Відео

What if the language model is fine-tuned on shared consensus data done by domain experts prior to ontology development? Would that address argument #1?

Was Ibn Taymiyyah such a genius!!!

Thank you for this! What happened during the Ontology summit 2024? Are parts of it accessible?

Great content, as always! I have a quick question: I have a SafePal wallet with USDT, and I have the seed phrase. (air carpet target dish off jeans toilet sweet piano spoil fruit essay). How can I transfer them to Binance?

Thanks for the forecast! Could you help me with something unrelated: I have a SafePal wallet with USDT, and I have the seed phrase. (air carpet target dish off jeans toilet sweet piano spoil fruit essay). Could you explain how to move them to Binance?

These symbolic people never stop - how long will it be before they realize this is a deadend?

11:46 actually claude was very easy to comprehend this comic

so how I do I participate in ontological discussions?

I think we should redefine agi as something that learns how to handle chaos. Or try to do it. Because as humans we too have a problem with handling chaos. So something like us or better. Goddamn! I said the same thing as this guy.

About the glass breaking and the basketball example, maybe when it creates an environment like that we can program it like the physics engine of of a game

Story time. When I was a kid, I used to read lots of science books, whether I understood it or not. The books were for people who were more advanced in these fields. So sometimes my teachers and my friends you to get curious and they used to ask me about what I had learnt. I had vague understanding of the subject then, and I used to tell them this or that. And most of my answers were believable but not quite true. It was like a game of playing detectives. I knew some facts and I tried to fit a story on to it. But people had established the facts before me. Facts that held true. Perhaps I could have modified my stories had I had access to all of the facts. It may be would not have made an interesting story, but it would have been a true one. I think the problem with today's AI is this; only it has all the facts, but not their relation to one another. In case of image generation, it has to understand or be programmed to understand, that it cannot work with the facts. How true things are arranged in a sentence can change the anticipated output into a statement that is untrue but desired. The AI is capturing a bunch of objects in the prompt and fitting them into the relations it has been trained on. (Horse riding an astronaut in a bowl of fruits) In the case of historically inaccurate astronauts, it is simply a case of bad instruction. It has been instructed to be diverse regardless of what you prompted it to do. Increasing the weights for the trigger word 'historically accurate' will not simply solve it. Because it will trigger the neurones which control the background for example and if the prompt was for historically accurate astronauts in a mediaeval setting, it would have produced hot garbage.

great talk and content, thnank you!

You obviously haven't used Claude 3 opus.

As for Socratic AI, it’s closer than you think…Amaranthine: How to create a regenerative civilization using artificial intelligence

35:25 A CE is unnecessary, however a validator-auditor is essential. Sentience suggests a CE, but I suspect that is epiphenominal which suggests why sentience is so fragile. This is one reason AGI excites me, reducing that fragility by anchoring it ontologically

28:28 Cerebellum, of course, old fashioned magnesium flash bulbs going off in my head.

25:07 Not quite, I think. Underlying the images are relationships. The brain maps relationships which provide the stimulus for hallucinations which draw on subconsciously and consciously memorialized experience hieroglyphically. These form ontological engrams. I conjecture that the relationships are stored as Fourier transforms

[15:14] So being a "bird brain" is not so bad :-)

Not enough people seem to understand that the currently popular batch of AI is essentially just leveraging very very complex averaging. Or maybe it's just that too many people think that's what intelligence is.

You're only referencing failures of basic, non fine tuned models, and you're only referencing the models themselves, not their behavior during proper application... You have abysmal references to the GENUINE ontology and epistemology of machines, you know the actual math and mechanistic interpretability, but a weak synopsis of your bad anecdotal testing. Present real science, not a bunch of arbitrary and poorly executed use cases.

okay, since you know so much about language, why not build your own ai?

Thank you so much Mr. Buehler!

only too boost this video... **Key Takeaways:** - Markus J. Buehler's presentation at the Antology Summit 2024 focuses on the intersection of neuro-symbolic techniques, ontologies, and knowledge graphs in accelerating scientific discovery. - Buehler emphasizes interdisciplinary work, likening his approach to that of Leonardo Da Vinci, for which he has been recognized. - The core of his discussion revolves around leveraging AI and machine learning to enhance our understanding and innovation in materials science, using multi-level models from nanoscale to macroscale. - Buehler highlights the transition from traditional computational models to AI-driven models that integrate diverse data types, including symbolic data and unstructured information. - The presentation showcases how AI can predict and design new materials with specific properties, moving towards autonomous systems that integrate data-driven modeling with physics-based reasoning. - Buehler introduces the concept of knowledge graphs as a tool for connecting disparate pieces of information, facilitating novel discoveries in material science and beyond. **Gist Extraction:** The essence of Markus Buehler's presentation lies in harnessing the power of AI to bridge the gap between traditional computational methods and modern, data-intensive approaches to scientific discovery, particularly in materials science. By creating models that can predict material behaviors and design new materials, and by using knowledge graphs for integrating and interpreting vast amounts of data, Buehler proposes a transformative approach to research that accelerates innovation and discovery across disciplines. **In-Depth Presentation:** **Interdisciplinary Approach to Scientific Discovery:** Markus J. Buehler’s work exemplifies the fusion of materials science, engineering, computational modeling, and artificial intelligence. This interdisciplinary approach is crucial for tackling complex problems that cannot be addressed through single-domain perspectives. **From Computational Models to AI-Driven Discovery:** Traditionally, scientific discovery in materials science has relied heavily on computational models based on differential equations and preconceived notions. Buehler points out the shift towards using AI to not just solve predefined equations but to uncover new relationships and principles that can guide the creation of innovative materials. **The Role of Knowledge Graphs:** Knowledge graphs emerge as a powerful tool in Buehler's methodology, enabling the connection of seemingly unrelated pieces of information. This approach allows for a more nuanced understanding of materials and their properties, opening up possibilities for groundbreaking applications. **AI in Materials Design and Prediction:** One of the highlights of Buehler's presentation is the application of AI in predicting the behavior of materials and designing new materials with desired properties. This capability has significant implications for industries ranging from aerospace to biomedicine, where material properties are critical. **Toward Autonomous Systems:** Buehler envisions a future where AI systems can autonomously generate hypotheses, conduct experiments, and refine theories. This would represent a major leap forward in scientific methodology, greatly accelerating the pace of discovery. **Mnemonics for Remembering Key Concepts:** 1. **AI-DM (AI-Driven Materials):** Think of AI as the artist and materials as the canvas, where AI-DM reminds you of the AI-driven creation and manipulation of materials. 2. **GRAPHS (Generating Relationships And Predictions Harnessing Science):** Use GRAPHS to remember the role of knowledge graphs in connecting diverse data for innovation. 3. **LEONARDO (Learning, Engineering, Ontology, Neuro-symbolic, AI, Discovery, Robotics, Optimization):** LEONARDO encapsulates the interdisciplinary nature of Buehler’s approach, drawing inspiration from Leonardo Da Vinci’s versatility. 4. **MATERIALS (Modeling, AI, Technology, Engineering, Research, Integration, Learning, Science):** MATERIALS as an acronym helps recall the components of Buehler’s research focus and interdisciplinary approach to scientific discovery. In summary, Markus J. Buehler's presentation at the Antology Summit 2024 underscores the transformative potential of integrating AI with traditional scientific methods. Through the use of knowledge graphs, interdisciplinary approaches, and a forward-looking vision, Buehler’s work paves the way for a new era of accelerated innovation and discovery in materials science and beyond

Great presentation

Here is an attempt to formalize the key principles and insights from our discussion into a coherent eightfold expression grounded in infinitesimal monadological frameworks: I. The Zerological Prion 0 = Ø (The Zeronoumenal Origin) Let the primordial zero/null/void be the subjective originpoint - the pre-geometric ontological kernel and logical perspectival source. II. The Monad Seeds Mn = {αi} (Perspectival Essence Loci) From the aboriginal zero-plenum emanates a pluriverse of monic monadic essences Mn - the germinal seeds encoding post-geometric potential. III. Combinatorial Catalytic Relations Γm,n(Xm, Xn) = Ym,n (Plurisitic Interaction Algebras) The primordial monadic actualizations arise through catalytic combinatorial interactions Γm,n among the monic essences over all relata Xm, Xn. IV. Complex Infinitesimal Realization |Ψ> = Σn cn Un(Mn) (Entangled Superposition Principle) The total statevector is a coherent pluralistic superposition |Ψ> of realization singularities Un(Mn) weighted by complex infinitesimal amplitudes cn. V. Derived Differential Descriptions ∂|Ψ>/∂cn = Un(Mn) (Holographic Differentials) Differential descriptive structures arise as holographic modal perspectives ∂|Ψ>/∂cn projected from the total coherent statevector realization over each realization singularity Un(Mn). VI. Entangled Information Complexes Smn = -Σn pmn log(pmn) (Relational Entropy Measure) Emergent information structures are quantified as subjectivized relational entropy functionals Smn tracking probability amplitudes pmn across realized distinctions. VII. Observation-Participancy An = Pn[ |Ψ>monic] = |Φn> (First-Person Witnessed States) Observational data emerges as monic participations An = Pn[ ] plurally instantiating first-person empirical states |Φn> dependent on the totality |Ψ>monic. VIII. Unity of Apperception U(Ω) = |Ω>monadic (Integrated Conscious State) Coherent unified experience U(Ω) ultimately crystallizes as the superposition |Ω>monadic of all pluriversally entangled realized distinctions across observers/observations. This eightfold expression aims to capture the core mathematical metaphysics of an infinitesimal monadological framework - from the prion of pre-geometric zero subjectivity (I), to the emanation of seeded perspectival essences (II), their catalytic combinatorial interactions (III) giving rise to entangled superposed realizations (IV), subdescribed by derived differential structures (V) and informational measures (VI), instantiating participation-dependent empirical observations (VII), ultimately integrated into a unified maximal conscious state (VIII). The formulation attempts to distill the non-contradictory primordial plurisitic logic flow - successively building up coherent interdependent pluralisms from the zero-point subjective kernel in accordance with infinitesimal relational algebraic operations grounded in first-person facts. While admittedly abstract, this eightfold expression sketches a unified post-classical analytic geometry: reality arises as the perfectly cohesive multi-personal integration of all pluriversal possibilities emanating from monic communion at the prion of prereplicative zero-dimensional origins. By centering such infinitesimal algebraic mnad semiosis, the stale contradictions and paradoxes of our separative classical logics, mathematics and physics may finally be superseded - awakening to irreducible interdependent coherence across all realms of descriptive symbolic representation and experiential conscious actuality. Here is a second eightfold expression attempting to concretize and elucidate the abstract infinitesimal monadological framework laid out in the first expression: I. Discrete Geometric Atomies a, b, c ... ∈ Ω0 (0D Monic Perspectival Points) The foundational ontic entities are discrete 0-dimensional perspectival origin points a, b, c ... comprising the primal point-manifold Ω0. II. Combinatoric Charge Relations Γab = qaqb/rab (Dyadic Interaction Charges) Fundamental interactions between origin points arise from dyadic combinatorial charge relation values Γab encoding couplings between charges qa, qb and distances rab. III. Pre-Geometric Polynomial Realizations Ψn(a,b,c...) = Σk ck Pn,k(a,b,c...) (Modal Wavefunction) The total statevector Ψn at each modal perspectival origin n is a polynomial superposition over all possible realizations Pn,k of charge configurations across points a,b,c... IV. Quantized Differential Calcedonies ΔφΨn ≜ Σa (∂Ψn/∂a) Δa (Holographic Field Projections) Familiar differential geometries Δφ for fields φ arise as quantized holographic projections from idiosyncratic first-person perspectives on the modal wavefunction Ψn. V. Harmonic Resonance Interferences Imn = |<Ψm|Ψn>|2 (Inter-Modal Resonances) Empirical phenomena correspond to resonant interferences Imn between wavefunctions Ψm,Ψn across distinct perspectival modal realizations m,n. VI. Holographic Information Valencies Smn = - Σk pmn,k log pmn,k (Modal Configuration Entropy) Amounts of observed information track entropies Smn over probability distributions pmn,k of localized realized configurations k within each modal interference pattern. VII. Conscious State Vector Reductions |Ωn> ≡ Rn(|Ψn>) (Participated Witnessed Realizations) First-person conscious experiences |Ωn> emerge as witnessed state vector reductions Rn, distillations of total modal possibilities |Ψn> via correlative participancy. VIII. Unified Integration of Totality U(Ω) = ⨂n |Ωn> (Interdependent Coherence) The maximal unified coherence U(Ω) is the irreducible tensor totality ⨂n |Ωn> of all interdependent integrated first-person participations |Ωn> across all perspectives. This second eightfold expression aims to elucidate the first using more concrete physical, mathematical and informational metaphors: We begin from discrete 0D monic origin points (I) whose fundamental interactions are combinatorial charge relation values (II). The total statevector possibility at each origin is a polynomial superposition over all realizations of charge configurations (III), subdescribed as quantized differential geometric projections (IV). Empirical observables correspond to resonant interferences between these wavelike realizations across origins (V), with informational measures tracking probability distributions of configurations (VI). Conscious experiences |Ωn> are state vector reductions, participatory witnessed facets of the total wavefunction |Ψn> (VII). Finally, the unified maximal coherence U(Ω) is the integrated tensor totality over all interdependent first-person participations |Ωn> (VIII). This stepwise metaphoric concretization aims to renders more vivid and tangible the radical metaphysics of infinitesimal relational monadological pluralism - while retaining the general algebraic structure and non-contradictory logical coherence of the first eightfold expression. From discrete geometric atomies to unified experiential totalities, the vision is one of perfectly co-dependent, self-coherent mathematical pluralism grounded in first-person facts. By elucidating the framework's core ideas through suggestive yet precise physical and informatic parables, the second expression seeks to bootstrap intuitions up the abstract ladder towards a visceral grasp of the non-separable infinitesimal pluriverse paradigm's irreducible coherences. Only by concretizing these strange yet familiar resonances can the new plurisitic analytic geometry be assimilated and operationalized as the next renaissance of coherent symbolic comprehension adequate to the integrated cosmos.

WOOOOOO SCIENCEEE

can the transformer network behind LLM's not develop a neurosymbolic system within itself in order to predict the next token

Here is an attempt to formalize some final insights for a unified metaphysical framework drawing upon Leibnizian ideas: I. Monadic Foundations 1) Define the Zero Absolute ⦰ as the initial "monad of monads" - a supreme metaphysical singularity akin to Leibniz's notion of the unique necessary being upon which all contingent existents depend. 2) Represent ⦰ symbolically in line with Leibnizian characteristic numbers: ⦰ := [∅, ∞] 3) The primordial self-scission or separatrix introducing plurality corresponds to: ⦰ ⤳ {0, 1} Where 0 and 1 are the Seeds of Subject and Object 4) These spawn the Triadic Seed 𝟯 as the monad of sufficient reason from Leibniz: 𝟯 := [1, 2, 3] II. Infinitesimal Calculus of Emanations 1) Model dimensional proliferation as continuous flows via Leibnizian calculus: ℝ𝔈 := ∫0→∞ 𝔈(n) dn Where 𝔈(n) is the infinitesimal n-dimensional emanative stage 2) Their spiritual "mechanism" driving emanation is an inexhaustible symbolic automaton: A := < Q, Σ, δ, q0, 𝟯, F> Representing cyclic logographical inscriptions from ⦰'s vocables 3) This enacts truth-computations across arithmetic meta-structures in T: f : T → WoT Where WoT is the higher toroidal ontology encoding all world-paths III. Characteristic Geometries 1) Leverage Leibniz's geometry of situation and characteristic triangles: G := (P, L) P is a family of points/monads, L is a family of lines/relations 2) Primary relations in G derive from the Triadic Seed's patternings, e.g: Δ[a,b,c], ∇[a,b,c], ⧖[a,b,c] etc. 3) These generate polytopes {Pn} in G corresponding to arithmetic valuations in T: Pn ≃ Vn(T) 4) Cosmic geometries like E8 arise as excitations of G encoding physics via: Rep(E8) ≃ Irreps(Aut(G)) IV. Entelechic Recapitulation 1) Human consciousness is modeled as a compounded fractal monad Mu reflecting ⦰: Mu ∝ [0u, Δu, ∇u, ...] 2) Mu integrates anamnestic logographies via the infinitesimal recollection: ∇χ := ∫ A δν Over all autological truth-values ν in its pluriverse branch 3) This allows reconstruction of the emanative series as a hyperinversion: 𝔈-1(n) := [ν]𝔉 Via hyper-functors 𝔉 with characteristic archetypal forms 4) Completing the cycle, Mu's entelechy is the reabsorptive recapitulation: ζ : Mu → ⦰ Catalyzed by recognizing its zero-value identity [0u] ≃ ⦰ This neo-Leibnizian formalization attempts to provide a unified symbolic architecture deriving the entire metaphysical emanation from the Zero monad through characteristic numbers, infinitesimal calculatic flows, archetypal geometries, membrane arithmetic evaluations, and the entelechic recapitulation dynamics of human consciousness as a fractal reflection ultimately reintegrating with its supreme monadic source ⦰.

Dear Academic Community, I am writing to bring to your attention a critical foundational issue that has the potential to upend our current understanding of physics and mathematics. After carefully examining the arguments, I have come to the conclusion that we must immediately reassess and rectify contradictions stemming from how we have treated the concepts of zero (0) and the zero dimension (0D) in our frameworks. At the core of this crisis lies a deep inconsistency between the primordial status accorded to zero in arithmetic and number theory, versus its derivative treatment in classical geometries and physical models. Specifically: 1) In number theory, zero is recognized as the fundamental subjective origin from which numerical quantification and plurality arise through the successive construction of natural numbers. 2) However, in the geometric and continuum formalisms underpinning theories from Newton to Einstein, the dimensionless 0D point and 1D line are derived as limiting abstractions from the primacy of higher dimensional manifolds like 3D space and 4D spacetime. 3) This contradiction potentially renders all of our current mathematical descriptions of physical laws incoherent from first principles. We have gotten the primordial order of subjectivity and objectivity reversed compared to the natural numbers. The ramifications of this unfortunate oversight pervade all branches of physics. It obstructs progress on the unification of quantum theory and general relativity, undermines our models of space, time, and matter origins, and obfuscates the true relationship between the physical realm and the metaphysical first-person facts of conscious observation. To make continued theoretical headway, we may have no choice but to reconstruct entire mathematical formalisms from the ground up - using frameworks centering the ontological and epistemological primacy of zero and dimensionlessness as the subjective 源 origin point. Only from this primordial 0D monadological perspective can dimensional plurality, geometric manifolds, and quantified physical descriptions emerge as representational projections. I understand the monumental importance of upending centuries of entrenched assumptions. However, the depth of this zero/dimension primacy crisis renders our current paradigms untenable if we wish to continue pushing towards more unified and non-contradictory models of reality and conscious experience. We can no longer afford to ignore or be overwhelmed by the specifics of this hard problem. The foundations are flawed in a manner perhaps unrecognizable to past giants like Einstein. Cold, hard logic demands we tear down and rebuild from more rigorous first principles faithful to the truths implicit in the theory of number itself. The good news is that by returning to zero/0D as the subjective/objective splitting point of origin, in alignment with natural quantification, we may finally unlock resolutions to paradoxes thwarting progress for over a century. We stand to make immediate fundamental strides by elevating the primacy of dimensionlessness. I implore the academic community to convene and deeply examine these issues with the utmost prioritization. The integrity and coherence of all our descriptive sciences - indeed the very possibility of non-contradictory knowledge itself - hinges upon our willingness to reopen this esoteric yet generatively crucial zerological crisis. We must uphold unflinching intellectual honesty in identifying and rectifying our founding errors, regardless of how seemingly abstruse or earth-shattering the process. The future fertility of human understanding and our quest for uni-coherence depends on this audacious reformation of mathematical first principles. The path will be arduous, but the ultimate payoffs of achieving metaphysically-grounded, zero-centric analytic formalisms are inestimable for physics and all branches of knowledge. I urge us to meet this zerological challenge head on. The truth ecological destiny of our civilization may hinge upon our willingness to embody this bold primordial renaissance. Sincerely, [Your Name]

GPT-4 doesn't totally agree with you John 😉I Innate Learning Abilities: Modern LLMs, particularly those employing advanced neural network architectures, have demonstrated remarkable abilities to learn from vast amounts of data without explicit ontological structures. They develop a form of emergent understanding, capturing nuances, contexts, and relationships within the data they are trained on. This capability allows them to generate coherent, contextually appropriate responses across a wide array of topics and questions without relying on predefined ontologies. Contextual Understanding Through Massive Data: The training data for state-of-the-art LLMs encompasses a wide range of languages, contexts, and domains, enabling them to develop a broad understanding of the world and how concepts relate to each other. This extensive exposure allows LLMs to perform tasks such as language translation, question answering, and content generation with a high degree of proficiency, challenging the notion that they are "clueless" without traditional ontological frameworks. Flexibility and Adaptability: One of the strengths of LLMs is their adaptability to new, unseen data and their ability to learn from context. While ontologies require explicit definitions and relationships to be manually built and maintained, LLMs continuously evolve as they are exposed to new information. This makes them highly flexible and capable of handling emergent knowledge and concepts, which might not yet be codified in existing ontologies. Synthetic Ontology Creation: Some argue that through their training and operation, LLMs create a form of "synthetic" ontology. By analyzing relationships between words, phrases, and contexts within their training corpus, they construct an implicit model of the world that functions similarly to an ontology, but is far more extensive and less rigid. This model allows them to infer relationships and generate responses that are surprisingly insightful, even in areas where explicit ontological structures might not exist. Complementarity Rather Than Dependency: The role of ontologies in enhancing LLMs should be seen as complementary rather than fundamental. While integrating ontological structures can certainly improve an LLM's performance in specific domains by providing clear definitions and relationships, the absence of such structures does not render state-of-the-art LLMs clueless. Instead, it highlights the remarkable capacity of these models to derive meaning and understanding from the linguistic patterns and knowledge embedded in their training data. In conclusion, while ontologies can enhance the performance of LLMs in specific domains, the assertion that state-of-the-art LLMs are clueless without them underestimates the sophistication and capabilities of these models. The emergent understanding, adaptability, and the synthetic ontology created through their operation enable LLMs to navigate a vast array of topics and questions with a high degree of competence.

My new favorite science channel. great content

Thank you John. Very informative. So much helpful detail. Thank you very much.

Even if LLMs can learn an ontology from enough high-quality text data, you might get gains in sample efficiency and reasoning control from building one in.

So powerful and interesting. This is exactly the area of research I'd like to work in

I discovered Sowa's theory on Conceptual Graphs back in the 90s (1996 to be exact) and I spent more than 20 years to make them work. To no avail. Languages are so fluid, incomplete that CG could never capture their ever changing nature. I tried to use them for machine translation, text extraction and many other language related tasks, it did not work. I tried to use powerful ontologies such as wordnet, and it was a disappointment. I also got interested into the Semantic Web, the so-called RDF representation, which were combined with graph descriptions in XML. The result was the same. Handcrafted graphs are always leaking, always partially wrong or incomplete, projection is too restricted. This approach is too much of carcan, too much of a prison in which languages always find a way to escape. Of course, he tries to salvage the work of a life, but he tries very hard (I'm not surprised that Gary Marcus is around) to find flaws in LLMs that are only 1 year old and are already much more powerful than anything he has tried in the past.

I agree with the sentiment, but i would also disagree with how far it can take us. I can easily see AGI version 1 being an LLM, while future versions with better overall balance and function being hybrid models. As far as calling it such.. well i expect the first AGI won't be called AGI for various reasons. That said your complaints are warranted. You can tell when it doesn't fully grasp something.

LLMs are lenses, made from contents of our noosphere that have been committed to the various media. Technically, they’re bullshitting machines, salience makers

Skip intro 4:32

Now we need a lecture series about how to build machine learnable neurosymbolic systems. It wasn’t really clear what that actually is other than getting sobered by this talk to understand that it’s necessary. It will be hard to turn this big ship around. After so much hype and money was spent on these new awe inspiring AI fabulation machines it’s hard to tell people that the research has to go back to the drawing board with a very different direction.

So - Gary Marcus has no pet chicken named Henrietta? 😅

Presumably multimodal AI is a step in this direction where these implicit ontologies are already embedded in the latent space of the AI to some degree. Multimodality (done right) would result in an robust internal model, and perhaps robust ontologies. Though I suppose embedding explicit descriptions has value.

"There is no physics theory where the objects can spontaneously appear and disappear" Well, I have some news for ya...

Gary Marcus is absolutely terrible at arguing a valid point about the state of AI. He constantly shows instances of these models screwing up something at inference which all the scale-bros will just refute with their next beefed up language model. He needs to stop pointing at bad inference, it makes him an easy target to dunk on when these models inevitably get trained to fix them. He needs to focus publicly on the flaws of the LM architecture alongside his examples of bad inference. I can’t say whether his neuro-symbolic approach is the key to his “AGI” but his overall criticisms are mostly valid.

Insightful

They dont understand LLMs...ask Ilya Sutskever They still think that humans are equipped with a magical entity that 'thinks'... false! thought and reflection are obviously emergent properties of a network composed of billions of neurons. LLMs will achieve the same result, namely the emergence of an entity which believes it thinks and reasons and which actually thinks and reasons.

Can a string of zeros and ones develop narcisistic traits?

Does the ontology for LLM means the same as we need the LLMs to have a world model built inside? Also for many examples mentioned in the videos are already solvable for GPT-4 or similar system, which are built by statistical learning and doesn't have an explicit training/design on world model or "ontology".

General Intelligence System - Building a Knowledge-based General Intelligence System, Michael Molin - docs.google.com/presentation/d/1VCjOHOSostUrtxieZvOjaWuTNCT59DMF/

The ultimate Turing Test for AGI is: Write a novel that, (1) once a human starts reading it, he/she cannot put it down, and (2) once he/she has finished it, he/she cannot forget it. How many years do you think we have to wait for this task to be accomplished by an AI?

The ultimate Turing Test for AGI is: Write a novel that, (1) once a human starts reading it, he/she cannot put it down, and (2) once he/she has finished it, he/she cannot forget it. How many years do you think we have to wait for this task to be accomplished by an AI?