantial group of people that believes the minimum wage is good for the economy and for you And another one which believes that it's bad and it's pretty much aligned with political parties It's not aligned with different understandings of economy because nobody understands how the economy works And if you are a nerd you try to understand the world in terms of what's true and false You try to prove everything by Putting it on some kind of truth and false level And if you're not a nerd you try to get to right and wrong you try to understand whether you're in alignment with what's objectively right in your society, right so, uh, I I guess that nerds are people that have a defect in your opinion forming system And uh, usually that's maladaptive and in under normal circumstances nerds would mostly be filtered from the world Because they don't reproduce so well because people don't like them so much And then something very strange happened the computer revolution cable long And suddenly if you argue with the computer, it doesn't help you have if you have the normatively correct opinion You need to be able to understand things in terms of true and false, right? So now we have this strange situation that the weird people that have these offensive strange opinion and that really don't mix well with Real normal people, uh get all these high paying jobs and we don't understand how how is that happening And it's because suddenly our male adaptation is a benefit But out there there is this world of the social norms and it's made of paper walls There are all these things that are true and false in a society that make people behave. It's like these japanese walls that they made palaces out of paper basically and These are walls by convention that exist because people agree that this is a wall And if you are a hypnotist like donald trump you can see that these are paper walls and you can shift them And if you are a nerd like me you cannot see these paper walls if you pay closely a close attention You see that people move and then suddenly In mid-air they make a turn Why do they do this? There must be something that they see there and this is basically a normative agreement And you can infer what this is and then you can manipulate it and understand it Of course, you can fix this you can debug yourself in this regard But it's something that is hard to see for nerds. So in some sense they have a superpower They can think straight in the presence of others, but often they end up in the living room and people are upset Learning in a complex domain cannot guarantee that you find the global maximum You can you know that you cannot find truth because we cannot Declanize whether we live on a playing field or in a simulated playing field But um, what we can do is we can try to approach a global maximum But we don't know if that is the global maximum We will always move along some kind of belief gradient We will take certain elements of our belief And then give them up and for new elements of a belief based on thinking that this new element is better than the one that we give up So we always move along some kind of gradient and the truth does not matter the gradient matters If you think about teaching for a moment When I started teaching I often thought okay, I understand the truth of the subject The students don't so I have this to give this to them And at some point I realized oh, I've changed my mind so many times in the past I'm probably not going to change it to stop changing it in the future I'm always moving along a gradient and we'll keep moving along a gradient. So I'm not moving To truth. I'm moving forward And when we teach our kids we should probably not think about how to give them truth We should think about how to put them onto an interesting gradient that makes them explore the world world of possible beliefs And these possible beliefs lead us into local minima. That's inevitable These are like valleys and sometimes these valleys are neighboring and we don't understand what the people in the neighboring valley are doing Unless we are willing to retrace the steps. They've been taken And if you want to get from one valley into the next We will have to have some kind of energy that moves us over the hill They have to have a trajectory where every step Works by finding a reason to give up a bit of our current belief and adapt a new belief because it's somehow more useful more relevant More consistent and so on Now the problem is that this is not monotonous We cannot guarantee that we are always climbing Because the problem is that the beliefs themselves can change our evaluation of the belief It could be for instance that you start believing In a religion and this religion could tell you if you give up the belief in the religion You're going to face eternal domination in hell As long as you believe in the religion it's going to be very expensive for you to give up the religion Right if you truly believe in it. You're now caught in some kind of attractor Before you believe the religion it's not very dangerous, but once you've gotten the attractor it's very very hard to get out So these belief attractors are actually quite dangerous You can get not only too chaotic behavior that you cannot guarantee that your current belief is better than the last one But you can also get into beliefs that are almost impossible to change And that makes it possible to program people to work in societies Social domains are structured by values Basically a preference is what makes you do things because you anticipate pleasure or displeasure And values make you do things even if you don't anticipate any pleasure These are virtual rewards They make us do things because we believe that there's stuff that is more important than us This is what values are about And these values are the source of what we would call True meaning deeper meaning There's something that is more important from us for than us something that we can serve This is what we usually Perceive as a meaningful life It's one which is in a service of values that are more important than I myself because after I'm not that important I'm just this machine that runs around and tries to optimize its pleasure and pain. It's just kind of boring So my PI has puzzled me my principal investigator in the Harvard department where I have my desk Martin Novak, he said that meaning cannot exist without God. You're either religious or you are a nihilist And this guy is the head of the department for evolutionary dynamics Also, here's the Catholic So, uh, this really puzzled me and I try to understand what he meant by this Typically if you are a good atheist like me Um You tend to attack gods that are structured like this religious gods that are institutional They are personal that there are some kinds of person. They do care about you They prescribe norms. For instance, don't masturbate. It's bad for you Money of these norms are very very much aligned with societal institutions For instance, don't question your question the authorities. God wants them to be ruling above you And uh be monogamous and so on and so on So they prescribe norms that do not make a lot of sense in terms of Being that creates worlds every now and then but they make sense in terms of what you should be doing to be a functioning member of society And this god also does things like they create roles They like to manifest as burning strawberry and so on there are many books that describe stories that these gods have allegedly done And it's very hard to test for all these features which makes these gods very improbable for us and Makes atheists very dissatisfied with these gods But then there is a different kind of god. That is what we call the spiritual god This spiritual god is independent of institutions. It still does care about you. It's probably conscious. It might not be a person There are not that many stories that you can consistently Tell about it, but you might be able to connect to it spiritually Then there is a god that is even less expensive. That is god as a transcendental principle And this god is simply the reason why there is something rather than nothing This god is the question that the universe is the answer to this is the thing that gives meaning Everything else about it's unknowable. This is the god of thomas of Aquinas The god that thomas of Aquinas discovered is not the god of abraham. This is not the religious god It's a god that is basically a principle that asks the universe into existence It's the one that gives the universe its purpose And because every other property is unknowable about this this god is not that expensive Unfortunately, it doesn't really work. I mean thomas of Aquinas try to prove god He tried to prove a necessary god a god that has to be existing and I think we can only prove a possible god So if you try to prove a necessary god, this god cannot exist Which means your god proof is going to fail. You can only prove possible gods Then there is an even more impoverished god and this is the god of Aristotle And he said if there is change in the universe something is going to have to change it There must be something that moves it along from one state to the next So I would say this is the primary computational transition function of the universe And Aristotle discovered it. It's amazing, isn't it? We have to have this because we cannot be conscious in a single state We need to move between states to be conscious. We need to be processes So we can Take our gods and sort them by their metaphysical gods The first degree god would be the first mover second degree god is the god of purpose and meaning The third degree is the spiritual god and the fourth degree god is bound to religious institutions, right? So if you take this statement by martin ovec, you cannot have a meaning without god I would say yes, you need at least a second degree degree god to have meaning, right? So objective meaning can only exist with a second degree god And subjective meaning can exist as a function in a cognitive system, of course We don't need objective meaning So we can subjectively feel that there's something that's more important to us And this makes us work in society and makes us perceive that we have values and so on But we don't need to need to believe that there is something outside of the universe to have this So the fourth degree god is the one that is bound to religious institutions It requires a belief attractor and it enables complex norm prescriptions If my theory is right, then it should be much harder for nerds to believe in a fourth degree god than for normal people And what this god does it allows you to have state building mind viruses Basically, religion is a mind virus and the amazing thing about these mind viruses is that they structure behavior in large groups We have evolved to live in small groups of a few hundred individuals. Maybe something like 150 This is roughly the level to which reputation works We can keep track of about 150 people and after this it gets much much worse So in this system where you have reputation people feel responsible for each other And they can keep track of their doings and society kind of sort of works If you want to go beyond this you have to write a software that controls people And religions were the first software that that did this on a very large scale And in order to keep stable they had to be designed like operating systems in some sense They give people different roles like insects in the hive And they have even a part of these these roles is to update the religion But it has to be done very carefully and centrally because otherwise the region will split apart and fall together into new religions or overcome by new ones So it's some kind of evolutionary dynamics that goes on with respect to religion And if you look at the religions, there's actually a veritable evolution of religions. So We have this Israeli tradition and the mesotomic mythology that gave rise to Judaism It's kind of cool, right? Also history totally repeats itself It totally blew my mind when I discovered this Of course the real tree of formal language is slightly more complicated and the real tree of religion is slightly more complicated But still it's neat So if you want to immunize yourself against mind viruses first of all you want to check yourself whether you are infected You should check. Can I go let go of my current beliefs without feeling that meaning departs me and I feel very terrible if I let go of my beliefs Also, you should check All the other people around there that don't share my belief Are they either stupid or crazy or evil? If you think there's chances are you are infected by some kind of mind virus because they are just part of the outgroup And um does your god have properties that you know, but you did not observe So basically you have a god of a second or third degree or higher In this case, you probably also got a mind virus. There's nothing wrong with having a mind virus But if you want to immunize yourself against this people have invented rationalism and enlightenment basically To act as immunization against mind viruses And in some sense, it's what the mind does by itself because if you want to understand how you go wrong You need to have a mechanism that discovers who you are some kind of auto debugging mechanism that makes the mind aware of itself And this is actually the self So according to robot keegan the development ourself is a process in which we learn who we are by making things explicit By making processes that are automatic visible to us and to conceptualize them So we no longer identify with them and it starts out with understanding that There's only pleasure and pain if you are a baby you only have pleasure and pain you identify with this And then you turn into a toddler and the toddler understands that they're not their pleasure in pain But they are their impulses And in the next level if you grow beyond the toddler age you actually know that you have goals And that your needs and impulses are there to serve goals, but it's very difficult to let go of the goals if you are very young young child and At some point you realize oh the goals don't really matter because sometimes you cannot reach them But we have preferences we have things that we want to happen and things that we do not want to happen And then at some point we realize that other people have preferences too And then we start to model the world as the system where different people have different preferences And we have to navigate this landscape and then we realize that these preferences also relate to values And we start to identify with these values as members of society And this is basically the stage if you are an adult being that you get into And you can get to a stage beyond that especially if you have people around this which have already done this And this means that you understand that people have different values And what they do naturally flows out of them and these values are not necessarily worse than yours They're just different and you learn that you can hold different sets of values in your mind at the same time Isn't that amazing and understand other people even if they're not part of your group If you get that this is really good But I don't think it stops there You can also learn that the stuff that you perceive is kind of incidental That you can turn it off and that you can manipulate it And then at some point you also can realize that yourself is only incidental That you can manipulate it or turn it off and that you're basically some kind of consciousness that happens to run on the brain of some kind of person That navigates the world in terms to get rewards or avoid displeasure And serve values and so on but it doesn't really matter There is just this consciousness that understands the world and this is the state that we could typically call enlightenment In this state you realize that you are not your brain But you are a story that your brain tells itself So becoming self-aware is a process of reverse engineering your mind It's a different set of stages in which realize what goes on So isn't that amazing ai is a way to get to more self-awareness I think it's a good point to stop here The first talk that I gave in this series was two years ago. It was about how to build a mind Last year I talked about how to get from basic computation to consciousness and this year We have talked about finding meaning using ai I wonder where it goes next Thank you for this amazing talk. We now have some minutes for q&a. So please line up at the microphones As always if you're unable to stand up for some reason Please very very visibly raise your hand. We should be able to dispatch dispatch an audio angel to your location So you can have a question too and also if you're locationally disabled if you're not actually in the room If you're on the stream, you can use isc or twitter To also ask questions. We also have a person for that We'll just start on microphone too Wow, that's me Just a guess what would you guess when can you discuss Your talk with a machine in how many years? I don't know as a software engineer. I know if I don't have a specification All bets are off until I have the implementation And so it can be of any order of magnitude. I have a gut feeling But I also I know as a software engineer that my gut feeling is usually wrong Until I have the specification So I the question is if there are silver bullets Right now there are some things that are not solved yet And it could be that they're easier to solve than we think but it could be that they're harder to solve than we think Before I stumbled on this cortical self organization thing I thought it's going to be something like maybe 60 80 years and now I think it's way less But again, this is a very subjective perspective. I don't know Number one, please Yes, I wanted to ask a little bit about meta cognition It seems that you kind of end your story Saying that it's still reflecting on on the input that you get and kind of working with your social norms and this and that But colberg for instance talks about What he calls a post cognitive post conventional universal morality for instance, which is Thinking about moral laws without context basically stating that there's something beyond The relative norms that we have towards each other, which Would only be possible if you can do kind of You know meta cognition thinking about your own thinking and then modifying that thinking So kind of feeding back your own ideas into your own mind and and coming up with stuff that actually can't get Thinking about while processing external inputs I think it's very tricky this project of defining morality without societies Exists longer than can't of course and can't try to give these eternal rules and others try to I find this very difficult from my perspective. We are just moving bits of rocks And these bits of rock they are on some kind of dust mode in the galaxy Out of trillions of galaxies and How can there be meaning it's very hard for me to say That's one chimpanzee species is better than another sympathy species or a particular monkey is better than another monkey This only happens within a certain framework and we have to set this framework And I don't think that we can define this framework outside of a context of social norms that we have to agree on So objectively, I'm not sure if we can get to ethics I only think that's possible based on some kind of framework that people have to agree on implicitly or explicitly Microphone number four, please Hi, thank you. It was a fascinating talk. Um, I have two thoughts That went through my mind and the first one is that um, it's it's so Convincing the models that you present, but it's kind of like You present another metaphor of understanding the brain Which is still something that we try to grasp on different levels of science, basically And and the second one is that um That your definition of the nerd who walks around and doesn't see the the walls is kind of a definition or remind me of Richard Wurte's definition of the ironist, which is a person who knows that their vocabulary Vocabulary is finite and there's other people who also have a finite vocabulary And then that obviously opens up the whole question of meaning making which has been discussed in so many other Disciplines and fields and I thought about Derry does deconstruction of ideas and thoughts and and butler and then down the rabbit hole to Nietzsche I was just wondering if you could maybe map out other connections where basically not AI helping us to understand the mind but where Already existing a huge huge field of science into under like cognitive processes coming from the other end could help us to understand AI Thank you The tradition that you mentioned Rorty and butler and so on Are part of a completely different belief attractor in my current perspective That is they are mostly social constructionists that they believe that reality At least in the domains of the mind and sociality are social constructs There are a part of social agreement Personally, I don't think that is the case. I think that the patterns that we refer to are mostly independent of our mind The norms are part of social contract constructs But for instance our motivational preferences that make us adopt or reject norms are something that Builds up resistance to the environment. So they're probably not part of a social agreement and The only thing that I can invite you to is Try to retrace both of the different belief attractors try to Retrace the different paths on the landscape all the things that I tell you a lot of this is of course very speculative These are things that seem to be logical to me at this point in my life And I try to give you the arguments why I think that they are plausible But don't believe in them question them challenge them see if they work for you. I'm not giving you any truth I'm just going to give you suitable encodings according to my current perspective Thank you The internet please So someone's asking If in this belief space you were talking about Um How do we how is it possible to get out of local minima? and Very related question as well Uh, should our should we teach some sort of Momentum method to our children so they don't get stuck in local minima I believe at some level it's not possible to get out the local minima In an absolute sense because you'll only get to get into some kind of meter minimum But uh, what you can do Is to retrace The path that you took whenever you discover that somebody else has a fundamentally different set of beliefs And if you realize that this person is basically a smart person that is not completely insane But has reasons to believe in their beliefs and that they seem to be internally consistent It's usually versed to retrace what they have been thinking and why and this means You have to understand what their starting point was and how they move from their current point to their starting point You might not be able to do this accurately And the important thing is also afterwards you discovered a second valley You haven't discovered the landscape in between But the only way that we can get an idea of the lay of the land is that we try to retrace as many paths that possible And if we try to teach our children what I think what we should be doing is To tell them how to explore this world on their own is not that we tell them. This is the valley this is Basically, it's given it's the truth But instead we have to tell them this is the path that we took And these are the things that we saw in between and it's important to be not be completely naive when we go into this landscape But we also have to understand that it's always an exploration that never stops And that might change everything that you believe now at a later point So for me, it's about for teaching my own children how to be explorers How to understand that knowledge is always changing and it's always a moving frontier We are unfortunately out of time so please once again, thank you, chef