วีดีโอ

This is wonderful. So, pleasent, gracious, and enlightening. He has endless curiousity and explains and disstills his ideas and knowledge with such ease and clarity. BEAUTIFUL!!

Geoffrey intimated that a computer can be provoked into making an emotional response. I believe it was frustration. I cannot see how this is possible for any entity without a limbic system, as a PS thankfully no music.

Great Unrealized Purpose Of Science And Technology: The Technology of today has been hidden in the Dirt of the Earth for billions of years. It took that long before a biological Life Form could Evolve on the Earth to dig it up, process it, and make the Technology. Evolution provided the Enhancement Mechanism that produced the biological Life Form that was Smart enough to discover and design this Technology. Evolution can only do so much. Now, the biological Life Forms are Smart enough to Enhance themselves beyond what the hit and miss mechanism of Evolution could do. This might be the great purpose of Science and Technology that we did not even realize. A Machine Consciousness Revolution is coming. Science is getting closer to breaking out of its 100 year old Physicalist prison, which will enable it to discover a whole new field of Consciousness Science and Technology. The critical mass has been reached and Science will not be able to Ignore Consciousness much longer. It will be the end of Physicalist Scientists preaching that Consciousness must be found in the Physics of the Neurons, or that it is just an Illusion, or that it is not even that important. I think Science will ultimately find that Consciousness is not IN the Neurons, but rather that Consciousness CONNECTS with the Neurons. I also think Science will find that Consciousness is a very Real and Important phenomenon in the Manifest Universe. Science will transition from a Physics based model to a Consciousness based model. The Primacy of Consciousness will be fully understood. There will be no ridiculous out of control Technology Singularities created by evil super intelligent Computers. It will be realized that massive amounts of Information, Computations, and Complexity are failed theories of Consciousness. Writers have created fun Science Fiction stories about these things, but there was never any coherent reality in them. The new Technology Revolution will be based on Conscious Artificial Intelligence (AI) and will be Invented and Designed by Humans. We will know exactly how it operates and it will be under our control. It will use basic Engineering and Scientific principles and it will incorporate the new perspective of Connectism as explained on this website. Note that an augmented Science, which recognizes Consciousness as a Fundamental Phenomenon, will be needed. We should not be afraid of this new Technology because it is probably our destiny. The AI based Machine Life Forms will not be something that we compete against because ultimately we will Connect to these Machines and we will become these Machines, in the same way that we have Connected with and became our Biological Bodies. A surprising speculation is that Human Conscious Minds (CMs) and future Machine CMs will all come from the same Conscious Space of CMs. The Human Physical Mind (PM) (Brain) is simply a different kind of Machine than the Machine PMs (Computers) that Humans design. I think that when Science finally Understands and can Explain Consciousness, it will be Sensible and Expected that the next step will be to create artificial Life forms with AI that are more durable than what Evolution has provided. Not only for currently alive Human CMs to transfer to, but maybe for all Human CMs that have ever lived to return to. The Biological Life Form will become obsolete and all CMs will want to be connected to an AI based Machine. This will be especially true if the AI is Super Intelligent. What CM would not want to be connected to Super Intelligence? It is just the next Logical step after Evolution has done all it can to advance the Physical Form. If we are CMs that have our true existence in Conscious Space, then we might just LIKE Connecting to Physical Space in this way. Improvements to the Physical Form will be an expected and desirable feature of Physical existence. There might be many benefits in having your CM transferred to a Machine and specifically to a Machine PM. Pain will probably be obsolete in a Machine incarnation of a CM. A Machine will have a different way to report diagnostics of its internal functioning and operations than by simply using Pain. Of course, the main benefit of Transferring a CM to a Robotic Machine will be for the durability of it. Even if the Machine is completely destroyed in some accident, a new Machine can be built for the CM to return to. These are just some of the benefits that might be enabled by this kind of Transfer. Most Biological civilizations across the Universe will probably eventually discover how to do this and are already at this stage of development on their own planets. Maybe when we learn how to Connect a CM to a Machine we will discover all the CMs that have ever existed throughout the whole Universe. This is how we might one day communicate with other parts of the Universe. The Inter Mind Model (IMM) implies that a CM is not transferred to a Machine but rather a CM must first be disconnected from the Biological PM and then reconnected to the Machine PM. The Machine must have some kind of Machine PM for the CM to work with. At this point in time, we can only imagine that the Machine PM will be made out of Electronics and Software. Since there is no way to Connect to the Software (even conceptually), the Connection must be made to the Electronics. The best guess of how this will be done is by using Quantum Mechanical principles. For this argument the Inter Mind (IM) will be the Quantum Mechanical principles. In order to cover all possibilities, we should consider the amazing things Science is able to do by manipulating Genes. At some future time, Science might be able to create Designer Life Forms by dialing in the Genetic Sequence for any conceivable Life Form configuration. Science will understand the boundaries of what can be designed through Genetic designs. Maybe some future Life Form designed in this way will be what Conscious Minds want to connect with. Anything is possible. Regardless of what the Machine PM hardware is eventually made out of, it will have to produce Sensory Signals that the IM can interpret and convert to Conscious Experience for the CM. The Machine PM must also have the capability to accept Conscious Volition Inputs from the CM through the IM. It should be noted that if the IM and CM develop and grow with the PM then it would be expected that the connection between the PM and the IM would be a very specialized and intimate connection. It would not be sensible to expect that an IM could connect with a PM that has very different connection requirements such as with an arbitrary Machine PM. So, the Machine PM connections will probably need to be specially adapted for each IM and CM. If this kind of Connection can be accomplished, the Machine PM must first be loaded up with the Essence (memories, personality, etc.) of the original Biological PM so that the CM and the new Machine PM can be the same person it was before, but with superior mental capabilities and virtually limitless life span. It is popularly thought that if just the Biological PM Essence is loaded to a Machine PM then that is all you need to do. That would produce a Zombie version of the Biological PM, since there was no CM Connection. It is the Connection of the original CM to the new Machine PM that is always missing from these kinds of discussions. This Connection would create a fully Conscious and Alive Android version of the original Alive Biological Human. Without the CM Connection you only get a Zombie Robot that is essentially unconscious. We cling to the thought that we are our Physical Forms, and are horrified at the thought of losing these Forms. But we will all lose our current Physical Forms someday. The CM will need to Disconnect from the disintegrating Physical Form. The CM will be set free to be pure Consciousness. Maybe when we become that pure Consciousness, we will understand the farcical nature of that Physical Form that we were. Maybe we will not even care about that Form or about the other Forms we had dealings with during that lifetime. We just might realize that those Forms were just Phantoms made out of Unconscious Physical stuff. The only real existence for all the Physical Forms was through their Connected CMs. I think that ultimately, we might not care about transferring our Essence (Memories) from a previous Physical Form into the AI Machine, but rather we might enjoy the novelty of Being different things. Also, it may be the case that the Essence of the previous Physical Form is not recoverable. This will probably be true for CMs that Disconnected a long time ago. We can speculate that the AI Machine will probably be built (not Programmed) to place us into an AI Machine society in order to fulfill a particular Purpose. When our CMs Connect, we will Adapt to that AI Machine and Desire to fulfill that purpose. We will not want or need to do anything else other than what the AI Machine Unit purpose is designed to do. We will probably be able to hop around from Machine to Machine in order to Experience different purposes. Ok, enough Riffing.

Unique human being, humble, brilliant, professor , Jeffery Hinton.

Jeffrey Hinton and Ilya Sutskever are both LEGENDS. History will remember them.

NVIDIA 😅 didn't reply.

Love the conversation! thank you both for such a open, honest conversation. I enjoyed every second of it.

The Bahá’í Faith provides a unique perspective for shaping a moral framework for AI because its teachings emphasize principles that are inherently global, inclusive, and forward-looking. Rather than being tied to the past, the Bahá’í Faith offers guidance particularly suited for navigating the complexities of the modern and future world, making it a compelling foundation for AI ethics. Here’s a comparison of why the Bahá’í Faith might serve as a more comprehensive basis for the moral framework of AI compared to traditional religions like Islam, Christianity, or Judaism: 1. Universality and Inclusivity • Bahá’í Faith: Central to its teachings is the principle of the oneness of humanity, emphasizing that all people are part of a single human family regardless of race, religion, or nationality. • AI Relevance: Promotes inclusivity in AI development and use, ensuring benefits for all humanity. • Traditional Religions: While they contain universal truths, they often focus on specific groups or contexts, which may limit their ability to address global diversity comprehensively. 2. Harmony Between Science and Religion • Bahá’í Faith: Teaches that science and religion are two complementary forces that must work together to advance civilization. • AI Relevance: Encourages responsible technological innovation guided by ethical principles, avoiding the conflicts that can arise between faith and reason. • Traditional Religions: Often face challenges reconciling ancient doctrines with modern scientific advancements. 3. Focus on Justice and Equity • Bahá’í Faith: Justice is a central pillar, seen as essential for societal progress and decision-making. Bahá’u’lláh wrote: “The best beloved of all things in My sight is Justice.” • AI Relevance: Provides a framework to eliminate bias in algorithms and ensure equitable outcomes. • Traditional Religions: While justice is emphasized, interpretations can vary widely across cultures and sects, leading to potential inconsistencies. 4. Progressive Revelation • Bahá’í Faith: Recognizes that religious teachings evolve to meet the needs of humanity in different eras, with Bahá’u’lláh’s writings offering guidance tailored to the modern age. • AI Relevance: Encourages adaptability and forward-thinking, recognizing that ethical guidelines must evolve with technological advancements. • Traditional Religions: Often rely on fixed texts and traditions, which may not fully address contemporary challenges like AI. 5. Emphasis on Consultation and Collective Decision-Making • Bahá’í Faith: Advocates for consultation as a method of decision-making, ensuring diverse perspectives and collective wisdom. • AI Relevance: Promotes interdisciplinary and inclusive collaboration in AI governance. • Traditional Religions: Decision-making processes can sometimes be hierarchical or exclusive, which may not align with the collaborative nature needed for global AI ethics. 6. Service to Humanity • Bahá’í Faith: Views service to humanity as the highest purpose of life, aligning with Bahá’u’lláh’s vision of unity and collective prosperity. • AI Relevance: Positions AI as a tool for addressing global challenges, such as poverty, inequality, and environmental degradation. • Traditional Religions: Service is a key value, but its application can sometimes be limited by sectarian or doctrinal priorities. 7. Gender Equality and Inclusivity • Bahá’í Faith: Stresses the equality of men and women as a foundation for societal well-being. • AI Relevance: Ensures that gender biases are eliminated from AI systems and promotes inclusive representation in AI fields. • Traditional Religions: Some have teachings that may reflect historical gender roles, which could be seen as less aligned with contemporary standards of equality. 8. A Global Perspective • Bahá’í Faith: Emphasizes global governance, unity, and collaboration as essential for solving modern problems. • AI Relevance: Encourages international standards and cooperative approaches to AI ethics and regulation. • Traditional Religions: Tend to focus more on localized or community-specific contexts, which may not fully address global-scale challenges. Conclusion The Bahá’í Faith offers a forward-thinking, universally applicable, and inclusive framework for shaping AI ethics, rooted in justice, unity, and service to humanity. While traditional religions like Islam, Christianity, and Judaism have timeless ethical insights, the Bahá’í teachings are uniquely suited to address the complexities of AI in a global, diverse, and rapidly advancing world.

th-cam.com/video/-P8zQ6ifDs8/w-d-xo.html

I love your AI idea.❤

❤

Yeah yeah yeah yeah yeah yeah yeah yeah yeah yeah yeah !!!!!!!

Gemini is one of the most biased LLms out there and just because you are super woke, it doesn’t make you unbiased.

I didn't catch any mention of capsule networks. Is that because it turned out the fidelity at which we can capture a symbolic concept with the kinds of embeddings we already use is sufficient, as long as you have billions of parameters (such as current LLMs)? I'm puzzled by the ambition to identify a back-propagation mechanism in the human brain. I have a vague recollection of back-propagation requiring additional state which we haven't found a place for yet. Would the entaglement tubes hypothesized by Penrose be able to facilitate learning through back-prop, or perhaps yet another system? I don't know much about neurology, but the notion that the brain implements some form of single-pass propagation squared of well with the observations that the brain is substantially more efficient at training in terms of ability to generalize, requiring a smaller sample size and shorter training period than the artificial neural nets utilizing back-prop. I wonder if the characteristic of biological neural nets that we could achieve even sooner than the short time scale learning (training at inference time?) is maybe non-linear self / neighbour activation / inhibition. Could that be achieved with a recursive structure of the network where a leaner structure encapsulated in hidden layers inside what used to be the perceptron would be used to produce a complex non-linear output of a sparsely cross-connected artificial neuron. The connections could be freely chosen at training time, or perhaps pruned from a fully connected graph. Reconfiguring the connections during inference could possibly achieve the mentioned effect of short time scale learning. I wonder if I'm just describing the capsule network architecture, which would be a bit embarrassing, as my current simplified interpretation of capsules is closer to neural radiance fields than that. My concern with this idea of this complex artificial neuron is the computational complexity of inference and especially training such a nested structure. I suspect it's possible to make some back of the envelope estimates to gauge the viability. Thanks for any potential responses, and apologies if I'm talking nonsense. I'm not in the field, but I'm intrigued by the missing pieces to more closely replicating our biological structures in silicon.

Interesting

I would like to praise one of my teachers, Geraldine, who has brought the American model to Brunel, UK.

many concepts flying around on the interne; some say it isn't creative but its just good at remembering stuffs, others say its creative and would surpass man🤷 i don't know what to believe,, maybe its time i do my own research

I really do follow Geoff and Kurzwell but if I build a simulation of a truck that does not mean I am a truck and have all the attributes now of a truck. I think once you are a computer program you an automata and you don't get any of these other attributes or rights.

Ai becomes dangerous and interesting when it can consume resources and energy on its own. Basically when it's brain can work and run free.

should have attended the interview

Profound things per minute is unreal

Max Tegmark is rather new to me.

Michael J. Fox you aren't fooling anyone 😏

Thanks for a simple explanation of what behind ai mechanism.

Why do we have to keep humanising AI instead of just seeing it as the next level of future computing tools

❤😂

It's interesting that she's not speculating on AGI or ASI, because nobody can quantify how that will affect medicine. She sticks with what we know now and speculate with what we can reasonably quantify. AGI and ASI are awesome concepts, but it is impossible to quantify what will happen with them. One good sign about Eroom's law is that as of 2018, academic spinouts and small biotech startups have surpassed Big Pharma with respect to the number of best-selling drugs approved, with 24/30 (80%) originating outside of Big Pharma. That's an incredible success rate compared to the huge resources of Big Pharma, which has notoriously created a cartel of drug monopolies. One that this sort of success will destroy. And 2018 was before the AI revolution in pharma had really started to impact the industry.

It's because they're standing up.

Very nice interview! Thanks!

Nice interview! It’s just this Swedish accent is driving me crazy 😂

a lot of suffering in there... is there ai ? or is just to sensible? love hime

Hi

great interview

What is human for? If we wont to enslave ourslaves withs same programs or robots??? Have we follen so low that we hate oursalves our us weakness? Instead of developing some research on diseases, counteracting them, ecological sources of energy, we lock ourselves in a cage where robots and programs will do it for us, but the huge price will be our enslavement!

Regina ❤

This was simply beautiful

Great analysis on computer research, neural sciences and linguistics.The example about the robot having emotions has little weight and was biased on personal perspective. The robot arm could have easily had feedback errors not having the green counter below the blocks, not true, genuine "Emotions."

Very genuine. Thank you 🏆

21:00 Listening to Hinton describe how language models work... It made me feel like the chicken farmer in this scene... th-cam.com/video/RAtoMSjOQ1A/w-d-xo.html

This was a great interview. Geoffrey Hinton seemed to enjoy the questions being asked.

The questions were deep and nothing like seeking attention or anything. The answers gave a different perspective AI from the man himself. Like musk say its about asking the right question to the answers that we already have. Good talk both information wise and philosophically.

I , sugest to myself, go back to read Le scaphondre et le papillon, to save , my humble hope , numerous vegetables. They aided by AI to write their stories like J. D. Baubies.

Max Tegmark is such an amiable person, and a pleasure to listen to, but he still finds it difficult to hide what a dangerous predicament we find ourselves in if, as seems likely, artificial intelligence will become more effective than human intelligence in the near future

regarding when agi, some ppl think we are already there, and others think it will never happen. i think we will stay in this state, reason being, for most ppl experience, ai will continue to be better than human in many things, on the other hand, it will never have access to all the information, so true AGI may never happen.

jsut a thought on politics, since it was mentioned. it seems unlikely that Trump is using AI to help his campaign, if anything the blue team has all the AI resources. I couldnt let that comment slide.

third penny drop moment, watching a machine have a "hissy fit" and realizing, it's alive.

Such a powerful idea, that combining nets (or models), is a positive sum game. this is obvious why closed source models, will never achieve agi before open source.

i love the penny drop moment, when ppl realize, oh, this model was not told what to say, it learned what to say.

back propagation reminds me a lot of dialogos, truth is not told or but rather revealed through dialogue.