AI automating all industries https://twitter.com/AISafetyMemes/status/1722977029778243833 how to learn the most effectively: depends on the field, depends on the part of the learning process gpt can ive intutive xplanations gpt can be a tutor, follow though equations gpt is more exact big body of strict knowledge for some you need concrete books or math/physcourses depends if you want mathematical or intitive understanding and of what so many variables IMO many tools have their advantages and disadvantages, from gpt prompt engineering to reading wiki/books to watching lectures to various courses online or offline its better to combine the various tools for more efficiency Bodhisattva of Playfulness! Curious playful exploratory enthusiastic hype childlike wonder acceleration! https://media.discordapp.net/attachments/991860230252134420/1167866715665870898/file-3Wj1rwUQ8fYkM2FVz7oJJmdz.jpg?ex=6558ea22&is=65467522&hm=e5ce28ebe04c5b16b841c4dd79ecbf5dd6e05a86da0d315619b0b69d90792c1b&=&width=597&height=597 Introducing Grimoire A GPT Coding Wizard 100x Engineer https://twitter.com/NickADobos/status/1722525397429174447 link parts from Joscha talks to different parts of the book the disintegrating mental health of people trying to solve quantum gravity https://media.discordapp.net/attachments/465999263059673088/1172585135875559495/f51v3ynxr7pa1.png [Episode 45: Leonard Susskind on Quantum Information, Quantum Gravity, and Holography - YouTube](https://www.youtube.com/watch?v=XTSdPSOcdjI) Sean Carroll Episode 45: Leonard Susskind on Quantum Information, Quantum Gravit Fullstack engineer conquering frontend and backend? Omnistack generalist engineer conquers all layers of reality, from atoms to hardware to software to ideas to changing the fabric of society or the whole universe [[2112.10510] Transformers Can Do Bayesian Inference](https://arxiv.org/abs/2112.10510) Transformers Can Do Bayesian Inference Truly having accelerationism + sustainability in terms of CO2 ppm requires abundant next gen fission and fusion. The way out is not by going dark, but by investing hundreds of billions more in global fusion projects like ITER, and gen 5 fission led by China. What is convolution [But what is a convolution? - YouTube](https://www.youtube.com/watch?v=KuXjwB4LzSA&t=693s) escaping “don’t exercise”->”low energy”->”don’t exercise” loop “don’t exercise”->”low energy”->”don’t exercise” loop https://twitter.com/uberstuber/status/1723051357609894161 criticality brain hypothesis Dopamine is not about the pursuit of happiness, it is about the happiness of pursuit. 'Does money predict happiness? Science 20 years ago: 'Well, sort of, but only up to about $80k/year; beyond that, it doesn't' Science now: 'Yep. More is better, all the way up, it scales logarithmically. But the effect is pretty small, less than half sigma on the whole scale.' https://twitter.com/primalpoly/status/1722727001981637003 “Over the next 10 years, we are aiming to build a fully autonomous AI Scientist, and to use that AI Scientist to scale up biology research.” [FutureHouse](https://www.futurehouse.org/articles/announcing-future-house) [The brain may learn about the world the same way some computational models do | MIT News | Massachusetts Institute of Technology](https://news.mit.edu/2023/brain-self-supervised-computational-models-1030) Two studies find “self-supervised” models, which learn about their environment from unlabeled data, can show activity patterns similar to those of the mammalian brain. multimodal robots [RoboVQA: Multimodal Long-Horizon Reasoning for Robotics](https://robovqa.github.io/) power of snythetic data for training [[2311.01455] RoboGen: Towards Unleashing Infinite Data for Automated Robot Learning via Generative Simulation](https://arxiv.org/abs/2311.01455) [[2310.17796] ControlLLM: Augment Language Models with Tools by Searching on Graphs](https://arxiv.org/abs/2310.17796) ControlLLM: Augment Language Models with Tools by Searching on Graphs LLM binding problem [[2310.17191] How do Language Models Bind Entities in Context?](https://arxiv.org/abs//2310.17191) https://www.facebook.com/xixidu Thermodynamic artificial intelligence computer [Normal Computing](https://blog.normalcomputing.ai/posts/2023-11-09-thermodynamic-inversion/thermo-inversion.html) https://twitter.com/ColesThermoAI/status/1625368583789334528 [Quantum mechanics - RationalWiki](https://rationalwiki.org/wiki/Quantum_mechanics) [RationalWiki](https://rationalwiki.org/wiki/Main_Page) pages from [RationalWiki](https://rationalwiki.org) quantum fluctations being the source of free energy is interesting train of thought but sadly doesnt work yea [Free energy (pseudoscience) - RationalWiki](https://rationalwiki.org/wiki/Free_energy_(pseudoscience)) [Zero-point energy - RationalWiki](https://rationalwiki.org/wiki/Zero-point_energy) "There have been speculations that usable energy might be extracted using this energy as a source using something called the Casimir effect, but this is almost certainly pseudoscience, as it would require using an extremely large collector device similar in some ways to an electronic capacitor, but much larger and thinner, with a vacuum dielectric; there is no knowledge currently extant that can allow the creation of such a collector cell, and even if it was possible, the zero point energy even in a volume the size of the Earth is so small as to be fairly useless." [/sci/ Wiki | Fandom](https://4chan-science.fandom.com/wiki//sci/_Wiki) kurzgesagt vids nootropics thread https://boards.4channel.org/sci/thread/15850475#q15850475 RationalAnimations uploading a mind [How to Upload a Mind (In Three Not-So-Easy Steps) - YouTube](https://www.youtube.com/watch?v=LwBVR68z-fg) if we can abstract most of functional behavior of a human brain into ANNs, we could run a whole brain emulation in about 10 years on the biggest supercomputer if exponential growth of computing power continues Mind upload [Reinforcement learning - Wikipedia](https://en.wikipedia.org/wiki/Reinforcement_learning) [Deep learning - Wikipedia](https://en.wikipedia.org/wiki/Deep_learning) [Psi-theory - Wikipedia](https://en.wikipedia.org/wiki/Psi-theory) [Conceptual blending - Wikipedia](https://en.wikipedia.org/wiki/Conceptual_blending) [Hopfield network - Wikipedia](https://en.wikipedia.org/wiki/Hopfield_network) [Spreading activation - Wikipedia](https://en.wikipedia.org/wiki/Spreading_activation) [Natural language generation - Wikipedia](https://en.wikipedia.org/wiki/Natural_language_generation) [Conceptual graph - Wikipedia](https://en.wikipedia.org/wiki/Conceptual_graph) [Meaning–text theory - Wikipedia](https://en.wikipedia.org/wiki/Meaning–text_theory) [Thematic relation - Wikipedia](https://en.wikipedia.org/wiki/Thematic_relation) [Argument (linguistics) - Wikipedia](https://en.wikipedia.org/wiki/Argument_(linguistics)) [Dependency grammar - Wikipedia](https://en.wikipedia.org/wiki/Dependency_grammar) [Boolean satisfiability problem - Wikipedia](https://en.wikipedia.org/wiki/Boolean_satisfiability_problem) [DPLL algorithm - Wikipedia](https://en.wikipedia.org/wiki/DPLL_algorithm) [Subgraph isomorphism problem - Wikipedia](https://en.wikipedia.org/wiki/Subgraph_isomorphism_problem) [Satisfiability - Wikipedia](https://en.wikipedia.org/wiki/Satisfiability) [Pattern matching - Wikipedia](https://en.wikipedia.org/wiki/Pattern_matching) [SPARQL - Wikipedia](https://en.wikipedia.org/wiki/SPARQL) [Genetic programming - Wikipedia](https://en.wikipedia.org/wiki/Genetic_programming) [Backward chaining - Wikipedia](https://en.wikipedia.org/wiki/Backward_chaining) [Forward chaining - Wikipedia](https://en.wikipedia.org/wiki/Forward_chaining) [Inference - Wikipedia](https://en.wikipedia.org/wiki/Inference) [Theory (mathematical logic) - Wikipedia](https://en.wikipedia.org/wiki/Theory_(mathematical_logic)) [Hidden Markov model - Wikipedia](https://en.wikipedia.org/wiki/Hidden_Markov_model) [Markov logic network - Wikipedia](https://en.wikipedia.org/wiki/Markov_logic_network) [Bayesian network - Wikipedia](https://en.wikipedia.org/wiki/Bayesian_network) [Cyc - Wikipedia](https://en.wikipedia.org/wiki/Cyc) [Datalog - Wikipedia](https://en.wikipedia.org/wiki/Datalog) [Type theory - Wikipedia](https://en.wikipedia.org/wiki/Type_theory) [Graph database - Wikipedia](https://en.wikipedia.org/wiki/Graph_database) [Relphormer: Relational Graph Transformer for Knowledge Graph Representations | Papers With Code](https://paperswithcode.com/paper/relphormer-relational-graph-transformer-for) Human-like systematic generalization through a meta-learning neural network [Human-like systematic generalization through a meta-learning neural network | Nature](https://www.nature.com/articles/s41586-023-06668-3) NNs optimizing for compositionality achieving both systematicity and flexibility [Mechanistic Interpretability — Neel Nanda](https://www.neelnanda.io/mechanistic-interpretability) https://medium.com/aiguys/decoding-ai-blackbox-mechanistic-interpretability-b64de4cbac91 https://www.lesswrong.com/tag/interpretability-ml-and-ai Physiological sigh Cold showers Book Unstructured notes Redundant Outdated [[1210.8447] Nothing happens in the Universe of the Everett Interpretation](https://arxiv.org/abs/1210.8447) [One-electron universe - Wikipedia](https://en.wikipedia.org/wiki/One-electron_universe) I'm curious why EM models of consciousness tend to not talk about other forces in the standard model. I suppose because they're much less common. What if they're also part of the consciousness dynamics equation as electromagnetism is everywhere in biological systems, but something still has to hold atoms together (strong force), certain radioactive isotopes are present in biological systems and decay (weak force), gravity is around every mass, and forces are interactions of matter so matter is needed in the equation as well. [Electromagnetic theories of consciousness - Wikipedia](https://en.wikipedia.org/wiki/Electromagnetic_theories_of_consciousness) [Why Relativity Breaks the Schrodinger Equation - YouTube](https://youtu.be/8Iu74b5iCuQ?si=PCW8DhGBJEeBS3O6) Why Relativity Breaks the Schrodinger Equation gravitons are interesting idea, i wish it wouldnt impossible with any physically reasonable detector, at least we can measure gravitational waves that can give us some properties about them [Graviton - Wikipedia](https://en.m.wikipedia.org/wiki/Graviton) There might be ways to test quantum gravity [The five most promising ways to quantize gravity - YouTube](https://youtu.be/5foUTeRdqII?si=v3mHnY_vbSaBMCsY) [How To Test Quantum Gravity - YouTube](https://youtu.be/HKtFU2Pxnm0?si=XpAhqz36VjDMhcL5) Selfimproving AI https://twitter.com/jkronand/status/1723155400571371988?t=mPCmv8lH14dtY4awmcjUbg&s=19 We can think of brain on hardware and software level. There is conscious and subconscious software that can be directly or indirectly programmed by ourselves by knowledge, metacognition, theraphy etc., or it can be programmed by our culture, environment, under the genetic constrains of hardware. Most of variance in IQ is explained by hardware (Twins separated at birth into different families study) https://imgur.com/k9uG9Cq Brain and its functions in one image AI for survelence [Nick Bostrom: How AI will lead to tyranny - YouTube](https://www.youtube.com/watch?v=_Oo-m893-xA) most of ethics people have is fuzzy wishy washy common good instead of utilitarian maximize exactly defined utility (whatever that is defined as) rlhf generalized to bigger systems smarter than us might not work anymore, is it afterwards learning to do what we want or what triggers humans for a thumbs up? intelligence inequality inequalities: social, economic, technological, intelligence May AGSI benefit all sentient beings Work on good causes, donate to good causes, meet people working on good causes not just offline but also online outside your bubble and country! Understanding the universe with science and AI, increase the wellbeing of all sentient beings! Ai can develop bioweapons more easily which was handled on Ai summit should we pay moral attention to digital minds moral weight to animals moral weight to all sentient matter living with potentially sentient very similar to us but diversely different or alien to us intelligent machines wont be shocking scifi fantasy very soon david pearce's moral vision https://fxtwitter.com/RuohanZhang76/status/1720525182732005683 https://fxtwitter.com/RuohanZhang76/status/1720525179028406492 NOIR is a general-purpose, intelligent BRI system that enables humans to command robots to perform 20 challenging everyday activities using their brain signals, such as cooking, cleaning, playing games with friends, and petting a (robot) dog. [The World Is Running Out of Data to Feed AI, Experts Warn : ScienceAlert](https://www.sciencealert.com/the-world-is-running-out-of-data-to-feed-ai-experts-warn) synthetic data to the rescue, and next optimizations in LLMs will probably be to do more with less data, getting closer in performance to how the human brain works Developing AGSI poses lots of harmful risks and potential failure modes for our system that may create suffering for all of sentience before or after AGSI is reached (opression, wars, extinction...), but if we get it right, there is a big potential for future where all of sentience flourishes with wellness. Risks will always exist, its about minimizing them in the aggregate. If we dont develop AGSI soon, there might be higher probability of getting killed by other factors, like nukes, biorisk, nanotech risk, climate change, natural causes, or getting into opressive dystopia or chaotic breakdown full of tribal wars, with radical social, economic, technological or intelligence inequality, which itself might even be accelerated by AI, but all tools are dual use aka omni use, it might also be the other way around if we direct it correctly into a future where all sentient beings, maybe even beyond humans, flourish in wellness! [Nick Bostrom: How AI will lead to tyranny - YouTube](https://www.youtube.com/watch?v=_Oo-m893-xA) Shuman https://cdn.discordapp.com/attachments/465999263059673088/1172988077308911707/dergfh4-7db1abd2-6fd4-443e-af79-07ad8075d025.png?ex=65625146&is=654fdc46&hm=99560f02fa2990ee4f646d77bd4b6fd7b44e9b05dbb7cd23ee770f98aa6065e6& Lucid dreaming vs Free wheeling hallucinations: [Free-Wheeling Hallucinations | Qualia Computing](https://qualiacomputing.com/2019/01/06/free-wheeling-hallucinations/) much more intensity, dimensions, psychedelic visuals/thoughts, more easily breakable and easier loss of control, bias towards better valence (can be bad valence if bad trip), bias towards less center, selfreferentiality,... [Active Dreaming is More Powerful than Meditation and Safer than Psychedelics](https://superbowl.substack.com/p/dreaming-is-better-than-meditation) List advantages (benefits), disadvantages (risks), benefity of technology and AI in general https://cdn.discordapp.com/attachments/991777510641651733/1173065827398983761/6zxcy41gq9wz.jpg?ex=656299af&is=655024af&hm=2b387310eef93faab36b78f28be8f3167a5688f03bca92177a1ca1742315d64b& In history we Went from tribal wars to kingdoms and many more, you can Google all entities mentioned here, all can be seen as different sociocultural economic technological interacting Markov blankets, and the current state is; list current (USA more capitalist libright, EU more regulátory China more totalatarian auth) and potential future humanity system equilibriums Across sociál, economic, technological laser, from individualistic to collectivist [ChatGPT](https://chat.openai.com/share/008d6501-5eab-4655-964d-a9e8dea7058d) Regulations and agreements on climate change, nukes, AI UN AI summit I "hold" the view that consciousness is both being real and an illusion and neither of these. All models are wrong but some are more useful for different goals than others. Both positions have different utility in different contexts. Easiest than ever to build apps with upgraded multimodal gpt api https://twitter.com/sairahul1/status/1723260273682006282?t=uKpn4MAJAobr8WOj6qYt-w&s=19 https://twitter.com/shushant_l/status/1723234162621255821?t=WX2OlIVZAUpVbrEAsJF-Ww&s=19 ML tools and models https://machinelearningmastery.com/a-tour-of-machine-learning-algorithms/ [Outline of machine learning - Wikipedia](https://en.m.wikipedia.org/wiki/Outline_of_machine_learning#Machine_learning_algorithms) [Category:Machine learning algorithms - Wikipedia](https://en.m.wikipedia.org/wiki/Category:Machine_learning_algorithms) [ChatGPT](https://chat.openai.com/share/07376f56-b187-4185-b2c4-40a28a73a116) [A comprehensive list of machine learning algorithms - Artificial Intelligence Stack Exchange](https://ai.stackexchange.com/questions/38093/a-comprehensive-list-of-machine-learning-algorithms) Lets convert the whole universe into wellbeingium (matter experiencing only wellbeing, meaning, selfactualization, truth, bliss, pleasure, peace, equaminity, safety, jhanas, positive valence) and not sufferium We should not overshoot fear from AGI, but we should be fully aware of its risks This guided meditation is lovely, the mind feels like calm paradise now after everything that wanted to express expressed itself 💞 May you fullfill your dreams! [The Simplest Thing - YouTube](https://www.youtube.com/live/_HG02hDl5R0) Turing machine can simulate directly or virtually arbitrary physical process given enough time and memory If you would want to simulate the whole universe perfectly, the computer would have to be as big as our universe otherwise there are time and memory constrains Can we comprehend all of physics, is it analyzable, is it computable? Even if we find undecidable functions in physics, like the spectral gap problem, we can still study the function's properties and therefore properties of the physical world. [Haag's theorem - Wikipedia](https://en.wikipedia.org/wiki/Haag%27s_theorem) "Haag's theorem shows that the interaction picture is always inconsistent, everyone must agree that as a piece of mathematics Haag’s theorem is a valid result that at least appears to call into question the mathematical foundation of interacting quantum field theory, and agree that at the same time the theory has proved astonishingly successful in application to experimental results." [quantum field theory in nLab](https://ncatlab.org/nlab/show/quantum+field+theory) [geometry of physics in nLab](https://ncatlab.org/nlab/show/geometry+of+physics) Biden and Xi ban AI in autonomous weapons https://twitter.com/AkashWasil/status/1723688001811706199 "The reason for the depression at all is because life is actually beautiful and sacred. And the depression is recognizing the destruction of the sacred." "You can't let yourself not act because you don't know how, but you also can't act doing things that you know won't work, so you have to learn." https://twitter.com/meta_crisis/status/1708988528175018453 "I was just following incentives" is the new "I was just following the orders" People having vested interest in AI regulating AI might not be the best idea https://twitter.com/meta_crisis/status/1708519228033106229 Act from service to the beautifulness and sacredness of life, not from anger, depression and fear https://twitter.com/meta_crisis/status/1708988528175018453 The peaceful period was period of proxy wars and nonphysical wars The embedded growth obligation [of the financial system] is totally incompatible with planetary boundaries. So do you have to change financial services to not have an embedded growth obligation? Absolutely." https://twitter.com/meta_crisis/status/1708526256541581612 we should make international nudist day, where we go on about our day as civilization but without any clothes, everyone is allowed to do that, all citizents, all politicians, everyone LLMs making bioweapons might be so far only just better than what one can do with a search engine, which might change with autonomous agents, but having a lab for it is still a big obstacle [[2304.05376] ChemCrow: Augmenting large-language models with chemistry tools](https://arxiv.org/abs/2304.05376) "Our agent autonomously planned and executed the syntheses of an insect repellent, three organocatalysts, and guided the discovery of a novel chromophore." Philosophers are people who sublimate their existential confusion by specializing on being confused in highly specialized ways about preliminaries history of technology fire wheel plow industrial revolution I just can't see it. You get corporate dystopia limiting your (not only) economic freedom in anarchocapitalism as a result. Let's assume anarchocapitalism starts with no inequality at the beggining. Some people are more lucky or better at capitalist games than others. Power laws distribution of wealth emerges and therefore giants with giant unregulated power of influence over others forms. You're now living in the Amazon empire. Which creates different levels of environmental, economical, technological, cultural sustainability, transformation and risks to different valued entities ethical dilemmas in genetic engineering loss of privacy global citizenship echnology and nature coexisting balance between centralized control (e.g., authoritarian governments) and decentralized systems (e.g., democracies, blockchain technologies). Overcoming nationalism, resource competition, and differing values to achieve global cooperation. cross-cultural collaborations AI playing minecraft https://twitter.com/arankomatsuzaki/status/1723882043514470629 place all ideologies and movements and risks etc. my spectrum google multimodal SotA https://fxtwitter.com/arankomatsuzaki/status/1723886046155272232 https://fxtwitter.com/teortaxesTex/status/1723887651974009222 wellbeing (eq in book, humberman), nootropics reshaping of globalized economy from direct china russia usa trades to indirect [Globalization Is Fracturing. So What Comes Next? - YouTube](https://www.youtube.com/watch?v=ttaaMw8wAy0) logistic regression Naive bayes Global governance vs individual indeoendent decisionmaking Global coordination vs global everyone does stuff without coordinating on issues affecting everyone Pirates party FTX as example of sociopath invading altruistic circles Risks that people assign different degree of severity and probabilities of happening to What happens after death Quantum immortality Gradient descent [Gradient Descent in 3 minutes - YouTube](https://youtu.be/qg4PchTECck?si=B6-EYBSUwwX6WQqF) Merging with government Autonomous GPT swarm [The Swarm that Builds the Swarm - Trending on GitHub! I accidentally created the Borg Collective... - YouTube](https://youtu.be/D4FuVrrRi0A?si=cXN7h5OvqribbBvV) Centralized vs decentralized control in AGI Map of brain https://www.facebook.com/100009171158831/posts/3510094649306206/?mibextid=rS40aB7S9Ucbxw6v Probabilstic models Probabilstic graphical models Biggest unsolved problems in chemistry (ask ChatGPT) ending global poverty by 300 trilion dollars [How to Eradicate Global Extreme Poverty - YouTube](https://www.youtube.com/watch?v=2DUlYQTrsOs&list=WL&index=4&t=268s) adjunctions from category theory are formalized analogies