Effective Accelerationism has freedom to build technology as fundamental value, giving less weight to the probability of risks, more weight to it stopping prokaigress more than it creating destruction. Certain cases of Effective Accelerationism posit that there is no way how will humans not selfdestruct themselves by various risks or crises and instead we need to build AIs as fast as possible so that they can be the next step in the evolution of intelligence as humans are already doomed (perhaps positing that any information processing is source of moral value by perhaps all of it being conscious) but i suspect in practice this would just selfdestruct US even faster without any AIs after unless exponential growth of AIs getting better is truly So big that instead of AIs fighting and killing everything it will be suddenly one AGI absorbing everything extremely fast that nothing else can grasp it. I think in practice we should really think about risks but also about not slowing down progress totally. Middle way! And we can still do as much as we can to prevent molochian dynamics by strenghtening cooperation as much as possible to prevent global catastrophic failure modes, but I understand why some people lost hope in that and see AGI Aš the only way to preserve interesting things (intelligence) happening in the universe. But having lost hope means it decreases those chances of hope selfrealizing itself in the real world even more, as that not only predicts Future but also forms those beliefs also form our collective actions! Predicting reality is almost impossible as uncertainity grows and chatic nature of systems having totally different results in evolution from slight nudge in initial conditions like in weather, as humanity can be to certain degree seen as one giant weather system with slight mathematical correpondences from dynamical systems theory or systems science in general. [Joscha Bach—How to Stop Worrying and Love AI - YouTube](https://youtu.be/YeXHQts3xYM?si=scYKfu38lOlCBmex) [x.com](https://twitter.com/BasedBeffJezos/status/1705736424274686129?t=9yYOBRii9b2qPwjCmNo1gw&s=19) Driving licence for language models is good idea, but in practice because of lack of concenzus and expertise for such a global infrastructure we can end up with complex bureocratic systems of unprofessional people that are happy they know how to navigate Windows 11 that are those who are making decisions about AI driving licences. Could education or AIassisted masseducation save that? Can enough amount of people grasp the technical functioning of AI technologies, will there be enough expertise fór such infrastructure without too much beurocratic molochian dynamics? Driving licence for language psychedelics is good idea Computational neuroscience of happiness RL equation Quantum has x meaning Evolutionary competition of Future forming memeplexes propagate and merge [Levels of Lucidity - Joscha Bach](https://joscha.substack.com/p/levels-of-lucidity?utm_source=profile&utm_medium=reader2) [The Existential Risk of AGI - Joscha Bach](https://joscha.substack.com/p/the-existential-risk-of-agi?utm_source=profile&utm_medium=reader2) Counterargument to cooperation [Notes on Nggwal — Traditions of Conflict](https://traditionsofconflict.com/blog/2019/2/23/notes-on-nggwal) Aligning AGI! Framework is that all moral intuitions and moral systems of humans can be abstracted as cooperation protocols, [Moral foundations theory - Wikipedia](https://en.wikipedia.org/wiki/Moral_foundations_theory) [Cultural universal - Wikipedia](https://en.wikipedia.org/wiki/Cultural_universal) patterns in behavior resulting increased cooperation, for example having shared goals and infering states of others. [Entropy | Free Full-Text | An Active Inference Model of Collective Intelligence](https://www.mdpi.com/1099-4300/23/7/830) These cooperation protocols have neural correlates - localisable features encoded in topology of information networks. [Toy Models of Superposition](https://transformer-circuits.pub/2022/toy_model/index.html) You can transfer those protocols as these features geometries between the human brain and the transformer and hardcode it in the architecture, or as priors, or as first training data, or as how training gets shaped evolutionary. [Activation vector steering with BCI | Manifund](https://manifund.org/projects/activation-vector-steering-with-bci) Alternatively you can align existing models by RLHFing by prompting by asking the LLM to have the context of all these linguistic cooperation protocols. Aligning AGI! Framework is that all moral intuitions and moral systems of humans can be abstracted as cooperation protocols, [Moral foundations theory - Wikipedia](https://en.wikipedia.org/wiki/Moral_foundations_theory) [Cultural universal - Wikipedia](https://en.wikipedia.org/wiki/Cultural_universal) patterns in behavior resulting increased cooperation, for example having shared goals and infering states of others. [Entropy | Free Full-Text | An Active Inference Model of Collective Intelligence](https://www.mdpi.com/1099-4300/23/7/830) These cooperation protocols have neural correlates - localisable features encoded in topology of information networks. [Toy Models of Superposition](https://transformer-circuits.pub/2022/toy_model/index.html) You can transfer those protocols as these features geometries between the human brain and the transformer and hardcode it in the architecture, or as priors, or as first training data, or as how training gets shaped evolutionary. [Activation vector steering with BCI | Manifund](https://manifund.org/projects/activation-vector-steering-with-bci) [The Evolution and the Neural Correlates of Cooperative Behavior--Center for Excellence in Brain Science and Intelligence Technology](http://english.cebsit.cas.cn/lab/wangzuoren/news/202111/t20211117_292424.html) Alternatively you can align existing models by RLHFing by prompting by asking the LLM to have the context of all these linguistic cooperation protocols. https://openai.com/research/learning-from-human-preferences [x.com](https://twitter.com/gabriel_ilharco/status/1603415656699162624?t=Cusii4PCPRVtiVR1210mIA&s=19) Cooperation neurotech? [The Evolution and the Neural Correlates of Cooperative Behavior--Center for Excellence in Brain Science and Intelligence Technology](http://english.cebsit.cas.cn/lab/wangzuoren/news/202111/t20211117_292424.html) Natural abstraction hypothesis Simulators LW Shard theory My motivation is wanting florishing of sacredness of conscious physics from not only 5meodmt trips Gibbard-Satterthwaite theorem Hyllands theorem Strategyfreeness (lying will get better local results) is almost impossible Social choice theory Voting theory Auction theory Hodge theory Von Neumann game theory I had an irl discussion with @niplav_site about his work based on this paper on incosistency resolution in a graph of preferences, resolving ontological paradoxes, AI alignment! Statistical ranking with combinatorial Hodge theory https://arxiv.org/abs/0811.1067 You can use Helmholtz decomposition and Hodge theory on a graph under Von Neumann's game theoretic axioms of our preferences aka utility functions to partition an inconstent graph (cats are better than dogs, dogs are better than unicorns, unicorns are better than cats) into a minimal cyclic graph and potential graph (without inconsistency) and discord the cyclic graph to achieve consistency. Achieving consistency after shattering ontological shifts! Maybe could be used to make LLMs have consistent preferences with enough interpretability research! I feel like most of human value systems are incosistent so maybe its about finding the most common graph of incosistent human preferences and encode it to LLMs. Noselflessness is more correlation between self agent and others' people subagents' in the social graph If consciousness is loopiness, then intuition for panpsychism, feeling aliveness in everything, is mental representations of all things having selfreferential loopy processing You can dissolve by having all safety put into in deepest just existence prior, seeing all local structures as not needed for safety, looking for the sense of looker and center, looking for sense of spacetime and deconstructing it, how everything is made of nothingness, neither existence nor not existence? My fav approach to consciousness is topological, either topology of the electromagnetic field or information flow geometries, with various other theories of consciousness analyzing concrete different properties of architectures, processes, or other shapes inside this framework Some useful psychotechnologies https://cdn.discordapp.com/attachments/991876598724833363/1155858156476432385/IMG_20230925_152448.jpg You can trace the core beliefs of your surface beliefs in a mental graph to locate what they're build on and use this to align your actions and thoughts with your goals and values by using the power of the intuitive mind Best political system? https://slatestarcodex.com/2013/05/15/index-posts-on-raikoth/ [MicrasWiki](https://micras.org/mwiki/Main_Page) AI regulations just test for "safety" of prompts with advetasial training and RLHF instead of deeper foundational alignment AI labs instead of just researching and just aligninging end up in capture of power by big corporations and end just releasing possibly unaligned models (AnthropicAI by Amazon) Redwood research AI alignment lab Arc Center for compatible AI Apollo research Davidad [Yes, You Can Eliminate Bureaucracy](https://hbr.org/2018/11/the-end-of-bureaucracy) decentralization [Amazon to invest up to $4 billion into OpenAI rival Anthropic - The Verge](https://www.theverge.com/2023/9/25/23888841/amazon-4-billion-investment-anthropic-claude-ai-openai-microsoft) AI Tristan Harris Misinformation In crisis you can push benevolent regulations to prevent risks, but malevolent agents will try to do the same and they probably have evolutionary advantage https://www.lesswrong.com/posts/WCevxhGtmnPhWH3ah/is-ai-safety-dropping-the-ball-on-privacy-1 Cambridge analytica is overstating its claims, AI safety people should not have public information to prevent AIs decieving them by finetuned targeting? Main structured linear text Alternative structured linear texts Is big part of alignment symmetrification of structure and information flow in the social graph? https://www.lesswrong.com/tag/crockers-rules maximizing information in communication by minimizing emotions in debates Consciousness models have to be mathematical and physical [OSF](https://psyarxiv.com/r5t2n) You can get easy out of body experience by offsetting constructed sense of location of center of experience via VR, adapting to it, taking VR headset off and still being in the old offset position Lots of neurons in autism papers Are shaky? Localized brain regions for emotion and depression And illnesses dont replicate? Nate Hagens presentation Future [Nate Hagens l The Superorganism and the future‍ l Stockholm Impact/Week 2023 - YouTube](https://youtu.be/mN87PWfj7LA?si=fkUDz-bSAf37GsFg) Visualizing philosophy Wiki [Visualizing SEP: An Interactive Visualization and Search Engine for the Stanford Encyclopedia of Philosophy](https://www.visualizingsep.com/#) [Home](https://www.sentienceai.org/) [The Theoretical Minimum - Wikipedia](https://en.wikipedia.org/wiki/The_Theoretical_Minimum?wprov=sfla1) book [Visual: Visualizing Complex Systems Science — New England Complex Systems Institute](https://necsi.edu/visualizing-complex-systems-science) [Neural coding - Wikipedia](https://en.wikipedia.org/wiki/Neural_coding?wprov=sfla1) Is epistemic collapse the most probable current biggest risk? [List of unsolved problems in physics - Wikipedia](https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_physics?wprov=sfla1) [Radware Bot Manager Captcha](https://iopscience.iop.org/article/10.1088/1741-2552/ac6ca7/meta) Emergence of associative learning in a free energy minimizing neuromorphic inference network Active Inference memory [List of psychotherapies - Wikipedia](https://en.wikipedia.org/wiki/List_of_psychotherapies) [The Qualia Structure paradigm & the Quantum Qualia hypothesis @ CatTheory Workshop 2023 Apr16 Oxford - YouTube](https://youtu.be/rZsdqD29E60?si=VD19WHw3yMbFXLnb) category theory qualia Omni/Acc = Steelman all accelerationism movements and synthetize them on a higher order of complexity as compatibely as possible for collective survival, wellbeing and flourishing of all beings Join! Microsoft + OpenAI Google + DeepMind Tesla + xAI Meta + Meta AI and now Amazon + Anthropic biggest shapers of our future? I want to respect as much value and goal systems as possible while minimizing their friction in their evolutionary ecosystem until it doesn't interfere too much with my individual and collective longtermist wellbeing enhancing values Let's accelerate truth, wellbeing, resilience, sustainability, progress List of free energy principle papers [AII ~ Knowledge Engineering Frontend](https://coda.io/@active-inference-institute/knowledge-engineering) [The Free Energy Principle & Active Inference: a Systematic Literature Analysis](https://zenodo.org/record/7449368) List of neuroscience neural manifolds papers [x.com](https://twitter.com/ari_dibe/status/1706603453990994156?t=3tJNrqzub_xPgvxmlK7xMw&s=19) [What will GPT-2030 look like? — AI Alignment Forum](https://www.alignmentforum.org/posts/WZXqNYbJhtidjRXSi/what-will-gpt-2030-look-like) Is epistemic collapse the most probable current biggest risk? In order to fully model the universe, we would have to be as computationally powerful as the universe as individual agents, otherwise there is always some degree of compression. The fact that we only approximate the dynamics of the world, and lots of our models tend to be in big part noise instead of signal, is fundamentally fueling Moloch. Petrubation theory approximates anslytical physics allowing easier mathematics and computability All models of consciousness are right, they're just studying the same physical system from different angles using different tools with different degrees of predictive power of empirical data in their niches, let's synthetize them and express them in the language of physics! If music is group theory, I want a song corresponding to the monster group. Pymdp Idealism is compatible with laws of physics, just hard problem of consciousness stops existing Dark night of the soul as grief that you're not what evolution wants you to think you are [x.com](https://twitter.com/nickcammarata/status/1704308108686762380?t=Jt_namfwr6s8Gr5wyLV9EA&s=19) Exponential tech accelerates molochian dynamics https://www.lesswrong.com/posts/LpM3EAakwYdS6aRKf/what-multipolar-failure-looks-like-and-robust-agent-agnostic Wellbeing is about doing better than expected at reducing uncertainity, resolving error slopes, of basic needs homeostats in interconnected bayesian graph, locally and globally, including higher order goals connected to the basic needs or goals maximizing arbitrary objective functions, https://journals.sagepub.com/doi/full/10.1177/17540739211063851 and about sense of freedom, in the ultimate case when the whole topological graph with needs, self, social subgraph, and other belief priors all dissolve into a more unified global underfitted dynamics of information flow [The Symmetry Theory of Valence 2020 Overview](https://qri.org/blog/symmetry-theory-of-valence-2020) We can statistically define emergent functional behavior of language models, such as metacognition, by running actions where metacognition is used vs where not, scan what lights up in the inference network, and compare the two sets of measurements using all sorts of mathematical machinery [Reddit - Dive into anything](https://reddit.com/r/okbuddyphd/s/1Pmbl2Pk4P) reductionism ti physics What i value is people that understand that their reference frame isnt objective and update their beliefs according to evidence (in bayesian manner) instead of resisting being wrong Easiest meditation instruction is "use the least amount of effort in paying attention to the breath" Game theory Graph theory Network analysis [Casper Hesp | Max Planck Institute for Human Cognitive and Brain Sciences](https://www.cbs.mpg.de/cbs-coconut/video/hesp) Southwell Milewski [Computational creativity - Wikipedia](https://en.wikipedia.org/wiki/Computational_creativity) [metacrisis.xyz](https://metacrisis.xyz/) https://media.discordapp.net/attachments/991777510641651733/1156588811862679552/image.png?ex=651584c5&is=65143345&hm=20672b6728b21d111e5b57943921505be05b857c7ced0ffa80cacdf755fccfdb&=&width=1185&height=605 [Xenobot - Wikipedia](https://en.wikipedia.org/wiki/Xenobot?wprov=sfla1) update aethetics I wonder if there is literal biological neural correlate difference between people that believe in illusionism vs on the contrary, consciousness/qualia isnt/is real/illusion. Maybe something like less conceptualizing linguistic connections to lowest raw sensory data layers in the cortical hiearchy? Science -> Social systems -> Concrete phenomena -> Current state of our world and its future -> Chaotic systems, replication crisis, predicting the future also forms the future The more overall symmetrical an experience is neurophenomenologically, the more free, the more relieving, the more pleasant it is, the more psychological valence it has. How is boundary of conscious physical systems determined mathematically? [Binding problem - Wikipedia](https://en.wikipedia.org/wiki/Binding_problem) [[The Closest We Have to a Theory of Everything - YouTube](https://www.youtube.com/watch?v=A0da8TEeaeE) The Closest We Have to a Theory of Everything by Sabine Hossenfelder on Youtube] https://media.discordapp.net/attachments/992213422274003066/1116711387394228224/20230609_145134.jpg?width=800&height=600 Solomonoff's theory of inductive inference is a mathematical proof that if a universe is generated by an algorithm, then observations of that universe, encoded as a dataset, are best predicted by the smallest executable archive of that dataset. Formalization of Occam's razor. [Solomonoff's theory of inductive inference - Wikipedia](https://en.wikipedia.org/wiki/Solomonoff's_theory_of_inductive_inference) You wouldnt exist without all evolutionary physical forces that nourishes your biological machine in a thermodynamic nonequilibrium pullback attractor metastable state resistant to most petrubations. All language and modelling is fundamentally relational to others and universe. https://imgur.com/MxYRcHP special cases of moloch: academic citations, overfishing, AI arms races, nuke arms races, technology arms races without testing safety [multipolar trap · Daniel Schmachtenberger's knowledgegraph · Stephen Reid](https://stephenreid.net/k/daniel/terms/multipolar%20trap) https://conversational-leadership.net/multipolar-trap/ [Turing completeness - Wikipedia](https://en.wikipedia.org/wiki/Turing_completeness) Godel's incompleteness theorems, cantor's theorem, russel's paradox, halting problem, Y combinator, liar paradox 3Blue1Brown More general uncertainity principle 3blue1brown [The more general uncertainty principle, regarding Fourier transforms - YouTube](https://www.youtube.com/watch?v=MBnnXbOM5S4) Put FEP everywhere: FEP is mathematical "truth", epistemological framework, scalefree classical or quantum physics information theoretic ontology, modelling language, generator of process theories for any physical systems across scales (mainly brain and society), machine learning architecture, psychology model, what else? Different fields speaking different languages require different types of explanations, such as for threorists, engineers, information theory, classical physics, quantum physics, machine learning people, philosophers, mathematicians, neuroscientists, biologists, complex systemists, psychologists, sociologists, laymen. [NCCR SwissMAP - Field Theory for mathematicians (1/2) - YouTube](https://www.youtube.com/watch?v=DHiqHM5Ur_I&list=PLR6S6vLGnUbwkVbtq1LhpYRlW8G8czdU-&) "mathematicians studying (quantum) field theory is like blind men touching the elephant as it can mean many things" [faq - Resource recommendations - Physics Stack Exchange](https://physics.stackexchange.com/questions/12175/resource-recommendations) [How to become a GOOD Theoretical Physicist](https://www.goodtheorist.science/) physics resource physics resource [Classical Physics Lecture - 092723 | PIRSA](https://pirsa.org/23090036) physics resource [Search | MIT OpenCourseWare | Free Online Course Materials](https://ocw.mit.edu/search/?q=physics) [Classical Mechanics | Lecture 1 - YouTube](https://www.youtube.com/watch?v=ApUFtLCrU90) What is the general grand framework in which all the various specific physical laws are framed in, what are the rules for the allowable laws? Simplest laws of motion for the simplest system possible (coin flipping each time update) https://imgur.com/DfFWbDB [Classical Mechanics | Lecture 1 - YouTube](https://www.youtube.com/watch?v=ApUFtLCrU90) The Deep End of Meditation - Rebooting the Brain | Delson Armstrong [YouTube](https://www.youtube.com/watch?si=wOxhTMj_HRCfzjW5) FINALLY! A Good Visualization of Higher Dimensions [FINALLY! A Good Visualization of Higher Dimensions - YouTube](https://www.youtube.com/watch?v=sZqGWy0hxe8) utok culture https://medium.com/unified-theory-of-knowledge/now-utoking-culture-e24e43274adb evolution - survival of the fittest, of the stablest, of the most resilient patterns in space and time [Entropy | Free Full-Text | A Variational Synthesis of Evolutionary and Developmental Dynamics](https://www.mdpi.com/1099-4300/25/7/964)