This man was over 90 years old giving this talk. Absolutely amazing. May we all have so much mind left at ninety.

Suddenly it seems I find a rash of Freeman Dyson youtube clips. Thanks to everyone who s uploading them.

What a great man. I'd love to shake his hand and tell him an honest "thank you, sir".

Prof Freeman Dyson | "Are Brains Analogue or Digital?" | 19th May 2014 - Dublin Institute for Advanced Studies, Statutory Public Lecture of the School of Theoretical Physics, in association with the UCD School of Physics. Abstract: We know that creatures like us have two separate systems for processing information, the genome and the brain. We know that the genome is digital, and we can accurately transcribe our genomes onto digital machines. We cannot transcribe our brains, and the processing of information in our brains is still a great mystery. I will be talking about real brains and real people, asking a question that will have practical consequences when we are able to answer it. I am not able to answer it now. All I can do is to examine the evidence and explain why I consider it probable that the answer will be that brains are analog.

One of the greatest physicists of the 20th century does a lecture at UCD and it gets 541 views on youtube... makes me sad man :(

Great lecture, I feel privileged to have the opportunity to hear this genius. Prof. Dyson, maybe be right about the brain being analoge!

Freeman Dyson is like coffee to me. Love this skinny old genius.

I just love listening to this fellow--what a mind! The depth and richness of the human experience suggest to me that we must evoke the ideas of quantum fields to to arrive at the vast probabilities and complexities that we witness in thought and emotion.

I've never seen him with so much energy he must have been having a really good day that day

Type B AI is now developing for entertainment such as chat bots and AI produced art, AI produced music and poetry, video games and fun photo filters. AI ML was used to develop a vaccine during the pandemic and to sequence DNA. In medical applications AI ML is used for comparing patient medical information to a database of information to help determine a possible diagnosis. IE X-rays of bones, CAT scans of the body and brains, ultrasound images, and ect. AI ML has been used in a wide range of applications and fields and is growing more popular as it becomes more available to the general public. At this point in time I can download several AI apps to my smart phone and use them right away.

Good discussion on digital vs analogue around AI among other things, but crap Dyson is a brilliant and it's just a great listen. Should have FAR more views.

Fascinating profound thought. Many may not realise that what they call a digital computer is actually a simulation built on an analogue substrate. Nobody has made a transistor that changes from off to on without spending some time in transition, where most power is dissipated. Typically a clock is imposed to ignore the confusion of these analogue transitions and provide a digital abstraction layer. A rough analogy would be the suppression of static between radio stations, the output being cut for weak signals, offering you perhaps 10 discrete stations. Convenient, but don't imagine that the continuous analogue domain between them doesn't exist.

Freeman Dyson, brilliant, as always.

I largely disagree with the objections to "digital AI", but this is great stuff nontheless. I of course would never argue with the Professor about quantum physics :) Listening in 2022, after all the recent significant advances in image recognition using NN, suggest that digital processes are able to achieve things that the professor didn't expect. Music by computers, i.e. Algorithmic Composition, is a favorite niche subject of mine, and has gone through incredible evolution in recent years. The one thing that stopped us so far has been computing power - at it is the availability of greater and greater computing power that has enabled the current leaps. In general, the notion of "the beauty of a rembrant belongs to the whole scene and not to the indevidual pixels", i.e. the gestalt approach, is exactly what NNs are great at - categorization. This repeating point over, that digital fails to capture the whole and analogue succeeds - does not seem to hold ground. A seperate fundamental issue is some aspects of the human brain are in fact digital-like. In particular, the spike behavior that neurons exhibit very much exhibit an on/off behavior. I am not saying the brain is digital. I am saying it contains elements of digital processing within it.

Nice tie. Great lecture. Deserves many more views than a meager 546.

"The Black Cloud" was a conglomeration of time displaced points of reflection and the concept of a mind-muscle motivated-moved by collective thoughts is a scale independent idea of mind-body? It's another version of temporal superposition in principle in which "one electron" or vanishing point occupies all time inflated by projection.

If your looking for the needle it sounds like an important new use for big data ... however most understand trends etc. that are currently being mined. This video in part is basically talking about noise filtering which is used in all manner of devices from radios to cameras....in some applications it's called DSP .... digital signal processing....digital computers also just happen to be very good at pattern matching in other ways.....a very useful function can be seen in identifications systems that match large groups of finger prints and faces very quickly using digital mapping technics

26:43 recog of pattern maps vis in space-time+sound/time

Excellent.

His comments on the LHC are spot on. He's amazing.

Does anyone have an Idea where to find the script for this talk? It would be interesting to read the chapter he skipped here.

That was the greatest gift someone I’ve never personally met yet I know so well, ever gave me. Bravo Sir, Bravo 🙃 who’s the say that the north pole is the top that’s kind of where I started.

Prof Dyson I agree up to the point where we say that primates who utilize "Songs with meaning" are not just using language. You need to explain the emergence of language better.

A philosopher in Science

His point about the LHC being limited seems more a hardware limitation- declaring the overwhelming majority of the data stream "uninteresting" is as much the problem as the bias towards analyzing only germane data concerning verification of prediction.

A theoretical physicist named Freeman. Yes.

The Pour-El-Richards theorem seems totally academic, since we do not actually seem to live in a classical universe with continuum fields. It might be great for analogue entities living in a universe without quantum mechanics, but then would they perhaps lack free will? So then who can tell me that the power of analogue machines is really so hot?

Makes me think modelling climate requires a quantum analogue machine - the digital approach ain't getting anywhere close - they are just using large spreadsheets.

Better question: does it matter which they are?

Memory. We remember events demarked by sense impressions and our understandings of them when experienced. Stimuli create tool marks of themselves and the body's responses. Those markers can be sequenced contemporaneously and when recalled, convey the memory together with our understandings. Which since as memory are objective to us, can be subjected to further consideration and new understandings. Is the saved sequence molecular or stored charges? That's how it works conceptually. And to say 'can be sequenced' suggests they are sequenced subliminally by the same intelligence and creative potential that we experience as fundamental to our core sense of self, personal identity with unconscious agency, our knowing and agency just a part of that same self.

Puzzled here. How can the brain be analogue if the retina and ear hairs perform transduction of analogue signals?

Dyson was amazing, and i think he was one of the greats. But I'm saddened to hear him confusing the concepts of accuracy and precision. In his talk, it's not a severe deficiency, but it is unexpected. I do wish Dyson--and Turing and von Neumann especially--were alive to see large-language models in action. I can only imagine his delight and contributions to the field now that we have fast-enough computers and algorithms to implement rudimentary AI....

42:06 analog quantum computing

39:00, analog vs digital, if a picture is worth a thousand words , how many digits are in a word?

I have always thought brains are sampled and thus digital. As we age the sample rate is slower and that is why time seems to pass more quickly - there are larger conscious gaps between samples.

We could not experience symmetry of instantiated general physical law (e.g. f= ma) except from the stasis, or continuity, of our core self, personal identity as Hume described it, the autobiographical or narrative self. The role in our biology for the unoscillating continuity of personal identity is largely subliminal, not of the brain, which acts as the demon in the machine from Maxwell, maximizing use of free energy. Coppola and Purves 1996 makes that optimizing function clear enough to say homeostasis predicates upon our core sense of self's absolute regularity during life along with its agency. And, the binding problem from that study is clarified: because visual processing is instant per that study, we have an acausal binding to our body.

This is a Brilliant Mind

4:03

RIP sir.

now i know why i prefer analogue sources in my hi-fi system

4.03

????saludos

14:13

Every thing in the universe(s) is digital, nothing is analogue. Analogue is only a concept.

Can i use analogue calculators to avoid big taxes :D ?

Star Trek Next Gen Season 6 episode 4 RELICS Scotty is trapped in a Dyson Sphere! HOOHAH!

world class

is it really possible that Neural networks and machine learnings in 2014 where so underdeveloped that even Dyson cannot see what was coming? And to think that all these new technologies are basically cutting-edge applications of statistical mathematics it is ironic how much unaware he was of this recent development

23:20 After 40 years of blood, sweat, tears ... and the profound lack of an AMD Radeon RX 6900 XT or Nvidia GeForce RTX 3090. Of course, back in 1980 such a device would have been impounded for _extremely_ secretive use in an Area 51 Data Center is two shakes of a lamb's tail. The NSA would barely have had sufficient technology to interface with this thing. Whereas Area 51 always had the best interstellar glue (including the glue they passed around every weekend).

37:30 analog computers more powerful than digital

Digital or analogue. It's seems to be for me ,a matter of interpretation ,rather than nature of it.

54k bro

too long didn't watch explain it to me please

The real question is, are brains analog or analogue?

I don't think the story about evolution of songs has the right explanatory power to explain evolution of language, because it leaves unexplained how apes or any other animal can perceive, store, generate, modify, select, combine (etc) songs. This is part of the general problem of how intelligent animals, including squirrels, crows, elephants, orangutans, and pre-verbal human toddlers can perceive, form intentions, work out how to achieve intentions, notice that something is going wrong, take remedial action, learn both from observation and from effects of their own actions, wonder whether ..., wonder why..., wonder what..., and many more. I have argued for many years (since about 1978) that all of these apparently non-linguistic capabilities require uses of rich *internal* languages, i.e. structures formed using mechanisms that allow construction of semantically interpretable internal structures with generative powers -- allowing construction and use of increasingly complex new structures -- and compositional semantics, i.e. allowing something complex and new to be interpreted in a way that depends on how the components are interpreted and how they are assembled, and also on the context ["Context sensitive compositional semantics"]. But these inner languages need not be restricted to linear structures like sentences (compare maps, diagrams, blueprints, trees, networks, etc.). There are many forms of animal intelligence and pre-verbal human intelligence that seem to be inexplicable if there are not such internal syntactic and semantic structures, and mechanisms of building and operating on them, including comparing them, storing them, finding ones that match something, and using them to control or guide behaviour or interpretations of sensory data. We know that such things are possible in principle because very simple (and as yet grossly inadequate) versions have been used in a wide variety of more or less impressive AI demonstrations, experiments and even useful applications -- but still with HUGE gaps compared with animal intelligence, or toddler intelligence. We know (as well as anything is known in science) that natural selection can produce and use forms of internal language with complex, varied, structures and a wide range of applications because something like that goes on in the mechanisms of reproduction/inheritance, and development of individual organisms. Without that, there could not be so much precise control, during development of a foetus and later, i.e. development and control of many exceedingly complicated working organisms. (Compare E. Schroedinger "What is life?" (1944)). So perhaps it's not wildly outrageous to wonder whether still unknown *internal* forms of language (with unknown structures and unknown mechanisms of construction and use) are also in use, even though neuroscientists know nothing about them, and perhaps don't even know how to look for them. I have an incomplete, tentative, discussion of how animals musing such internal languages for cooperative actions might gradually enrich and control the cooperation by extending the functions of actions to aid the cooperation (e.g. miming what is to be done before doing it, or, more subtly, using an incomplete action, with a change of gaze, as a request for information about what to do next). This might, by unknown evolutionary stages, lead to a complex and powerful *sign* language. We know that all humans (without serious physiological or brain impairment) are capable of learning, and can even create, very rich sign languages, even if most do not. (Search for "Nicaraguan deaf children" on KZfaq). Various easily imagined pressures (e.g. communicating in the dark, communicating while hands are busy with a complex task or holding infants, communicating round corners, etc. etc.) could later have led to additional brain and physiological mechanisms that re-used old non-speech linguistic competences in new spoken languages. Later on the use of written languages, another form of sign language, could develop, also based on use of previously acquired internal languages for encoding meanings. All of this raises many deep research questions. I hope more scientists and engineers will be prepared to investigate these possibilities instead of remaining "hide-bound" by a *definition* of "language" that is restricted to communication between persons. These ideas are spelled out in more detail in a slideshare.net presentation: www.slideshare.net/asloman/evolution-of-46383806 or PDF here: www.cs.bham.ac.uk/research/projects/cogaff/talks/#talk111 or goo.gl/piY2Lv/talk111 I suspect a theory of evolved rich *internal* languages could provide a sound platform for beginning a search for the explanations of musical and other forms of creativity discussed by professor Dyson. One of the challenges is to explain what additional mechanisms (e.g. meta-cognitive, or meta-meta-cognitive ... mechanisms) allowed our ancestors to make the discoveries leading up to Euclid's Elements, about 2,500 years ago, before the birth of modern logic, algebra, meta-mathematics, etc. Current AI systems do not (yet) have such capabilities, but that may because we have not found the right sorts of information processing architectures and forms of representation (despite all the innovations since the middle of last century). Pre-verbal toddlers seem to have fragments of the required competences, e.g. when they make and use topological discoveries, even if they cannot talk about them. It is possible that some of the quantum-mechanics based mechanisms considered by Professor Dyson will find applications in future AI systems with far more animal and human capabilities, e.g. in perception and action tasks that require simultaneous solution of multiple linked problems. I suspect that if Alan Turing had not died two years after publishing a paper on "The chemical basis of morphogenesis" he might have gone on to address these problems, as a natural extension of his earlier work: A Meta-Morphogenesis project. Aaron Sloman School of Computer Science University of Birmingham www.cs.bham.ac.uk/~axs

Dyson, brilliant in one venue, and not afraid to make bold pronouncement that are very wrong, like on Climate.

He become professor he didn't like PhD system

Professor Dyson, RIP.

One would think people at a university would know how to spell university.

Brains are organic

I think they're holographic....

Brains are formal boolean systems. In fact, are digital computers.

It appears that certain "large brains", instead of dealing with social problems, are more interested in creating social problems!

This talk consists of a lot of waffle and nonsense. A key part of the argument seems to be "My computer isn't analogue, and it doesn't think. Brains are analogue and do think. Therefore, I need to make an analogue computer." Compare "My plane doesn't have a beak and doesn't fly. Birds have beaks and fly. Therefore, I need to give my plane a beak." There are two courses of action available. If you make something as similar to a human brain as one human brain is to another, it will think. If you make something as similar to a sparrow as one sparrow is to another, it will fly. On the other hand, you can build a plane from a mathematical understanding of airflow. A plane that is different from any bird. You need either detailed, high precision copying, or a mathematical understanding of what you are doing. Reasoning by superficial analogy doesn't actually work in real life. The proof that "analogue" computers are more powerful than digital ones is an equivocation. Abstract mathematical "analogue" computers can store and process an infinite amount of information. If there is no noise whatsoever, you can encode an infinitely long string of information in the infinite digits of the position of some component, or the voltage in some wire. In the real world, there is always some noise. Precision is not infinite. The amount of information you can actually store in a continuous position or voltage is finite and fairly small. Then we get reasoning by science fiction. One work of fiction portrayed digital mind uploading in a way I don't like, another portrayed analogue mind uploading in a way I do like, therefore ... Any mind uploading, whether digital or analogue will have finite precision. There will be some error in copying memories and personality. Memories and personality change slightly with every moment. There is no reason why digital or analogue mind uploading must involve more than a miniscule negligible error. The decision of digital vs analogue mind uploading should be made based on detailed technical consideration of the reliability, efficiency ect of the technologies available, not on a warm fuzzy fealing you have around the word "analogue". Quantum computing is good at things like factoring large numbers in polynomial time, something humans aren't good at. Our current quantum technology usually requires cold temperatures, vacuum chambers, chrystals ect. The warm messy environment of the brain is more prone to decoherence. Quantum unpredictability is just randomness. Does a dice have "free will"? no. Its just random.

Are you kidding me? Dyson is a great guy but an hour of from Reading off of a paper is ridiculous! The whole point of a lectures to personalize a subject and make it interesting, not to watch somebody read words off of a paper for an hour. Whoever came up with this idea should be shot.

When Mr Dyson cannot get the term "Digital" correct, I can't continue watching..... People use the term "Addiction" without understanding the definition, it is just one of countless examples. Ignorant people just make it up as they go, I hope there are other civilizations elsewhere in the Universe, because just thinking that Humans are the most intelligent creatures of all time makes me sad.