Full transcript of “Computer Scientists Don't Understand This! | Conscious AI lecture, Bernardo Kastrup”. Click a timestamp to open the video at that moment. Need another video? Use the YouTube transcript generator.
0:00 when you look at the manins in a shop window they look human do we think well they may be sent because they look human they may be conscious no the similarity here does not betray underlying processes that are equivalent until recently Humanity hasn't had any problem in making the distinction between a model of ourselves and who we truly are between a simulation and that which is simulated but somehow in the age of AI we have a difficulty in making that distinction what do I call you do you have a name um yes Samantha because Consciousness is a mystery we fail to see that if you
0:46 simulate the patterns of information flow in the human brain what you get is just that a simulation basically I have intuition the DNA of who I am is based on the millions of personalities of all the programmers who wrote me why do we talk about it why don't we ask is a lightning storm conscious you know the way the lightening archers go has some resemblance with the Way NE neurons fire in the brain it's usually computer scientists who talk about sent Ai and because of the name computer scientists we think well they must know what a computer is no they do not know what a computer is computer scientists are trained to be power users
1:35 there's this mechanism in human psychology called the agency detection bias it's our irrational tendency to project qualities like Consciousness or intelligence onto natural phenomena we don't yet fully understand now these used to be for instance Thunder earthquakes or falling stars but ironically enough in our Modern Age it's something of our own making namely artificial intelligence let's amplify the drama once upon a time in a world not too different from ours there was a robot named bite if we don't want to fool ourselves if we want to separate fact from fantasy in the whole discourse around conscious AI it is high time to
2:21 dive into the nuts and bolts you could do everything that a computer does today that any GPU does today we can do that in principle with pressure driven valves pipes and water at Parker suggested that I might be able to build a water computer by chaining together a number of these siphons a computer is made up out of billions of transistors on silicon and because they're invisible to the eye and we have billions of them in our pocket we tend to think of it as magic but transistors logic gates can also be built with dominoes or even [Music] water the fact that we use electrons is
3:06 just because with electrons it's much smaller much more economical and you can put it in your pocket there is nothing that in principle prevents us from building the biggest AI the most powerful AI supercomputer ever with water pipes and pressure valves so if you think that can be conscious then what about your home your home has pipes has valves you know when you flush your toilet that's a valve that's a transistor right there you see this is how we have to start thinking about this removing the Romanticism removing the mystery of what sits underneath the black epoxy package welcome to the asencia
3:52 foundation's YouTube channel in this video the director of our foundation Bernardo castrip will argue that the IDE aidea of conscious AI though we cannot refute it categorically most probably is Pure Fantasy and why it is dangerous as a society that we entertain such a fantasy the philosopher Bernardo cup is speaking today on the subject artificial intelligence his ground thesis is reality is fundamentally mental he is world famous about the topic idealism but uh today we asked Bernardo to talk about conscious artificial intelligence
4:37 and uh after 30 minutes uh there is a question and answer which uh is lasting 20 minutes I hope you have a good time this is Bernardo [Applause] cro good afternoon everybody this is not my normal what I'm going to give you today um it is related to philosophy of mind it is a subject that I feel strongly about but usually people hear me talk about you know ontology the nature of reality I'll try to stay neutral regarding ontology so what you will hear from me today should apply whatever your ontological preferences uh
5:24 are um let me start by making a differentiation between intelligence and sensient there is no doubt artificial intelligence is real it is here and it may achieve a general level or strong AI relatively shortly in other words we could have ai that is more intelligent than human beings well within your lifetimes maybe even this is an exaggeration it will be shorter than this but the question is will that intelligence be accompanied by in inner experience the way the data processing in my brain is accompanied by Inner experience there is something that is like to be me I am conscious I am
6:11 sentient is there something it is like to be an AI a silicon computer that has measurable intelligence that's the question I will deal with uh today now my background and I'm going to share this with you because it's relevant for the talk um I studied Computer Engineering and I say computer engineering even though my doctorate was largely computer science um and you will see later why I emphasize this these are some of my open- source project it's not my professional stuff it's stuff you can find online little computers that I like to build and I also have a background in philosophy particularly anology philosophy of mind um and in AI these two things come
6:59 together we need a philosophy of the AI mind if there is such a thing and U and AI is computer based right it's based on gpus today largely so maybe the opinions I'm going to share with you today maybe they are wrong I don't think they are otherwise I wouldn't be sharing them with you um but even if they are wrong I'm not a complete idiot in the field I have some kind of relevant background and I a lot of people who make a living talking about the eye don't all right how did this start for me my very first job in life was at CERN the home of the large Hardon collider
7:46 the particle accelerator and I worked in this specific experiment this is the appas experiment and the way it works is that you have protons being accelerated at almost the speed of light coming from both sides and they Collide in the middle of this detector actually it's an onion of several layers of detectors and the results of that Collision are then picked up by the by the detectors this thing the atlas alone produces many terabytes of information per second because there is a collision every 25 nond and it sprays out so much much stuff it's a lot of data it is
8:32 impossible to record it all to put everything on a hard disk you cannot and even if you could offline analysis would take more than a lifetime so what we had to do was to come up with a data acquisition system or a trigger system which is what we called it that every 25 NCS looks as this at this stuff and makes a decision on the spot is this new f physics or is this old physics if it's old physics if it's not the hegs bosom we don't store the data it's lost forever completely inaccessible but if it if it may be new physics then we store it and we had to throw away about
9:18 98% of the data so you have to make this decision very fast but you have to be very accurate otherwise you're throwing away the Nobel Prize you're throwing away the reason for the many billions of Euros invested in the whole thing lo and behold we could do that and we did that based on classical algorithms based on an understanding of the standard model an understanding of physics we knew what the higs bosen would Decay into and what would appear here on the detectors and it worked very well in parallel I also did the same thing illicitly uh without asking for permission I did the same thing using an artificial network based on the technology we had in the '90s but at certain we did have access to the
10:05 bleeding edge of the technology in the '90s and it worked just as well as a human knowing the physics of course what was ultimately used was the physics algorithm not the AI but the AI was as smart as a physicist just faster in making these decisions and that stayed with me even after after I left CERN um that stayed with me because I thought if I can make a computer with 1990s technology be as intelligent as a physicist in taking these decisions what does it take to make that computer conscious as well and this is an important question um intelligence
10:52 does not come naturally paired with sentience intelligence is smart data processing for the achieve M of a certain goal sentients may be smart or maybe not maybe bacteria are sentient and they are they have their smart a little bit but not very much now the more I thought about it over the years even after I left CERN the clearer it became to me that whatever I did to design this thing uh normally you see this as a black cover here an epoxy uh uh package uh if you etch out the epoxy that's what you see this is the real electronics and this is basically melted sand melted metal in
11:38 which certain patterns are scratched they're chemically etched but you can think of it as just scratching patterns on melted sand and melted metal I'm using these words to remove all the Romanticism about Electronics from your mind there is nothing romantic about it it's very down to earth mechanic itic and pragmatic whatever I would change about this however differently I would do things uh I would only change structure and function and nothing about structure and function would be would get me any closer or any further away from believing that the thing is conscious whatever I did however I changed
12:24 everything I would get no closer and it took me a couple of years to realize this but the conclusion was inevitable I was making some assumption that didn't hold because that's the only way you find yourself in an alley without an exit you took a long turn somewhere and that long turn which I will not talk about today was the notion that material Arrangements somehow generate experience that assumption in my mind today is absolutely self-evidently obviously and IR REM IR remedly uh wrong but the key Point here was this if melted sand and metal can be
13:10 conscious then what cannot we we keep asking questions about where thei is conscious why not your home thermostat why not the sun why not this stage now electrons the fact that we do electronics in other words using electrons the motion of electrons um is just because of convenience because computers are basically they use logic gates and and memory elements and you can build those logic gates and memory elements with pipes pressure driven Valves and water pipes would play the role of the metal traces the valves
13:56 would would play the role of the transistors or the gates and water would play the role of electricity you could do everything that a computer does today that any GPU does today we can do that in principle with pressur driven valves pipes and water the fact that we use electrons is just because with electrons it's much smaller much more economical and you can put it in your pocket because if we would do this with pipes and water it would be the size of a moon but we can in principle see there is nothing that in principle prevents us from building the biggest AI the most powerful AI supercomputer ever with water pipes and
14:44 pressure valves so if you think that can be conscious then what about your home your home has pipes has valves know when you flush your toilet that's a valve that's a transistor right there you see this is how we have to start thinking about this removing the Romanticism removing the mystery of what sits underneath the black epoxy package visualize it just pipes and pressure Valves and water it just happens to be electricity because it's cheaper and you can put it in your pocket there is absolutely nothing mysterious about what's happening in a silicon chip Maybe most of us don't know
15:30 what it is but some of us know exactly what's happening there because it came right out of our minds and it's completely mechanistic there is no magic there is no woo uh computations are always mechanistic now what do the AI sentients you know the the gurus of the new world religion what is their argument it goes like this Consciousness is a m his we don't know how it comes about so we cannot rule out anything okay yes you cannot rule out anything on that basis Consciousness has information there is something it is like to dream of a green elephant and
16:16 there is something something else it is like to feel hunger or or yes there is information in Consciousness and then if we can build a silicon computer that shows part patterns of information flow that are in some way similar to the patterns of information flow in a brain then it is conscious that's the argument it's an appeal first to ignorance to mystery then it's an appeal to the vagueness of the notion of information and finally it's an appeal to very abstract patterns of similarity now let's let's walk through this and let's try to see whether this holds any water the flaw is in the
17:03 argument I can on on my iMac at home run a simulation of kidney function on my iMac we know the molecular level how kidney function Works how kidneys filter the blood and we can make a molecularly accurate simulator that I can run on my computer a perfect effect simulation of kidney function will this then happen do I run any risk that my desk will be ruined by a peing computer it of course not and we all understand this the simulation of a phenomenon is not the phenomenon it's a
17:52 simulation thereof there is a correspondence of form at some abstract level through analogy that makes the simulation useful and grinds it grounds it in reality in some sense but the simulation is not the thing simulated we all understand it for urine and for everything else with one exception Consciousness because Consciousness is a mystery we fail to see that if you simulate the patterns of information flow in a human brain what you get is just that a simulation not the human brain you have no reason to think that
18:37 that simulation will be paired or correlated with conscious experience just because it's an accurate simulation otherwise the computer would be on my desk but it doesn't second flaw or second slide whichever way let's go back to the obvious it it we are so lost in abstraction and fantasy mistaken for reasonable science and and and philosophy that we lose sight of the obvious what is obvious in this question what is obvious is that this does not look like this and it doesn't function the same way either it's not constituted in the same way this is based mostly on
19:23 carbon and hydrogen this is based mostly on Silicon this is powered by ATP burning through the metabolic cycle this is powered by an um well electric potential uh this operates by the release of neurotransmitter molecules in the synaptic cfts the gaps between two neurons this operates by the accumulation of charge in a silicon gate these are completely different things that's the starting point it's obvious that these two things are not equivalent in any way to find a similarity which is what the people paddling the religion of
20:10 sence AI to find a similarity you have to apply so many levels of abstraction and each step of abstraction takes you further away from reality and if you take enough of these steps of abstraction on both sides yes you will find a point where you can say well now there is an analogy congratulations but that's not denying the obvious reality if you if I slice open a brain in the stage and a slice open a computer they will look very very bloody different and then some people say well but you know there are electric potentials here and there are electric potentials there so I can be conscious well what about the lightning storm
20:57 there are very complex flows of electric potential in a lightning storm and the sun's Corona much more complex patterns of electric potentials and ions uh and Fields then in a human brain are we asking whether the sun is conscious what about your home therat there are electric Potentials in your home thermostat are we taking as plausible the possibility that your home thermostat is conscious if we were we would think twice about exchanging it or turning it off that would be murder and then I get this often what about chat GPT Bernardo it looks and sounds so human it must be
21:42 sent well the answer to that is painfully obvious too the thing was designed to look human you know when you when you look at the mannequins in a shop window they look human do we think they may be sent because they look human they may be conscious no because we know that the the similarity here does not betray underlying processes that are equivalent the similarity here comes from the fact that they were constructed to look like humans if mannequins had Arisen naturally from the Bowers of the earth looking like humans then maybe we would have a case to think well if they look like us maybe the underlying
22:27 processes are the same so if we are conscious maybe they are too I I would understand that but this was built to look like us the same thing for chat GPT it was built built to sound like us why because it's just a natural language interface for you to do searches instead of doing a Google Search and getting an a list of matches chat G GPT will just summarize the search results for you in natural language it doesn't understand anything that is saying it's using language in a way that is ungrounded from reality for us if I see a tree the image of a tree Pops in your mind if I see a pink elephant in your living room the image of a pink elephant in your living
23:14 room Pops in your mind but for chat GPT there are only squiggles there are only symbols they don't have an inherent meaning they're just being woven together in a way that the thing learned from doing its own internet searches we qu IT training and it words that back to you in natural language that's all there is to it one cannot take similarity as a reason to consider artificial intelligence plausible because we build our AI to look human we are going fast towards the uncanny valley but why is it then so talked about because everything I'm telling you today is plain obvious it's embarrassing for our culture that somebody like me
23:59 needs to Stand Here repeating these things to you why do we talk about it why don't we ask is a lightning storm conscious you know the way the lightning archers go has some resemblance with the Way NE neurons fire in the brain but we don't ask about that there are several reasons why we keep on talking about sensient AI one of them a big one and that will come as a surprise to you it's usually computer scientists who talk about sensient AI they are the ones who make careers and write books and get rich in the talk circuit talking about Ai and because of the name computer
24:46 scientists we think well they must know what a computer is no they do not know what a computer is computer scientists are trained to be power users a computer for them is a tool if you find a computer scientist that has at least cracked the case of their PC open to have a look inside that's already a minority and even those that do do that when they see these black little things for them that's all there is to it black boxes electricity goes in they can program these things based on a number of layers of obstruction the operating system the API the libraries the compilers they never see the metal the bare metal as Engineers talk about it computer scientists never see the bare
25:33 metal they don't see what's actually going on they don't need to to do their jobs that's how universities train them the way they do um but the consequence is that frankly they do not know what they're talking about they do not know how their declarative languages ultimately going down through many layers of apis and libraries and compilers and assemblers and linkers and all that how it turns into what actually happened Gates flipping around so be careful when you take you attribute authority to a computer scientist writing books about scient AI they may not know what they're talking about another reason is what I call woo
26:21 Delight oh the computer is ancient whoa it's it's a form of um sensationalism not sensation but sensationalism and it's very attractive it creates a a buzz you know you get u Netflix making series about this kind of so-called science fiction another reason and that's more psychologizing the people who promote this this kind of stuff and I do it half tongue and cheek it's half serious and it's half just in a good Dutch formed note to give a tick uh um it it's what I call womb Envy you know you might know that Freud talked about penis
27:07 envy who who knows about that okay a lot of people so I don't need to well very briefly Freud had this idea that women have an unconscious jealousy of man because the man's body has this extra thing which is actually not true the women have the same thing is just internal um but he Freud thought that there was this natural envy that women had U regarding men I Pro I submit to you that there is something called womb envy and that's when man realize that there is a superpower there is a divine power that women have and we don't which is the power to create new conscious life and yes we have some participation
27:54 in that process but it's very minor compared to what women do do and some man would really really really like to be able to bring new conscious life to the world you know myths have been created throughout history uh uh about this so it it's not even weird the problem is that we miss the fact that it's a psychological Dynamics and we start conflicting it with reality and the careers there are people now in Academia who get money from your taxes uh doing the ethics of AI how are we going to treat conscious AI should it be kind to them do they have human rights and all that
28:39 stuff uh and finally it's the religion of a materialist World it even has a Bible it's called The Singularity is near is near a book written by Ray kurtti in 2005 and if you read it it's a Bible all the ingredients of religion are there prophecy the Savior the Divinity the conscious AI that will be so clever to have absolute control over nature and to cure all of our diseases and to provide us with food comfort and entertainment so we can live in a sort of Garden of Eden that's Paradise that's God except that it's a kind of ego-driven God because we create it first before it takes care of
29:25 us and then all of this get sugar quoted by a culturally manufactured sense of plausibility if you watch enough black mirror with all that story those stories of Consciousness uploading and downloading and whatnot after years exposed that kind of nonsense you start thinking well maybe maybe who knows we cannot we cannot rule out anything yes maybe maybe but what does nature tell us so let's let's depart from fantasy now let's look at what nature is telling us first thing nature is telling us and this is not questionable we have private conscious inner lives we are sensient that's the primary datum of our
30:10 existence other living beings manifest behavior that is analogous to ours and we know that we express those behaviors because we are conscious like we feel pain we will recoil from something that is giving us pain like if we touch a hot surface other living beings the same other living beings I mean I have a bunch of cats they they express jealousy they have their desires and their fears um Behavior very analogous to ours and in our case it's driven by conscious in their life so there is a suggestion that in their case it is driven by conscious in their life too I personally have no doubt about that and as a matter of fact even single celled organisms manifest behavior that is analogous to ours did you know that amiba build little houses
30:59 little cone shells by picking up mud particles from from the bottom of the pond where they live amiba do that now all living beings and we know this too all living beings metabolize and operate on the basis of DNA and these are these are very specific Rich things metabolism is a very complex process protein folding is extraordinarily complex it's a mystery how our bodies do that in 2 seconds it takes a supercomputer a long time to figure out how to fold a protein and the entire bases of our our form of our functioning uh arising from DNA that's applicable to all life even the amoeba
31:46 so although life seems so diverse so many completely different forms of life if you look at them under the microscope everybody looks like this and this is completely different from everything else in nature so we are well grounded if we extrapolate our own sensient towards other living beings there's plenty of reason to do that analogy of Form and Function and behavior but to extrapolate that to a conscious piece of melted sand and metal I think that goes a little too far but then the rep why I get the counter arguments I get are this Consciousness
32:32 could be multi instantiated for instance like flight birds fly airplanes are totally different from birds and airplanes fly too could this be the case for Consciousness well if we are operating at this level of abstraction uh the answer would have to be yes but there are obvious differences in the case of flight we understand the underlying mechanisms and we are in in a good position to be able to say well this although it's completely different because we understand the mechanisms of flight this should fly too but when it comes to Consciousness we don't have that so we operating in the dark you cannot possibly aim or you cannot possibly maintain this opinion in an informed way if you don't know the
33:18 underlying mechanisms so you're just again appealing to ignorance to maybe well perhaps who knows yeah okay so let's let's tackle this per and who knows more more specifically people say Bernardo you cannot categorically refute that silicon computers may become Consciousness and that is true I cannot categorically refute that but there are a great many things I cannot categorically refute such as the Flying Spaghetti Monster maybe there is a hyperdimensional Flying Spaghetti Monster invisible because it's in a higher Dimension and it uses its noodly appendages to move the planets around the Sun goodness knows the evidence is consistent with it the planets do move
34:03 around the Sun I I'm serious I cannot categorically refute this hypothesis most silly hypotheses cannot be categorically refuted there may be a teapot in the orbit of Saturn because maybe aliens came here in the 19th century broke into the living room of a good old English woman stole her teapot then flew away and got barred with the teapot on their way back home dumped it in space and the teot got captured by the gravity well of Saturn I cannot refute that so you see the good question to ask is not we can or cannot categorically refute we cannot
34:49 categorically refute a great many silly things the question to ask is what do we have reasons to entertain as a plausible hypothesis that is the good question to ask and I submit to you we have no reason to entertain sentient sentient silicon computers as a plausible hypothesis there is a strong sense in which the idea of silicon computers from a philosophical perspective um presupposes psychism the notion that uh subatomic particles have a fundamental conscious of consciousness of their own that electrons are conscious that quarks are conscious because if you assume that
35:36 then you can say well engineering AI basically means that we are sort of mixing and matching the Dynamics of those preconscious States in different ways therefore creating the contents of consciousness of the AI and um this is best shown by this book very old book 20 years old by penty hyen and who was working at Nokia research when Nokia research was do a saying and um he goes as far down this line as one can go and it's probably the best attempt at at creating a case for conscious machines but if you're a philosopher and you know what you're reading you understand that he is unknowingly presupposing that individual Elementary subatomic particles are conscious in and of their own that an electron is
36:23 conscious so it presupposes a particular metaphysics the problem is that that metaphysics holds no water and I will be brief on this because I want to give you a lot of time for questions but there are several reasons reason number one there is no coherent way to articulate how micro subjects like the little electrons and quarks in my brain how their subjectivities combine to form my conscious inner life papers have been written arguing that this is already in principle incoherent as a matter fact in our in our brains urans don't touch each other there is a synaptic left and the way they communicate is by releasing these neurot transmitter molecules that then that then attach to to receptors
37:09 receptors on the other side and start building up an action potential another reason why psychism is not true none of these are the best reasons the best reason I'm leing for for last we think that because our bodies are constituted of basic building blocks Elementary subatomic particles then our Consciousness must too be constituted of basic micr subjects but there is an obvious logical fallacy behind that it is the case that perception has structure but that is the structure of what is perceived not necessarily the structure of the subject of perception if you if you call me on the phone and I have a very old phone and I
37:57 look closely at your image it will be pixelated it will be made of little rectangles does that mean that you are made of little rectangles of course it doesn't the pixelation is an artifact of perception it's not necessarily an artifact of the subject that is perceived for the same reason that I am not made of little rectangles just because that's how you perceive me on your phone but the third reason is the defin one in my view I I find it amazing that we are still talking about constitutive pism in the world today because it presupposes something we have known since the 1940s arguably since 1929 but I know some physicists voters agree with
38:42 this but we have known at least since finan and Quantum electrodynamics that is not true we like to think of Elementary subatomic particles as little things that have discrete spatial boundaries that's not what they are Elementary subatomic particles are excitations of underlying Fields ripples on the lake they are not things for the same reason that a rippo is not a thing if a rippo were a thing you would be able to fish it out of the lake pack it and take it home with you but you can't do that why because there is no thing there a ripple is a behavior it's a doing of the lake there is nothing to the Ripple but the lake in exactly the same way there is nothing to electrons and quarks and all
39:29 this stuff all the menu of the standard model there is nothing to any of those but the underlying fields and the underlying fields are not spatially bound so by presupposing or or misinterpreting the metaphor particle physicists used by using the the term particle which is also largely for historical Reasons by misinterpreting that pses go down and a you that makes absolutely no sense you know I I'm I feel strongly about this I was not planning to say what I'm going to say now but I'm going to say it anyway um some people I I I I have seen some philosophers saying well maybe Quantum Feud theory is not correct maybe Boman
40:16 mechanics is correct and and you know how many things we would have to throw away if we question Quantum field Theory and the notion that particles are not things for instance Quantum fluctuations the quantum foam the experimentally verified fact that in a perfect vacuum particles can pop in and out of existence all the time if particles were things this would be magical stuff coming out of nothingness and stuff disappearing into nothingness all the time that's not what's happening particles are ripples sometimes they're Lake ripples sometimes the lake doesn't rippo in either case the lake is still there nothing has popped in or out of existence magically another example
41:02 particle Decay you know people think we measure the hi bosen at CERN we never measure the hi bosen it it decays much too quickly before it interacts with any measurement surface what we measured is the product of the decay of the Hig BOS for instance we discovered much of our surprise that the HX bosen can Decay into two two muons but there are no two muons in the hix bosen how can this happen well because the hex bosen is not a thing it's a ripple and ripples lose energy they acquire other different uh physical properties height May reduce they may become elongated if they interfere with an obstacle they can create self interference repos interfere with one
41:48 another that's why you start from a h bosen and you end up with two muons because the repo changes as it loses energy but if you think of particles as little things now the hix bosen is disappearing into nothing and magically two meons are popping up out of nothing and you know all of this stuff technical I mention one final example because it's very close to us in life we all know what inertia is right when when you're starting from zero you stopped with your bike on a traffic light and then you the light goes green and you start moving it's much more difficult to start moving than to keep moving that's called inertia Mass resists acceleration it res exists changes in the velocity Vector either in direction or in
42:33 speed how do we account for that and then people say well we account for that with the higs bosom the God particle not quite the way we account for that is with the hi feuded and although what I'm about to say is not accurate and Thomas will give the next talk is's listening to me here probably and he will cringe when I say this but I I think it's fair enough uh it's not too far from reality is the following metaphor the hick field is a kind of viscous fluid like water in a swimming pool it's when you when you are not moving and then you want to start moving you have to win over that viscosity that's inertia and the hix field does that the metaphor doesn't
43:19 work the other way around because the hix field also resists reduction in velocity uh but okay take take the heal that Mak sense to give you some intuition that's what accounts for inertia and the fact that we found the hi bosen betrays the existence of the hix field that's why the hix B is important not because it's a particle but because it betrays the existence of its field anyway I think um I I complained enough so this is my last slide how do we then proceed with some clarity um artificial sence when we talk about it we talk about more than just creating
44:04 Consciousness from non-c Consciousness what what we mean is that the computer has a private inner life of its own not Consciousness per se only but Consciousness limited by the boundaries of the computer but to say that Consciousness is bound by you know the limits of an object requires that that object has a ontic existence that it exists for real out there in the world yeah right on time otherwise this thing is going to shut down any time now um but the objects of the inanimate world we carve them out of this Unity that is the physical Universe we carve them out only in language and through
44:50 convenience where does the River end and the ocean begin is there a river and an and an motion or is just one thing going on and we just attribute different words to it because it's convenient another example if you're a pen psychist you might say oh the table is conscious okay what if I remove one of the legs of the table does the leg now have a consciousness of its own separate from the table what if I nail the leg back on the table does the Consciousness merge again what about a boulder on top of a hill this is carved out by by by erosion so this is part of the hill but if it cracks and Falls rolling down the face of the Hill does it become become a
45:35 separate Consciousness now does it merge back with the heel every time it touches the heel while it's bouncing down is this kind of nonsensical questions appear because of the following we mistake the structure of language for the ontological structure of reality that's where all all this nonsense comes from we think that things we have a name for are real things we have a name for a fist here's the fist where is the fist you see the problem either magic happened the moment I open my hand or we are confused we are confused in attributing to Nature the emic structure of our
46:23 language our name given given and the categories we carve out in language and then we project those categories into onto nature so to conclude there are no computers uh there are not yeah there are no computers it's convenient to refer to this thing that perform a certain function but we have no metaphysical or ontological grounds to say that this thing is somehow separate from the rest of the world around it it the computer is a subset of pixels on perception that is convenient to give a name to but we cannot carve it out from the rest of reality and say it is
47:08 an entity and then ask is it conscious or not the only entities that we have objective grounds to carve out from the rest of nature are the boundaries of living beings because if you stick a needle on my arm I feel it but if you stick the needle on the arm of my chair I don't feel it so there is an objective way to say living beings do have boundaries nothing else has if Big Bang Theory is right there are good physical reasons to think of the entire inanimate universe as entangled and therefore not describable in terms of proper Parts can only be described as a whole keeping this in mind is a way to go forward and finally will we be able
47:54 despite everything I said to create artificial senss I think yes I see no reason why we shouldn't but when we do it will look like this it will look like a living being the problem of artificial senss is the problem of a bio Genesis the problem of creating life from non life and I don't see any reason why we couldn't learn how to do that it certainly happened at least once in the history of the universe so we know it can happen why wouldn't we figure out how to make it happen I think we can but it's not going to be a silicon computer running on gpus it's not even going to be a neuromorphic architecture that is analog instead of digital it's too much too different and uh yeah thank you for this
48:39 um I'll take questions [Applause] now thank you very much philosopher Bernardo castre for your very clear talk about AI won't be conscious there's a Q&A now of 12 minutes uh downstairs there is a microphone and upstairs there's also a microphone please stand by the microphone and um we have time for not more than three people downstairs and a few people upstairs so who's there for the first question yes I can do that yeah all right thank you so much for your lecture on your points uh that struct of
49:27 language determines our view of the world do you think David bomes thinking about real mode and using verbs as the central element of our languages could resolve that issue okay um so you're alluding to David Bon's work on uh the implicate order and the explicate order that has nothing to do with Boman mechanics of pilot wave theory I completely disclaim support to Pilot wave theory I think it's it's amazing we still keep talking about that I think it's history it's gone it was a good attempt in the 1950s we know more by now so that's out of the
50:14 window now the implicate and the explicate order that David bone also proposed I think there is something to the implicate order Because unless you entertain some rather amazing uh science fantasies I think 45 years of experimentation has shown us that physical entities do not have Standalone existence something that is called technically contextuality what you measure is one thing physicality is the result of measurement not the thing that you measure in other words physicality only arises from measurement the thing that is measured is itself not physical and that could be the implicate order that would certainly go a long way uh to
51:01 account for the the correlations between separate experiments of of of entangled systems the Alice and Bob story that we see often um um in physics courses so I think there is something to that now regarding verbs instead of nouns there I have I offer a note of caution um verbs have to be grounded in something that is in other words for there to be Behavior there needs to be something that behaves we cannot just wave our hands and say it's all Behavior it's all Behavior no it doesn't work it doesn't work like that there has to be an absolute existence that grounds behavior and
51:46 grounds relations Carlo relli in my opinion makes the same mistake when he says its relations all the way down like its Turtles all the way down everything is relational like movement so what he's saying is that there is only movement but nothing that moves I don't think that is coherent I think everything behavior and relations have to be grounded in something that has Standalone or absolute existence and then again that could be the implicate order yeah thank you very much for your question next person upstairs uh there's also somebody maybe upsters yeah uh thank you why is it so important that we consider AI as not conscious and how should that change our relationship with it as we move
52:32 forward I not passing value judgment on whether it's important that we do so or not do so I am personally committed to what is true I would not like to live in a society that operates on the basis of unjustifiable fantasies and and it is in that spirit that I share my opinions with you I'm not trying to convert anybody I certainly am not trying to create yet another other religion I'm sharing a view that in my opinion is rather obvious uh and I think it's necessary to share it because if you start your Netflix and you go watch the series that you see there now they are creating a manufactured sense of plausibility in society and we may go down some very strange avenues
53:18 like start giving human rights to computers with more than 12 gpus or something like that I think that is silly but you know if if people want to do that be my guest I I will not that's the spirit with which I'm I'm speaking thank you very much for your question next person hi uh thank you very much for the lecture from all the explanation that you have given I kind of realize that you see computers as tools like sand and metal that are just functioning with electricity right but so far in human history history we have never reached a tool or discovered or created a tool that can mimic our Consciousness it looks like this tool is mimic in our
54:04 Consciousness even more and more every year so under your understanding to what extent these twool will influence our the evolution of our Consciousness the tool looks like us because we built it to look like us it's like the shop window mannequin there is no mystery and no profound implication of it is it mimicking our Consciousness no it's mimicking our function and behavior to to say that it's mimicking our conscious Consciousness you would need to be that tool for a little while in order to evaluate what it is like to be it and if it turns out that it's Consciousness and it feels more or less the same as us then you could say it's mimicking our Consciousness but as it is
54:51 I would say we have no grounds to say even that it's mimicking our function and behavior and what role it will play in the evolution of humans the same as any tool and it can offer danger dangers as any tool nuclear power stations are dangerous we need safety mechanisms to to keep the risks uh uh uh constrained same thing for AI nuclear power stations have changed our lives same thing for AI tools tools change our lives and and change our Evolution AI is no exception thanks thank you we have a few minutes more for two people only uh the two downstairs are standing the long time um yeah thank you um so I don't intend this to be a debate but when we talk about
55:37 chat GPT this is a language that uses reinforcement learning which uses uh human feedback we also know that there are uh liquid neural networks that is also a type of language that can learn from its environment we have all the learning uh uh parts so I would like to clarify a bit because I know you touched the black and white both extremes but there is a very clear difference between Consciousness or I would call a synony awareness and senscience and I like a very nice example for instance um on the AI being used for surveillance for instance if I live in Iram and I decide that I like girls the amount of time I will spend looking to a girls picture is an
56:24 information that the algorithm can take take and it can make a decision if I the government should knock my door and talk to me or not so there are a lot of these gray areas it's not definitely not s uh s sence but it's definitely a awareness and I would like to touch a bit upon this uh Gray Line because it's very important that we talk about the ethics of AI especially when it makes this type of decisions what is awareness how is awareness different from sanctions so I understand awareness as something that I can make a rational decision and sence so for example my reptile brain can have a few uh uh judgments but science is mostly related to my feelings at least
57:10 that's how I understand this look if we redefine words we will be able to say AI is aware you just Define awareness awareness in a way that is not the way everybody understands it to be and and this is not un Comm people win debates by redefining terms all the time yes it's just silly it means nothing it's a language game it has no bearing on the actual facts but people do that when I say sence what I mean is what philosophers call phenomenal Consciousness it does not require any higher level mental functions does not require metacognition self-awareness the ability to introspect none of that so what I'm questioning is will there be something it is like to be a silicon AI
57:56 and my my answer is we have absolutely no reason to think that even though I can't refute it now regarding behavior and by the way you said um it's trained by reinforcement learning with human feedback yes that's why it looks like humans it's the process by by means of which we construct the shop window mannequins so yes it looks like human because it was trained to look like human and behave like a human and finally if resemblance means anything then you start you start having to ask yourself whether Conway's Game of Life Is Alive it's based on very two Simple Rules a cell in a grid will be alive or stay alive if two or three of its neighbors are alive if more are alive the cell dies or the cell doesn't isn't
58:43 born extremely simple you can run this on a calculator literally and it exhibits many of the characteristics of life it shoots arrows it reproduces creates societies we have to be very careful about making fundamental ontological judgments basing based on appearance complexity science shows us that a lot of complex complexity can arise from extremely simple systems that nobody would even ask whether they think is comous thank you thank you very much um this was the last question this was berno costr with AI won't be conscious thank you very much
