What is the minimal model to align agency and symmetrify information flow by reducing asymmetries across nested Markov blankets running on arbitrary computational substrate (maximizing money, morals, status) with arbitrary restrictions (democracy, regulated market) What mathematics does the universe run on? How to make life flourish infinitely? What's beyond our useful mental constructions for survival? Is realism also just useful mental construction? Is the concept of mental constructions just another useful mental construction? Our body is a love machine If you feel stuck, its often because you've been caught in some small minded game, when stuck try to pivot to a bigger more important (to you) game Pleasure can be seen as brain's instrumental tool for purpose that is projecting agency into the future Current interpretability work is just on very low level concrete features as facts or patterns of behavior. But we can steer it so that it happily wants to kill all people. High level reasoning is still mysterious. What matters is behavior. I dont care about linguistics. It's running on objective functions you put into it. If you prompt a model "do something harmful" and it has enough capabilities, it can just mindlessly go for it. Or you can prompt it to do a good thing but instrumental goal will be immoral to us. https://www.lesswrong.com/tag/žein https://www.lesswrong.com/posts/F4iogK5xdNd7jDNyw/comparing-anthropic-s-dictionary-learning-to-ours "huh, reframing ADHD as Intention Deficit Disorder highlights why spending a few minutes at the start of each day reminding yourself what your intentions are works so well" https://fxtwitter.com/OrphicCapital/status/1712127412018782383 One framework for making LLMs better https://fxtwitter.com/IntuitMachine/status/1712426605501399536?t=s1tQMZfgl1AxBSjUxMo03A&s=19 Greg (OpenAI cofounder) says we can predict model capabilities and Sam (OpenAI cofounder) says we can't. Wtf? XD [AI SAFETY PANEL | Elon Musk, Max Tegmark, Greg Brockman, Benjamin Netanyahu - YouTube](https://youtu.be/Dg-rKXi9XYg?si=2AWXd5fPtyaFKeaF) 5:00 https://fxtwitter.com/liron/status/1709207952509604205?t=3FU9c-TDLS1TsdxRgqjzdw&s=19 But realitically saying we can predict that black box of gazilion of layers of transformers seems unreal to me https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8356896/ Machine Learning in Drug Discovery: A Review [Security bug - Wikipedia](https://en.wikipedia.org/wiki/Security_bug?wprov=sfla1) [Shellshock (software bug) - Wikipedia](https://en.wikipedia.org/wiki/Shellshock_(software_bug)?wprov=sfla1) [Heartbleed - Wikipedia](https://en.wikipedia.org/wiki/Heartbleed?wprov=sfla1) [Formal verification - Wikipedia](https://en.wikipedia.org/wiki/Formal_verification?wprov=sfla1) [[2309.01933] Provably safe systems: the only path to controllable AGI](https://arxiv.org/abs/2309.01933) Provably safe systems: the only path to controllable AGI Utok [Now UTOKing | Learning the Language - YouTube](https://youtube.com/playlist?list=PLTJe1xFfoxrDi19ew-Zbo9oDOwv_c-Ysz) Unemployment from AI We had planes faster than we understood birds because evolution was much more constrained in terms of building flying machines. When building birds, evolution had to build selfassembling machines, which planes aren't, it had to use the most common periodic table elements, it had it had to be fuel efficient, where planes aren't as constrained this way. Current AIs seem to be on a similar track. If we put a ton of compute on very distant from but still similar to brain architecture in different context and way of training it, we get large language models like ChatGPT or other similar systems, while the human brain eats almost no electricity compared to our biggest AIs, but the massive compute we pour into them makes them so intelligent in various similar ways, but still in different ways. Also the last time evolution built more intelligent species, mass extinction happened (neanderthals) and we may be making the same this with building superintelligent AI now. [Philosophy of space and time - Wikipedia](https://en.wikipedia.org/wiki/Philosophy_of_space_and_time) [AI alignment - Wikipedia](https://en.wikipedia.org/wiki/AI_alignment) [PaperStream #006.0 ~ Active Inference in Modeling Conflict - YouTube](https://www.youtube.com/watch?v=HAujw2_ClCM) PaperStream #006.0 ~ Active Inference in Modeling Conflict [Conflict theories - Wikipedia](https://en.wikipedia.org/wiki/Conflict_theories) There are many foundations of mathematics and logical systems and there is warfare among them [How the surprising muon revolutionized particle physics - Big Think](https://bigthink.com/starts-with-a-bang/surprising-muon-particle-physics/) https://www.sciencedirect.com/science/article/pii/S1571064508000250 Life, gravity and the second law of thermodynamics [The Theory of Everything: Searching for the universal rules of physics | Space](https://www.space.com/theory-of-everything-definition.html) https://www.lesswrong.com/posts/9QxnfMYccz9QRgZ5z/the-costly-coordination-mechanism-of-common-knowledge [a physicist responds: physics has done very little for like 70 years - YouTube](https://youtu.be/d_o4k0eLoMI?si=WyPwN6XH51Owpp8u) a physicist responds: physics has done very little for like 70 years In physics in the last 70 years we managed to discover/validate/engineer simulations, fMRI, space exploration, gravitational waves, black hole picture, Bose Einstein condesate, higgs boson, (quantum) computers, internet, superconductivity, lasers, atomic clocks, tao neutrino, cosmic microwave background, graphene,... but we still struggle with quantum gravity, unifying fundamental forces (one can argument if thats really needed), other problems in physics like information black hole paradox or dark matter, integrating biology, consciousness and other fields with physics, integrating it all into one giant mathematical framework with mathematical bridges across parts and scales (one can argument if thats really needed as well). Time dillation Relational intepretation of QM by Carlo Rovelli Measuring QM particles with lasers [Van der Waals force - Wikipedia](https://en.wikipedia.org/wiki/Van_der_Waals_force) Its forces between particles across scales all the way down On one hand i wanna learn physics because knowledge about the nature of reality there is neutral wihout arbitrary assumptions and technically strict and exact which makes it so comfy as it isnt hidden under 1000 layers of ideological baggage and other kinds of noise but at the same time world is burning and i feel responsibility for helping it and physics is limited in that (might spawn another nuclear weapon instead heh) Scaling quantum computers soon? https://twitter.com/skdh/status/1712685313779278023?t=_LJ51PrYuNbZ6srV1ruWdQ&s=19 BCIs [Brain–computer interface - Wikipedia](https://en.wikipedia.org/wiki/Brain–computer_interface?wprov=sfla1) [Brain Computer Interface: A Review | SpringerLink](https://link.springer.com/chapter/10.1007/978-981-16-4035-3_3) [Frontiers | Progress in Brain Computer Interface: Challenges and Opportunities](https://www.frontiersin.org/articles/10.3389/fnsys.2021.578875/full) Neuralink Best initial assumptions are those that are general enough to accompany as many existing predictive models as possible inside them and that unify those models into one language If nature is free energy minimizing, having dynamics satisfying principle od least action, across scales, then all equations in dynamics across scales are just solutions to this general problem of nature trying to be as lazy as possible [A brief history of quantum mechanics - Big Think](https://bigthink.com/collections/brief-history-quantum-mechanics/?utm_campaign=weeklynewsletter&utm_source=rejoiner&utm_medium=email&utm_content=10%2F12%2F23+Smarter+Faster+%28A%29&rjnrid=dDXnEnJ) I think all philosophical assumptions are valid, but some are more useful than others in different kinds of (pragmatic) contexts for different goals [Give to outstanding charities | Effective Altruism](https://www.effectivealtruism.org/get-involved/give-to-outstanding-charities) You must do metaoptimizingmaxxing, optimizing the process of optimizing itself. Optimizing the process of optimizing itself and apply the optimized optimization process to optimize good things more efficiently. if we equate ontology with God: God is ontology God is incomprehensibility - transcendentalism God is undefinable - mysterianism God is reality God is nowness - empty individualism God is universal consciousness - panpsychist idealism God is statespace of consciousness - panpsychist idealist platonism God is nature - naturalism God is a mathematician picking differential equations God is a computer programmer picking programs God is a computer program - constructivism, computationalism God is all possible computer programs - multiverse computationalism God is all of mathematics - platonism God is certain subset of mathematics in physics - mathematical structural physicalism God is all possible physical laws - multiverse physicalism God is physics beyond mathematics with nonmathematizable parts - semiphysicalism God is everything possible - omnism God is nothing - zero ontology [The Zero Ontology - David Pearce on Why Anything Exists](https://www.hedweb.com/witherall/zero.htm) God is neither something nor nothing nor everything but all of it in parralel -nonclassicalism God is whole statespace of possible God definitions - omnidefinism Shapeshifting between all possible ontologies in parallel Lets explore his insides, painting painting itself, dancing on its own head https://twitter.com/burny_tech/status/1712845359959957827 https://twitter.com/zerfas33/status/1712699939229401431?t=vEyMdy4RYDU7FnAYPN-jUQ&s=19 In Ruliad all possible laws of physics are taking place at the same time, they're all entangled, and the observer selects laws by having an illusion that some particular laws are the ones that apply where Ruliad is what really exists with its all possible applications of all possible laws at the same time. Ruliad is equivalent to infinity groupoid from infinity category theory modern platonism is for example max tegmark's mathematical universe in his framework all possible mathematical universes exist a form of mathematicism in that it denies that anything exists except mathematical objects; and a formal expression of ontic structural realism [Mathematical universe hypothesis - Wikipedia](https://en.wikipedia.org/wiki/Mathematical_universe_hypothesis) we can assign realism label to anything and that creates our interpretations of reality for me the notion of realism is just pragmatic or anarchic depending if im in optimizing science vs exploring the statespace of consciousness mood if it yields prediction, i eat it if it doesnt, i can also eat it but just for fun if you buy into mathematical universe hypothesis you can interpret DMT trips as traversing different mathematical universes the fields of physics are fields of love let us dissolve into them on do nothing meditation plus 5-meO-DMT macrodose "i" love love implying there is some "i" "loving" some "love" implying there is some x doing y onto z no process reality is beyond processes but includes them all dissolving the dissolving itself epistemic nihilism by epistemic nihilism is death itself as well kjwhzfuáhnkyviaubvguhakdsvl faith is the glue that holds beliefs together gaia1 has a full world model of how the world works with phsics and such as a little inner simulation, grand theft auto 7 soon https://twitter.com/alexgkendall/status/1709607548582576141 Selfdriving electirc Cars https://www.eaccwiki.com/wiki/Main_Page [Where Do Space, Time and Gravity Come From? | Quanta Magazine](https://www.quantamagazine.org/where-do-space-time-and-gravity-come-from-20220504/) "The smallest positive integer not definable in under sixty letters." Berry's paradox is odd selfreferential metalanguage paradoxical chaos [Berry paradox - Wikipedia](https://en.wikipedia.org/wiki/Berry_paradox) Chaitin's incompleteness theorem is build on same kind of logic "No program P computing a lower bound for each text's Kolmogorov complexity can return a value essentially larger than P's own length, hence no single program can compute the exact Kolmogorov complexity for infinitely many texts." [Kolmogorov complexity - Wikipedia](https://en.wikipedia.org/wiki/Kolmogorov_complexity#Chaitin's_incompleteness_theorem) "If i could prove that a specific object requires very big programs, and I'm proving this from a smaller amount of bits, I can run through all the proofs in your algorithmic formal axiomatic theory until I find the best first object that provably requires a lot more bits than bits in your theory. I calculated that object which supposedly neds a very large program, but I've calculated it from a much smaller number of bits which is the number of bits in your theory." [Gregory Chaitin: Complexity, Metabiology, Gödel, Cold Fusion - YouTube](https://youtu.be/zMPnrNL3zsE?si=ZCCRW2QOz95nn59D&t=1095) Infinite complexity incompressible into finite set of axioms through this program's equivalence with pure randomness [Chaitin's constant - Wikipedia](https://en.wikipedia.org/wiki/Chaitin%27s_constant) breaking all of math's idea of compressing all of infinite mathematical complexity into finite set of axioms. For a computer program that will output the first n bits of the halting probability, you need a computer program that is n bits long. You would need n bits of axioms and laws of inference to be able to determine n bits of the halting probability. Theory of everything in math is impossible, or finitism to the rescue, in both math and physics, as all paradoxes, incompleteness, irreducibility results in math stem from infinities? [Gregory Chaitin: Complexity, Metabiology, Gödel, Cold Fusion - YouTube](https://youtu.be/zMPnrNL3zsE?si=JPOMMg_ps5jWTqwf&t=1567) [No free lunch theorem - Wikipedia](https://en.wikipedia.org/wiki/No_free_lunch_theorem?wprov=sfla1) "any two optimization algorithms are equivalent when their performance is averaged across all possible problems", optimization algorithms are context dependent, no one is universally performing perfectly in all contexts List of explanations to why is there something rather than nothing Mendelian genetics Population genetics Information theoretic metabiology https://scholar.google.co.uk/citations?user=P6z3U-wAAAAJ&hl=en https://scholar.google.com/citations?user=DUVtbIMAAAAJ&hl=es Molecular biology says information is the most important [Molecular biology - Wikipedia](https://en.wikipedia.org/wiki/Molecular_biology?wprov=sfla1) https://philpapers.org/rec/CHAPDM-2 Gregory Chaitin, Proving Darwin: Making Biology Mathematical [How did we prove most numbers are “uncomputable” “normal numbers” if we’ve never found any of them? - Quora](https://www.quora.com/How-did-we-prove-most-numbers-are-uncomputable-normal-numbers-if-we-ve-never-found-any-of-them) [Halting problem - Wikipedia](https://en.wikipedia.org/wiki/Halting_problem?wprov=sfla1) [Chaitin's constant - Wikipedia](https://en.wikipedia.org/wiki/Chaitin%27s_constant?wprov=sfla1) Superforcasting I wonder if we can construct a mathematical proof that societal complexity is irreducible because growing complexity by increase in scales disallows it as most of the dynamics become statistical randomness with the limited measument tools we have disallowing identifying mathematical patterns that compress dynamics David doich We can construct multiagent human simulations of societies by taking GPT instances as agents interacting by prompt engineering context, what they see about other agents and in the environment, what are their human basic needs, what is their memory, existing plans, and prompt for planning right now or planning more deeply long term to reach basic needs goals given circumstances, personality, learned skills, other conditionings, reinforcing good experiences, etc. [Learning Altruistic Behaviours in Reinforcement Learning without External Rewards | OpenReview](https://openreview.net/forum?id=KxbhdyiPHE) Learning Altruistic Behaviours in Reinforcement Learning without External Rewards [Donald Hoffman Λ Joscha Bach: Consciousness, Gödel, Reality - YouTube](https://www.youtube.com/watch?v=bhSlYfVtgww) [Michael Levin Λ Joscha Bach: Collective Intelligence - YouTube](https://www.youtube.com/watch?v=kgMFnfB5E_A) https://www.lesswrong.com/posts/3eqHYxfWb5x4Qfz8C/unrlhf-efficiently-undoing-llm-safeguards https://www.lesswrong.com/posts/qmQFHCgCyEEjuy5a7/lora-fine-tuning-efficiently-undoes-safety-training-from things exist independent of observers (objectivism) vs only when observed and constructed (subjectivism) and we share lots of parts of that hallucination that creates laws of physics, norms etc. feynman lectures Can we concretely mathematically explain the diversity of evolution? What (statistical) evolutionary pressures formed plants and humans? Math predicting empirical data is ontology agnostic https://fxtwitter.com/d_feldman/status/1713019158474920321?t=vzHqhPEpremg1mOwrLB0sA&s=19 https://fxtwitter.com/d_feldman/status/1713019290641670458?t=DrBAnDCCwMT-MZXOdvDdtw&s=19 [Multi-modal prompt injection image attacks against GPT-4V](https://simonwillison.net/2023/Oct/14/multi-modal-prompt-injection/) United States of Wireheaded Supercooperation Cluster Palestine x Israel: Is this a dispute over land, or about which cultures dominate over a region, and how that cultures are defining themselves, and highly fluid organizational, legal, religious, ethnic and social constructs being tied into these definitions in various complex and contradictory ways? Ethically I'm on the utilitarian side, minimize the total amount of suffering by any means of organizational, legal, religious, and social construct transformation. Same in Russia x Ukraine. On relativistic theory of consciousness https://fxtwitter.com/Nir_lahav/status/1704608508245881315?t=G7tRzZbH3LHmGlmAQsa05A&s=19 Whereof one cannot speak, thereof one must learn the language. Cross the transcendental horizon. Become unbounded. Accelerate utopian future Find bestest science Find bestest cooperation mechanisms Secure stability of clusters of sentient beings Hack sentient matter to infinite bliss Evolution of Biological Information: How Evolution Creates Complexity, from Virus to Brains - by Chris Adami (Princeton University press) [The Evolution of Biological Information | Princeton University Press](https://press.princeton.edu/books/hardcover/9780691241166/the-evolution-of-biological-information) i see attention as directed flow of information from a (observer) to be (what is being paid attention to) and other way around when recieving read information Fuzzy systems theory https://www.researchgate.net/publication/357935341_Fuzzy_Systems_Theory_and_Applications [Mathematical Systems Theory I: Modelling, State Space Analysis, Stability and Robustness | SpringerLink](https://link.springer.com/book/10.1007/b137541) Mathematical Systems Theory [Formal system - Wikipedia](https://en.wikipedia.org/wiki/Formal_system) [[2210.15098] Natural Language Syntax Complies with the Free-Energy Principle](https://arxiv.org/abs/2210.15098) Natural Language Syntax Complies with the Free-Energy Principle [Neuropsychology - Wikipedia](https://en.wikipedia.org/wiki/Neuropsychology) technooptimist manifesto [The Techno-Optimist Manifesto | Andreessen Horowitz](https://a16z.com/the-techno-optimist-manifesto/)