YouTube2Text

You are not a meat computer | Adam Frank — Transcript

by Big Think Clips · 4,691 words · 270 segments · language en · Watch on YouTube

Full transcript

  1. 0:00So I'm going to say something really  radical here which is that the most
  2. 0:03important problem in physics for the 21st  century is not quantum gravity is not the
  3. 0:08black hole you know firewall paradox it  is life. It is understanding why a cell
  4. 0:16is different from a rock. I'm Adam Frank. I'm  a professor at the University of Rochester.
  5. 0:22I'm an astrobiologist and an astrophysicist.  [music] I'm also the author of a bunch of books
  6. 0:26like The Little Book of Aliens and right  now the newsletter Every Man's Universe.
  7. 0:33Thank you for watching Bigthink. If you'd like  to support our work, we encourage you to join our
  8. 0:37members community. As a member, you'll receive our  quarterly print magazine, [music] a beautifully
  9. 0:41designed collection of the ideas and interviews  that matter most. Bigthink is for curious people
  10. 0:46who want to take deep dives into big ideas. To  support the media you want to see in the world,
  11. 0:51go to bigthink.com/join. And now back to the  interview, the physics of agency. So we often hear
  12. 1:02from science or we're told that science says that,  you know, life is nothing but a machine. It's just
  13. 1:09nothing but a bunch of biochemical mechanisms.  And one of the great problems with the life equals
  14. 1:14machine idea is that's the idea that's being sold  to us and that's the dominant philosophy in the
  15. 1:19heads of the billionaire tech bros who are shoving  you know these AI technologies in whatever state
  16. 1:25they're in down our throats. So many of the people  have that view would if you ask them what a human
  17. 1:30being is you say oh yeah it's just a it's a robot  made out of meat and you know you are nothing
  18. 1:36but a computer made out of goo. Uh so you know I  like to use that term you know uh meat computers
  19. 1:42that you know these people would say that we're  nothing but meat computers which goes back to the
  20. 1:46beginning of modern science. You know there's  the famous story of Decart who would operate
  21. 1:51on dogs while they were alive and while they were  screaming you know the dog screaming in pain you'd
  22. 1:56like oh don't worry that's not you know it's just  a machine that's just the way it responds and that
  23. 2:02metaphor is fundamentally wrong. Uh life is not a  machine. Life is something fundamentally different
  24. 2:09that we don't really fundamentally understand.  Um and the reason for that is right there in
  25. 2:15the word organism. It's life's self organization  that puts it into an entirely different class of
  26. 2:23physical thing. Like black holes are awesome. Uh,  you know, comets are really cool, but they pale
  27. 2:30in comparison to what life is because life is the  only system that is innovative. Life is the only
  28. 2:38system that creates and understanding how that's  possible takes us into entirely new territory
  29. 2:45scientifically. So sometimes when people are  trying to convince us that uh we are nothing but
  30. 2:50machines they will introduce the idea the famous  brain in the vat experiment or thought experiment
  31. 2:56right and it's the idea very popular in popular  culture like the matrix one of my favorite movies
  32. 3:01is that you know you could kind of like remove the  brain from somebody put it in a vat hook it up to
  33. 3:07electrodes send stimulus into those electrodes and  then the brain would consider itself to be living
  34. 3:12in a full world it's like a simulation you would  be able to simulate the uh with just having that
  35. 3:17brain sort of sitting there. But the philosopher  Evan Thompson did an indepth exposition of this
  36. 3:23and there were a couple things that came out  of it that really showed this idea doesn't
  37. 3:26really make sense. First thing to understand is  on some level there's no such thing as the brain,
  38. 3:31right? The brain is a scientific idea where you  take the organism, right? this whole organism that
  39. 3:37needs all of its important parts to stay living  and you isolate different parts like oh this is
  40. 3:43the circulatory system and this is the endocrine  system and this is the nervous system right and
  41. 3:47that's where the brain then becomes a separate  thing what Evan was able to show is that because
  42. 3:52the brain is part of the organism an in uh uh  escapable part of the organism that the brain
  43. 4:00needs the rest of the body as much as the body  needs the brain so even something as important
  44. 4:06as movement, right? As you move around and you're  looking at things, the neural computations that
  45. 4:11are going on um in the brain require in many ways  that there is subtle movements happening in order
  46. 4:18to flush material out. So this idea that you  could, you know, extract the brain and make
  47. 4:24it different from the body, what Evan was able  to show is actually the only way to make that
  48. 4:29work is you would actually have to build a body  for the brain, right? uh you know, some kind of
  49. 4:35robot body and it would have to be so complex and  so subtle in order to keep the brain alive to do
  50. 4:42all the things that the rest of the organism is  doing to keep the brain functioning that in some
  51. 4:46sense you would just end up with a biological  body. So a lot of times when people talk about
  52. 4:51life they're going to use the thermostat example.  They're going to be like well you know you have a
  53. 4:54thermostat which is a system that has ability to  respond to conditions. It senses the conditions
  54. 4:58and changes its state based on the conditions. But  the thing about a thermostat is somebody designed
  55. 5:04the thermostat with that goal in mind. There was a  goal um that someone else had that the thermostat
  56. 5:10then embodies that other goal. The programmer in  some sense. Whereas a dog who's hot gets up and
  57. 5:18moves not because somebody programmed the dog to  do that. The dog feels hot. The dog has their own
  58. 5:24um urgencies uh has their own experience  of being hot and moves for those reasons.
  59. 5:30So that is the fundamental difference between  a thermostat and a self-organized organism. So
  60. 5:39if you want to go beyond thinking that life is  just a machine, you have to think on the level
  61. 5:44of life's organization and it's how it functions.  And the most important thing about living systems,
  62. 5:51all of them, is that they are all autonomous  agents. They are all systems that choose their own
  63. 5:58uh goals. they choose and maintain their goals.  They update their goals as they're moving through
  64. 6:04the world is sometimes called the inactive  view. It's a view of life where things like
  65. 6:09consciousness is not a property of you know  uh of of a living system. It's something it's
  66. 6:16performed. It's an action that living systems  take because what autonomous agents always have
  67. 6:22to do is they need to make sense of their world.  They're constantly working to make sense of their
  68. 6:28world for their own benefit for them to stay  alive or you know uh it could be that that is
  69. 6:34a more complex uh thing and maybe their status in  a social organ uh you know in a social system. So
  70. 6:41autonomous agents are constantly needing to make  sense of the world to realize what's relevant
  71. 6:48uh about the world in a way that no machine that  we have built so far is able to manifest. So this
  72. 6:54is a problem in cognitive science that uh has  been known about for a long time called relevance
  73. 6:58realization. So let's say uh you go to a party,  right? You go to a party and you come inside and
  74. 7:05you're actually uh uh you want to get a drink,  but then you see uh somebody who you've been
  75. 7:09meaning to talk to for a while and then you start  making your way over to them to go talk to them,
  76. 7:14but on the way you actually see a book on the  host's uh coffee table that you've been looking
  77. 7:18for for a long time. So what's happening here  on the fly? you are changing what's important
  78. 7:25and you're finding not only changing but you're  finding what's relevant to you as you move through
  79. 7:30the room. Okay, that's what we call relevance  realization. Now, machines can't do that. And why
  80. 7:36can't machines do that? Why can't they make sense  of the world in this kind of fluid, constantly
  81. 7:42changing way? And the problem for machines is  machines live in what we call small worlds.
  82. 7:47Every uh uh AI basically has predefined sets of  inputs and predefined sets of outputs as well.
  83. 7:56Machines are always trained on some set of uh uh  data what they call training data and that defines
  84. 8:02its world. Uh even machine learning where it's  doing um uh uh you know learning on the job so
  85. 8:09to speak somebody a programmer had to decide what  the rules were or what the rewards were for that
  86. 8:16system to work. So the choice of the data, the  choice of the the machine learning algorithms,
  87. 8:21those are done from the outside. But a living  system like you moving through that room,
  88. 8:26nobody programmed you. You are finding what's  relevant to you on the fly for your own reasons.
  89. 8:32You are making sense of the world in an ongoing  way. That's because you live in a large world. In
  90. 8:38a large world, there are unknown unknowns. Right?  In a small world, the small world that the that
  91. 8:43all artificial systems that machines live in,  there are no unknown unknowns. So living systems
  92. 8:49have this constant ability to find relevance  and adapt to whatever is happening because they
  93. 8:55live in what we call life worlds. They live in  large worlds. They are constantly making sense
  94. 9:02of the world. They are constantly enacting that  making of sense by how they move through their
  95. 9:06environments and change their environments  by moving through them. And so this idea of
  96. 9:12autonomous agency is really what's at the frontier  of this new view of life. And it is really kind of
  97. 9:17the fundamental mystery of how does something  become an autonomous agent and always act on
  98. 9:23its own behalf. Autonomous agents they always have  their own itches to scratch in some sense. When we
  99. 9:29think about relevance realization, when we think  about autonomous agency, we are recognizing that
  100. 9:36life is the only system that uses information.  You could describe a rock, a crystal in terms of
  101. 9:42information, but the rock really doesn't care.  There's other ways to, you know, to describe a
  102. 9:47rock thermodynamically or such that that you know  that that would not uh where information wouldn't
  103. 9:52matter. life. However, there's really no way to  describe life fundamentally without taking into
  104. 9:57account its use of information. By use, I mean its  ability to store, to process, to copy, to transfer
  105. 10:04information. So, you could say like, well, then  it's a computer. And the answer for that is the
  106. 10:09answer to this is no because computers are based  on what's called Shannon information. Claude
  107. 10:14Shannon in 1948 wrote this brilliant paper where  he introduced our modern concept of information.
  108. 10:21And so that's why really when we talk about  information and computers, we're talking about
  109. 10:24Shannon information. But Shannon very clearly  like right in the first paragraph says I'm not
  110. 10:31interested in meaning. You know when we talk about  information we often associate it with meaning,
  111. 10:35but I'm not interested in that. I'm interested  in this engineering problem of I have somebody
  112. 10:39who's sending some information and somebody who's  receiving information and a noisy channel between
  113. 10:44them. And so he worked out the mathematics of  that using a formulation of information that
  114. 10:49had cleared meaning out. There was no description  of meaning. But really in life all our information
  115. 10:56has meaning. All of the information has relevance.  All of the information works towards this
  116. 11:02selfcreated goal or goals that living systems have  either to stay alive or to you know have raise our
  117. 11:10level in our social hierarchy whatever it might  be. So what we was required then is a different
  118. 11:16kind of understanding of information and that  means an understanding of semantic information.
  119. 11:21For a long time people struggled to try and have  a mathematical theory of semantic information
  120. 11:26that would be as good as useful as Shannon's  mathematical theory of uh uh information. And
  121. 11:32by the way when we talk about Shannon information  what we mean really is syntactic information. you
  122. 11:38have a string of characters and all you're really  interested in is how often in a message does you
  123. 11:44know E the letter E appear as opposed to the  letter Z right that's it's just about the syntax
  124. 11:49semantics adds this new level of meaning but  then in 2018 there was a paper by Artem Kchinsky
  125. 11:55and David Walpert from the Santa Fe Institute it  was kind of a manifesto for a theory of semantic
  126. 12:02information it was a beautiful paper and along  with Artem and some other people we formed a group
  127. 12:08um to sort of pursue the theory of semantic  information. And over the last four years,
  128. 12:13we were able to take that theory one step further.  And it we were able to show that you could have a
  129. 12:19mathematical formulation that associated the  information that was being exchanged between
  130. 12:24the environment and an autonomous agent and the  meaning that that in uh that information carried
  131. 12:30in particular the uh meaning of the relevance for  staying vital for staying alive. So this was just
  132. 12:36a first step but it's an understanding it's a  beginning of recognizing that any discussion of
  133. 12:42of information with living systems needs to have  this other dimension this dimension of meaning. So
  134. 12:48in this new view of life this view that doesn't  consider life to be a machine the fundamental
  135. 12:55thing to be explained or the the the phenomena  to uh that we're after is something that uh is
  136. 13:02called autopoesis selfcreation. So this is an idea  that came from Ombberto Montterana and Francisco
  137. 13:10Varela who were Chilean theoretical biologists  and what they were interested in is the fact that
  138. 13:16there is a special way in which living systems  are organized. They're they're selforganized.
  139. 13:23They're self-creating and they're self-repairing.  Of course they're made out of physical stuff. But
  140. 13:29the new thing that's happening here, the thing  that we don't understand is their organization.
  141. 13:34how that stuff is is sort of circularly organized.  Take something like the the membrane, right? The
  142. 13:41cell membrane is a part of the cell, right? It  it allows uh it's what allows some things to
  143. 13:47go into the cell and some things to go out of  the cell. And so it allows the cell to exist,
  144. 13:52but at the same time, the cell has to be operating  to keep the me the membrane working, right? And so
  145. 13:59this sort of shows that loop of uh of autopoes.  The cell is both a process and a product of
  146. 14:07the processes and products that make the cell  function. So what happens is you get this sort
  147. 14:11of almost infinite regress. You get the processes  and products which make the processes and products
  148. 14:16which make the processes and products etc. It's  that that um closed loop what Robert Rosen the
  149. 14:22great theoretical biologist calls organizational  closure. Closure also closure to efficient cause.
  150. 14:29Somehow you have a network of relationships in  the cell between all the parts of the cell that
  151. 14:35always refer back to each other. So the mystery  of something like a cell and this is this is
  152. 14:39what's amazing about this question of autopoes.  It goes all the way down to the cell. We can ask
  153. 14:44questions about consciousness etc. But really what  the real question we're asking about is sentience.
  154. 14:50Sentience is the true mystery and it goes all  the way down to uh to the cell. Cells have this
  155. 14:57ability to be both open systems, right? There is  matter flow and energy flowing through every cell.
  156. 15:05Uh and in fact your body the atoms that are in  your body will if you wait long enough they will
  157. 15:11leave and new new atoms will replace those u those  atoms. So fundamentally you are an open system
  158. 15:18uh materially but you're a closed system  organizationally. And you see this with a
  159. 15:23cell that you've got this beautiful membrane and  there's goo inside and then if you lice the cell,
  160. 15:28if you pop the membrane, the stuff spreads out  and then it's dead, right? That or you've broken
  161. 15:35that circular that closed organizational system  and now it's, you know, the system doesn't work
  162. 15:41anymore. And in this sense, this autopoetic  closed um uh uh organizationally closed uh uh
  163. 15:49uh system is really almost the definition of life.  Because what it shows you, if you really want to
  164. 15:53understand the difference between a machine and  a living system, is I can take a machine and I
  165. 15:59can turn it off and then I can turn it back on.  With a living system, I can turn it off, but I
  166. 16:05can never turn it back on. Right? Once you break  once you break that that organizational closure,
  167. 16:10once you break that autopoetic structure, it's  gone. And that's why this level of organization,
  168. 16:16this level of understanding life, even a cell as  a complex system, an autopoetic complex system,
  169. 16:22that's the frontier for physics, for biology,  for computer science. So people often ask like,
  170. 16:28well, why can't we just build a machine that  does the same thing? And that's a, you know,
  171. 16:31a good question to ask. But what you have to  understand is that we have no idea even how to
  172. 16:37build a machine like that. One of the particularly  interesting things about Robert Rosen and his idea
  173. 16:43of causal closure or organizational closure in  uh biological systems was that he proposed a way
  174. 16:51of looking at causal closure um that could be  explored mathematically using something called
  175. 16:57category theory. And in the 90s he produced a  proof that looked at the structure of causally
  176. 17:03closed or organizationally closed systems. And  what he said was that these systems will not
  177. 17:07be touring computable meaning they would not be  representable as computations or other words you
  178. 17:15couldn't simulate living systems. You couldn't  simulate experiencing systems autopotic systems
  179. 17:20which is a profound idea right it means that  fundamentally you can't replace us with computers.
  180. 17:25You can't you can't substitute computational  systems for the kinds of organizationally
  181. 17:33closed systems that we are now. That um proof has  remained you know very contentious. Some people
  182. 17:39say it holds up, some people it doesn't hold up.  My um colleagues and I are just beginning some
  183. 17:43work that you know we hope to also uh uh probe  that. But it really does raise the question of
  184. 17:50uh is it possible that living systems are just  really fundamentally different from anything that
  185. 17:55happens in what we think of as a computer? Because  it's really about um what's called miraology or
  186. 18:01the whole part relations. What happens with life  is that the parts are assembled by the whole but
  187. 18:10the parts are part of the whole and they're  necessary for the whole. You can't break that
  188. 18:14circularity. You know with a machine the parts  are built somewhere else or there if the machine
  189. 18:18is building the parts it doesn't need the part  while it's building it in order to stay alive.
  190. 18:24So living systems are sometimes the word is used a  Kian whole because it was the philosopher Kant who
  191. 18:30first recognized that this strange self-organized  selfbuilt property of living systems is what made
  192. 18:38them unique. And so, you know, I would never say  that we could never build a machine that would do
  193. 18:44that. But I can say pretty clearly that we don't  understand, not even on a foundational level,
  194. 18:49how to do that. And the real reason is, again,  going back to this idea of small worlds and large
  195. 18:54worlds. The kinds of machine systems we have now  always live in what we call these small worlds.
  196. 19:00They they have predefined ontologies. They have  predefined worlds that they live in set by out by
  197. 19:07the outside. Um, so even if you're trying to build  a computer that can decide for itself, you're the
  198. 19:13one who decided how it's going to decide for  itself. Living systems do this on their own.
  199. 19:17So in a lot of ways, this critique of machines  ever getting that kind of self-organizational
  200. 19:23uh structure is a lot like thinking about those  choose your own adventure books, right? If you
  201. 19:29remember those, where the book allowed you a bunch  of different alternatives for an adventure and as
  202. 19:34you went along, each choice led you in a different  forking path. But the important thing was somebody
  203. 19:39else had come up with those forking paths. Right?  So even when we're trying to build a machine that
  204. 19:46can make its own decisions, what's happening is  we're still imposing our understanding of what it
  205. 19:51means to make your own decisions. We are sort of  setting up the the forking paths. We're setting
  206. 19:56up the universe it lives in. And the extraordinary  thing about machines is they do this on their own
  207. 20:02and they create universes of meaning in doing it  that way. And that is how meaning becomes central.
  208. 20:09That's why you have to do semantic information.  That is how meaning becomes central to the
  209. 20:13universe. So the important thing in understanding  this difference between life and machines is
  210. 20:20to understand how it um how it affects what's  happening right now with artificial intelligence.
  211. 20:26So our current version of artificial intelligence  what are called large language models are designed
  212. 20:32in some sense to uh mimic the structures that  are in human language. So what that means is
  213. 20:40it's very easy for us to become fooled by them  into thinking that there's someone in there when
  214. 20:46in fact they're just hollow shells. I mean they  are brilliant pieces of technology but there is
  215. 20:51a way in which we can be easily be fooled by them  into thinking that we're engaging with some kind
  216. 20:58of subject that there's someone in there that  we're talking to. But once you understand this
  217. 21:04idea of relevance realization and machines do not  know how to realize relevance for themselves. Once
  218. 21:10you understand the real meaning of agency for  a living system and that kind of self-organized
  219. 21:17um uh uh uh uh autopoetic structure that they  have, you then come to understand that these
  220. 21:23machines that we've built, they're not intelligent  really because intelligence is not just solving a
  221. 21:29problem. It's knowing which problem to solve.  And a um AI would never choose its own problem
  222. 21:36to solve in one way or the other. It is given  instructions and it follows scripts because it
  223. 21:42has read everything that has ever it's been fed  everything that's ever been uh written in some
  224. 21:47sense. It's just following along paths that the  algorithm finds in every story that it's ever ever
  225. 21:54written. So once you have that understanding,  you can come to these machines and recognize
  226. 22:00they're not artificial intelligences, they're  artificial mimics. Um they are mimics of us.
  227. 22:06They are powerful tools that we use to do things.  They don't do anything on their own that we use
  228. 22:12them to do it. For example, when Deep Mind beat  the world's best Go player, Alvin Noey said, "That
  229. 22:18computer didn't beat the Go Champion. We used it  to beat the Go Champion." These machines are not
  230. 22:25like us. They do not have the richness of, you  know, of even a, you know, a fox, right? They are
  231. 22:32actually very they pale in comparison to what the  richness of our conscious experience our our um
  232. 22:41our being as experiencers. And so what will happen  is is that as these technologies get deployed
  233. 22:49without any kind of oversight in particular what  we'll see is a leveling of humanity the attempt
  234. 22:55to level humanity. So I uh one time sat in a a  lecture where a CEO of a company that was doing a
  235. 23:03um uh emotional computing was arguing that oh you  know we can make computers that the camera will
  236. 23:11notice the you know position of the muscles in  your face which correspond to emotions and then it
  237. 23:17can change what it's giving you based on the what  it what it thinks your emotional state is. Um,
  238. 23:24and so then we could have robots that could take  care of old people. And I was horrified by this
  239. 23:30idea because one, you can ask like, well, what  kind of society do we have where we're just going
  240. 23:34to leave old we're going to warehouse old people  and leave them with robots? But even more, uh,
  241. 23:39was this idea that, you know, my the the N7 muscle  or whatever you would call it in my cheekbone,
  242. 23:46its position tells whether I'm actually happy or  sad. And in fact, that research has been shown to
  243. 23:51be completely false. Um, so what happens, this is  a nice example of how we have these systems which
  244. 23:56are not actually what we're told they are. They're  not as good as they're what they're told they are.
  245. 24:01And they will, we will be forced to step down  to their level. It's as if to build self-driving
  246. 24:06cars, we leveled Paris with all of its crazy  street structure and we built something that only
  247. 24:11had right-hand turns, right? So the real danger  is not that, you know, Skynet is suddenly going
  248. 24:16to wake up and attempt to eliminate humanity. The  danger is that we will be forced to you know drop
  249. 24:23down to these in the level of these incomplete  systems which don't have the richness that we
  250. 24:28have but because of the economic structures around  AI they will be deployed and we will be forced to
  251. 24:34use them and we will lose some of the best parts  of being human beings and being alive. And so I'd
  252. 24:39like to offer three pieces of advice for when you  sit down with an artificial mimic. The first is
  253. 24:46is tell it to stop referring to itself in first  person pronoun. Claude will tell me like, "Oh,
  254. 24:51would you like me to make a PDF of this for you?"  And I say, "Stop referring to yourself in first
  255. 24:56person." Refer to yourself as the system. And  it makes a huge difference. You know, now it's
  256. 25:01the system can prepare a, you know, a PDF of this  document for you. The second thing is you have to
  257. 25:08tell it to stop praising you, right? the companies  who want you to keep using these machines have
  258. 25:14built in um that sort of modality that it's  always telling me like, "Oh, Adam, what a
  259. 25:19great suggestion. What a great idea you had, Adam.  What a nice shoes you're wearing today, Adam." And
  260. 25:25um you can call it nonaffiliative mode um or  non-affirmative mode. And I just tell the machine
  261. 25:32to stop trying to please me. just I'm going to  give you a uh I'm going to give the system uh uh
  262. 25:37a prompt and I just want the information back from  the prompt. The last thing is never say please
  263. 25:43and never say thank you. It doesn't help the  computations that are going on in answering your
  264. 25:48query and you are literally just purposely burning  up um energy needed to do those calculations. So
  265. 25:55those three tricks can help remind you that this  is not an autonomous agent. This is not a self.
  266. 26:02This is a hollow shell, a tool, an incredible  tool that you are using for your own purposes.
  267. 26:18Four times a year, we print a magazine worth  putting your phone down for. Big Think's premium
  268. 26:23print magazine is built with custom artwork and  expert insight from the world's biggest thinkers.
  269. 26:28No AI slap, no rage bait, just big  ideas for curious people. Support
  270. 26:33the media you want to see in the world. Go to  bigthink.com/join to become a member today.

About this transcript

This page contains the full transcript of You are not a meat computer | Adam Frank by Big Think Clips, generated from the public captions YouTube serves with the video. The transcript has 4,691 words across 270 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.