the complexity of the brain as a complex system goes beyond our comprehension i think its useful to assume you can get causal 1:1 mathematical correspondence between physics of the brain (including all the levels of abstraction that emerge) and qualia and constantly look for it (because of certain theoretical limits in math and practical limits in measurement we will probably always have just an approximations that get better and better overtime exponentially) and lots of it can be compressed using statistics that assumption already got us extremely far in sciences, for example all neurotech relies on it ELLIPTIC CURVES OVER FINITE FIELDS, THE ZARISKI TOPOLOGY, PERSISTENT HOMOLOGIES ON DATASETS, TENSOR CALCULUS ON MANIFOLDS, PACKING OF N-SPHERES, A BIO OF RAMANUJAN, KOSZUL OBJECTS IN TRIANGULATED CATEGORIES [Electron density-based GPT for optimization and suggestion of host–guest binders | Nature Computational Science](https://www.nature.com/articles/s43588-024-00602-x) https://arxiv.org/abs/2403.04704 bro, what do you mean the graph isomorphism problem is hard? bro, you just need to build a full fault-tolerant quantum computer and put space-time into a superposition it's that easy bro choose your character [top in nLab](https://ncatlab.org/nlab/show/top) [bottom in nLab](https://ncatlab.org/nlab/show/bottom) ještě existuje tenhle top, nebo hot top z topologie [Top in nLab](https://ncatlab.org/nlab/show/Top) [Ho(Top) in nLab](https://ncatlab.org/nlab/show/Ho%28Top%29) tohle je přes teorii kategorií zobecněný do hodně matematických vesmírů, ne jen kategorie Set/Hask co používájí programovací type systémy [Types and Functions |   Bartosz Milewski's Programming Cafe](https://bartoszmilewski.com/2014/11/24/types-and-functions/) mít vlastní motivační systém/cíle? na zásadě toho být autonomní agent? dost lidí argumentuje že to je to co z dosavadních AI systémů chybí, a být autonomně curious např Jürgen Schmidhuber [Artificial Curiosity Since 1990](https://people.idsia.ch/~juergen/artificial-curiosity-since-1990.html) to je AI researcher co inventnul hodně věcí co se dneska v AI používají, co věří, že artificial conscious systémy jsme tu už měli dávno https://fxtwitter.com/SchmidhuberAI/status/1765769164709371978 pro něho consciousness je byproduct datový komprese při problem solvingu https://arxiv.org/abs/2403.04121 Apply as much of these fields as possible to reverse engineering transformers in mechanistic interpretability: - **Foundations** - Category Theory - Information Theory - Mathematical Logic - Philosophy of Mathematics - Set Theory - Type Theory - **Algebra** - Abstract - Commutative - Elementary - Group Theory - Linear - Multilinear - Universal - Homological - **Analysis** - Calculus - Real Analysis - Complex Analysis - Hypercomplex Analysis - Differential Equations - Functional Analysis - Harmonic Analysis - Measure Theory - **Discrete** - Combinatorics - Graph Theory - Order Theory - **Geometry** - Algebraic - Analytic - Arithmetic - Differential - Discrete - Euclidean - Finite - **Number Theory** - Arithmetic - Algebraic Number Theory - Analytic Number Theory - Diophantine Geometry - **Topology** - General - Algebraic - Differential - Geometric - Homotopy Theory - **Applied** - Engineering Mathematics - Mathematical Biology - Mathematical Chemistry - Mathematical Economics - Mathematical Finance - Mathematical Physics - Mathematical Psychology - Mathematical Sociology - Mathematical Statistics - Probability - Statistics - Systems Science - Control Theory - Game Theory - Operations Research - **Computational** - Computer Science - Theory of Computation - Computational Complexity Theory - Numerical Analysis - Optimization - Computer Algebra https://twitter.com/burny_tech/status/1766164653011296261?t=QBWg9i2SfcrGwYvN3dCE6w&s=19 "💥“what causes polysemanticity?" was accepted to re-align/bgpt @ iclr!💥 tl;dr: with more features than neurons in a model, each neuron has to represent many features -> interpreting that is hard but even with enough neurons, you may still get one neuron for many features!" https://twitter.com/tmychow/status/1766147136369127514 [incidental polysemanticity](https://tmychow.com/posts/incidental_poly_0.html) Structured prompt enfineering Chollet model of generalizing discussion https://twitter.com/AndrewLampinen/status/1766135260327416228?t=A-MqA5r9arEFwT2gQIAJ-g&s=19 [Hamilton–Jacobi–Bellman equation - Wikipedia](https://en.wikipedia.org/wiki/Hamilton%E2%80%93Jacobi%E2%80%93Bellman_equation) Etale topology [The GPT-4 barrier has finally been broken](https://simonwillison.net/2024/Mar/8/gpt-4-barrier/) https://www.lesswrong.com/posts/SwcyMEgLyd4C3Dern/the-parable-of-predict-o-matic Climate change denial is harmful to our civilization. Climate change also doesn't mean stop growing or degrowth of civilization, that would be very harmful. Climate change can be solved using technology like bioengineering or adaptation. Claude cybersec https://twitter.com/JasonDClinton/status/1766233772805288006?t=A8CkYNDrzUwtmhGiOl5XDQ&s=19 https://arxiv.org/abs/2403.03925 Personally it can feel like that on meditation and psychedelics (or sometimes for longer in normal existing) but I don't see how assuming immeasurable things is any useful in the context of science and engineering, because you can postulate absolutely anything immeasurable existing, it feels arbitrary, mind can dream up so many arbitrary constructs there. I think that feeling can still be explained in the domain of physicalism and brain and that way engineered not only internally but also externally. I like this ontology too. Classical Spinoza's ontology. But I don't see why one has to do all these extra assumptions if you want to stay practically grounded in reality. I'm even for the possibility that observers are fundamental and the physics we observe in brains and outside of brains is one of many possible ones But I never really understood why are people so confident in nonphysical claims when you can dream up absolutely arbitrary constructs there and that doesn't seem weird to them, why would their lens there have to be the true one and others that go against it not. Unless it's much less motivated by scientific accuracy and relatively much more by spiritual euphoria. If guess if you contextualize this in cognitive science, I probably never gotten my brain to overly strengthen concrete beliefs in that domain And my objective functions possibly optimize for slightly different things in terms of what kind of models I reify As my main goal is the fusion of as predictive models for engineering as possible combined with them feeling euphoric (it can often go hand it hand, this doesn't have to be disconnected, as many people think) If you mean this kind of generalized entanglement (which unifies with just correlation) in quantum information theory and not from quantum mechanics in fundamental physics then you can look into Chris Fields. For waves, one can look at models seeing brain as classical nonlinear electrochemical wave computer. I think it's possible to be in as scientific/engineering mindset as possible while swimming in spiritual euphorias from models of reality being orgasmic Like, I've been there so many times, feeling like there is only deeply meaningful full of awe ieffable aliveness when deconstruction into void, but at the same time I believe that's a brain state corresponding to that qualia that you can explicitly engineer using neurotechnology, not just by our "will", and swim in it for how long you want. We're getting better and better at manipulating experience using neurotechnology, I don't think this is practically far off. Since Black hole is an object with the highest entropy, it contains the most saturated information in the universe We gotta act quickly I'm afraid that if we don't globally accelerate AI and other exponential tech enough we will fall into one of these: 1) Technology will get regulated to oblivion and we can forget about transhumanistic future with humanity growing and becoming intergalactical civilization. 2) Those with the biggest control will get there first and create monopolies of power essentially creating new kingdoms that nobody has power to disrupt if they go malicious without mechanisms for power checks and balances. 3) Malicious agents get there first and create controlled dystopia without collective growth for everyone. 4) Current unsustainable civilizational status quo configuration won't get destabilized and changed enough as a phase shift and all kinds of selfmade or natural existential risks from the planet or the universe will eventually kills us. We can become resistant to all of them, we are highly adaptive, maybe we can even fight the second law of thermodynamics in its limit. But acceleration can also go bad, like: 1) Disruptions so hardcore that the whole civilizational system cant adapt fast enough and get resistant fast enough and some existential risk will kills us, or tensions between individuals and all gigantic powerful corporations and nation states on earth get so tense that we bomb each other out of existence. 2) Creating new autonomous species million times smarter than us that will take control (like we asserted dominance over all the other animals) that won't share the desire to populate the universe and play the longest games by surviving infinitely I believe is a possibility too, but maybe not as likely one. We must mitigate all of these bad local minima to accelerate sentience to the stars. Someone: This is my perspective as an optimist systems thinker in case you'd like to hear 1) Not possible IMO due to competition between nations and corporations. 2) and 3) Monopoloy of power or bad guys taking control thanks to AGI is indeed a big cause for concern, I'll give you that. But you need to realise that the root cause of exploitation is not because of some evil human nature. It's because of lack of superabundance. If there is true superabundance, there is no longer any economic incentive for exploitaton. And believe it or not, exploitation has only existed due to the economic benefits it provides. 3) Current status quo not being flipped upside down as a result of very strong and cheap embodied AGI is exceedingly unlikely in my opinion. As I said, anything that leads to superabundance makes the current system obsolete. 4) True, it's always a possibility that we nuke ourselves to death. Instability will increase that chance. 5) I've shared this idea a few times. AI systems have no incentive to control or dominate a priori, whereas biological entities like humans do. AI does not have a survival agenda of their own. Any 'agenda' it does have is insofar as what we program into them. So as long as we're extremely careful about how we program AI, they won't suddenly develop the urge to destroy humanity, regardless of how much intelligence they possess. It's just not in their training. Me: Thanks, great perspecitve. I would argue that (out of distribution) emergence in AI systems is still a gigantic mystery, but maybe desire to dominate just won't have enough incentives to emerge aka instrumental convergence thesis being false, where proposed basic emergent AI drives include utility function or goal-content integrity, self-protection, freedom from interference, self-improvement, and non-satiable acquisition of additional resources. Or maybe all of this is on a spectrum. “The beauty of mathematics only shows itself to more patient followers.” - Maryam Mirzakhani Mathematician John Nash struggled with schizophrenia. His life, marked by his brilliant academic achievements and his battle with mental illness, was depicted in the biographical film “A Beautiful Mind.” Alan Turing significantly contributed to biology with his theory on how patterns like stripes on animals can form naturally. This work helped establish the field of mathematical biology. "I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted." — Alan Turing Alan Turing on mathematical reasoning ✍️ Mathematical reasoning may be regarded rather schematically as the exercise of a combination of two facilities, which we may call intuition and ingenuity. The activity of the intuition consists in making spontaneous judgements which are not the result of conscious trains of reasoning... The exercise of ingenuity in mathematics consists in aiding the intuition through suitable arrangements of propositions, and perhaps geometrical figures or drawings. - as mentioned in Proceedings of the London Mathematical Society (1939) Nature shows us only the tail of the lion. But there is no doubt in my mindthat the lion belongs with it even if he cannot reveal himself to the eye all at once because of his huge dimension. - A. Einstein Science, like art, not a copy of nature but a recreation of her Albert Einstein on Mahatma Gandhi ✍️ Gandhi's views were the most enlightened of all the political men in our time. We should strive to do things in his spirit... not to use violence in fighting for our cause, but by non-participation in what we believe is evil. Just by studying mathematics we can hope to make a guess at the kind of mathematics that will come into the physics of the future. A good many people are working on the mathematical basis of quantum theory, trying to understand the theory better and to make it more powerful and more beautiful. If someone can hit on the right lines along which to make this development, it may lead to a future advance in which people will first discover the equations and then, after examining them, gradually learn how to apply them. - Paul Dirac To me, the universe is simply a great machine which never came into being and never will end. The human being is no exception to the natural order. Man, like the universe, is a machine. Nothing enters our minds or determines our actions which is not directly or indirectly a response to stimuli beating upon our sense organs from without. - Nikola Tesla [AI/ML+Physics Part 3: Designing an Architecture [Physics Informed Machine Learning] - YouTube](https://www.youtube.com/watch?v=fiX8c-4K0-Q) [The String Theory Wars and What Happened Next - YouTube](https://www.youtube.com/watch?v=eRzQDyw5C3M) [The Area of the Mandelbrot Set - Stokes' Theorem - YouTube](https://www.youtube.com/watch?v=sZx7JAngPWM) [How Vector Addition Keeps Your Computer From Crashing - YouTube](https://www.youtube.com/watch?v=IzSs_gJDVzI) quanta magazine https://www.lesswrong.com/posts/JEhW3HDMKzekDShva/significantly-enhancing-adult-intelligence-with-gene-editing Claude learning prompt https://twitter.com/mattshumer_/status/1766520388308160894 Claude coding refactoring prompt https://twitter.com/mattshumer_/status/1766515138335584397 360 papers origin of life https://twitter.com/architectonyx/status/1766491411623657691 https://royalsocietypublishing.org/doi/10.1098/rsta.2020.0410 overview and mathematics of emergence