Reasoning skill is a spectrum for both biological and artificial systems [Validation of biomarkers of aging | Nature Medicine](https://www.nature.com/articles/s41591-023-02784-9) [Stable Diffusion with Brain Activity](https://sites.google.com/view/stablediffusion-with-brain/?pli=1) Automate the automatization of automatization [[2402.10790] In Search of Needles in a 11M Haystack: Recurrent Memory Finds What LLMs Miss](https://arxiv.org/abs/2402.10790) [Tutorial 16: Meta-Learning - Learning to Learn (Part 1) - YouTube](https://youtu.be/035rkmT8FfE?si=498erNTpjxoHWlKo) [9 New Gemini Leaks, Code Llama and A Major AI Consciousness Paper - YouTube](https://www.youtube.com/watch?v=ad5SZeANuJE&list=WL&index=1&t=762s&pp=gAQBiAQB) [Artificial Intelligence played Wargames. The result isn't reassuring. - YouTube](https://www.youtube.com/watch?v=mh_hNvGTmIg&list=WL&index=3&pp=gAQBiAQB) [Human languages with greater information density have higher communication speed but lower conversation breadth | Nature Human Behaviour](https://www.nature.com/articles/s41562-024-01815-w) Rabbithole so deep that you form a circle in spacetime https://twitter.com/fly51fly/status/1759520523334341108?t=0NyxRCplj0ds9mcRYnp5Gw&s=19 [[2402.10877] Robust agents learn causal world models](https://arxiv.org/abs/2402.10877) [The Shift from Models to Compound AI Systems – The Berkeley Artificial Intelligence Research Blog](https://bair.berkeley.edu/blog/2024/02/18/compound-ai-systems/) Did anyone made multiagent frameworks to write multiagent frameworks (like Autogen, AgentVerse or CrewAI) using all prompt engineering hacks and access to documentation (maybe RAG over downloaded, or Fleet Context, or search via Tavily or PerplexityAI) [GitHub - zjunlp/LLMAgentPapers: Must-read Papers on LLM Agents.](https://github.com/zjunlp/LLMAgentPapers) So many constructs that consciousness theorists see as fundamental meditators deconstructed into ieffable void! Self model, selfreflection, logic, language, distinctions, center of experience, space, time, boundaries, background awareness, percieving,... Are all just evolutionary useful mental constructs that can be deconstructed! We are accelerating towards global flourishing https://www.scientificamerican.com/article/brains-are-not-required-when-it-comes-to-thinking-and-solving-problems-simple-cells-can-do-it/ https://psycnet.apa.org/record/2022-77903-001?doi=1 The omni techbro cult: Unifying all techbro cults into one allencompassing megacult technooptimism is rising again "i" love installing 42069666 different software memeplexes installing different worldviews and observing them fight for dominance from the point of view of the biggest background most dominant memeplex that gets dissolved or transformed if overly destabilized by all incoming information with maximal openness Cortical activity spirals like eddies. This may stabilize the brain and keep neural computation on track. [Stability from subspace rotations and traveling waves | bioRxiv](https://www.biorxiv.org/content/10.1101/2024.02.19.581020v1) [Frontiers | Capture of fixation by rotational flow; a deterministic hypothesis regarding scaling and stochasticity in fixational eye movements](https://www.frontiersin.org/articles/10.3389/fnsys.2014.00029/full) my mind is a cage match where the most schizophrenic ontologies are forced to duel for their survival https://twitter.com/slimepriestess/status/1672366059570216961 categorical systems theory http://davidjaz.com/Papers/DynamicalBook.pdf [The Four Wars of the AI Stack (Dec 2023 Recap)](https://www.latent.space/p/dec-2023) https://twitter.com/swyx/status/1744467383090372743/ symbolic generalization [Automated discovery of algorithms from data | Nature Computational Science](https://www.nature.com/articles/s43588-024-00593-9) Eureqa Turingbot https://twitter.com/MikePFrank/status/1759979234246725820 Reversible computing [Power Causes Brain Damage - The Atlantic](https://www.theatlantic.com/magazine/archive/2017/07/power-causes-brain-damage/528711/) [Solving the Problem of Observers & ENTROPY | Stephen Wolfram - YouTube](https://www.youtube.com/watch?v=0YRlQQw0d-4&feature=youtu.be) If you want AI tools to give you good sources about their claims, don't use vanilla free ChatGPT. Force paid version that has search to search the web. Or use new Google's Gemini which is better which is trained on sources and has Google search functionality, PerplexityAI searchers the web much better than ChatGPT's Bing, and sources that way, and ConcensusAI searches and sources through 300 million studies (I love that one). Here are all the tools showcased asking for the neuroscience of wellbeing as an example, it's orders of magnitude better with giving links to concrete studies or articles depending on the tool and prompt, than vanilla free ChatGPT. They give links that lead to valid sources. Footnote, citation or green color means link to a source that's valid there. Or you can connect your any LLM to search APIs via GPTs, Langchain GUI editor, or Python code through APIs or locally (with or without Langchain or Llamaindex). https://twitter.com/burny_tech/status/1760268925592293431?t=cHP6fOTvmJ4W02CNOM1mPA&s=19 We're this crazy amalgamation of trillions of little cellular factories and proton motors, so don't get upset if you wake up one day and you just "feel off." It's a miracle you even work. https://twitter.com/burny_tech/status/1760271731661340695?t=1fkKvJanI11t4yJtbSt62w&s=19 It's not that exercise beats out SSRIs for depression treatment, but that *just* dancing has the largest effect of *any treatment* for depression. https://twitter.com/eshear/status/1760361698278744193 [[2402.12188] Structure of activity in multiregion recurrent neural networks](https://arxiv.org/abs/2402.12188) [[2311.06189] Syntax-semantics interface: an algebraic model](https://arxiv.org/abs/2311.06189) [[1912.05088] An Introduction to Complex Systems Science and its Applications](https://arxiv.org/abs/1912.05088) https://www.lesswrong.com/posts/gBHNw5Ymnqw8FiMjh/ai-51-altman-s-ambition Let's use Sora to create a better newest Matrix movie we all wanted [A Techno-Industrialist Manifesto](https://www.piratewires.com/p/techno-industrialist-manifesto) If you dig deep enough into the majority of people's minds about their fulfilment from life, you find out that there're deeply unsatisfied with what they have to do everyday in order to survive and feel disconnected from the whole system which makes them suffer deeply. As a result they numb, escape and distract themselves in all sorts of ways. This is not how the system should work. This has to and will change soon. I want to help technooptimist movements that try to technologically, politically, economically and culturally change the status quo dynamics using disruption and other methods to change this. Stuff like allowing biological systems to pursure meaningful fullfilling tasks they are evolutionary optimized for without having to worry about paying rent, or reprogramming the biological machine. Are LLMs conscious [[2402.12422] Simulacra as Conscious Exotica](https://arxiv.org/abs/2402.12422) https://twitter.com/mpshanahan/status/1760336183068950765?t=7XQie5tjVLnVC5jfcA3Dmw&s=19 [[2402.14020] Coercing LLMs to do and reveal (almost) anything](https://arxiv.org/abs/2402.14020) [Attention schema theory - Wikipedia](https://en.wikipedia.org/wiki/Attention_schema_theory) The attention schema theory (AST) of consciousness (or subjective awareness) is a neuroscientific and evolutionary theory of consciousness. It proposes that brains construct subjective awareness as a schematic model of the process of attention. Attention schema is like the body schema. Just like the brain constructs a simplified model of the body to help monitor and control movements of the body, so the brain constructs a simplified model of attention to help monitor and control attention. [ArXiv Machine Learning Landscape](https://lmcinnes.github.io/datamapplot_examples/ArXiv_data_map_example.html) https://fxtwitter.com/leland_mcinnes/status/1760470403632300476?t=O9Rojzf-aOSJh9rTsvpL1A&s=19 arxiv Machine learning landscape What worldview memeplex software did you pick from collective worldview wardrobe today [Synthetic Sentience: Can Artificial Intelligence become conscious? | Joscha Bach | CCC #37c3 - YouTube](https://youtu.be/Ms96Py8p8Jg?si=S79dwRVQdrDtMBVZ) https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4927579/ “Restricting your speech back propagates to restricting your thoughts.” [What if Singularities DO NOT Exist? - YouTube](https://www.youtube.com/watch?v=HRir6-9tsJs) It's not too often that a giant of physics threatens to overturn an idea held to be self-evident by generations of physicists. Well, that may be the fate of the famous Penrose Singularity Theorem if we're to believe a recent paper by Roy Kerr. Long story short, the terrible singularity at the heart of the black hole may be no more. [ALL OF PHYSICS explained in 14 Minutes - YouTube](https://www.youtube.com/watch?v=ZAqIoDhornk) My body is a machine that turns suffering into compassion Define defining of defining [Richard Sutton on Pursuing AGI Through Reinforcement Learning - YouTube](https://youtu.be/zZuh8YUBeDY?si=hV-m1oVvPX__FAqo) [[2402.14083] Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrapping](https://arxiv.org/abs/2402.14083) [Crocker's Rules](http://sl4.org/crocker.html) [The Singularity Is Near, The Movie - Homepage](https://singularity.com/themovie/) [[2402.14396] Quantum Circuit Optimization with AlphaTensor](https://arxiv.org/abs/2402.14396) Pokud to chceš z pohledu vědy, tak realita je taková, že zjišťujeme víc a víc jaký konkretní struktury v mozku určují tvojí gender identitu (pocit či se cítíš spíš jako holka nebo kluk) https://academic.oup.com/cercor/article/30/5/2755/5669983 což je dost propojený ale odlišný od biologickýho sexu. Pro některý lidi je cesta z utrpení z týhle disharmonie z mismatchu mezi genderovou identitou a biologickým sexem fyzická tranzice a tím alignout ten mismatched neuropsychologický pocit s tím jak vypadají fyzicky jako biologický sex. Brzo možná bude alternativní cesta z tohoto mentálního distressu upravit tuto mozkovou strukturu genderový identity přes neurotechnologie. I když někteří chtějí, tak "příjmout" genderovou identitu co sedí s biologickým sexem je fyzikálně v jejich mozku nemožný, tím jak je ta druhá silně hardcoded na úrovni softwaru a možná hardwaru, a mentální vliv na tuto strukturu s tím skoro nehne a nepřepíše to, ale u některých to možný být může. Tato část mozku určující jakou cítíš genderovou identitu je ovlivněná jak geneticky, tak psychologicky, tak psychiatricky, tak environmentálně, tak sociálně - a tam jde zkoumat a diskutovat, jaký všechny kauzální faktory vývoj tyto mozkový a ostatních napojených biologických struktur uvlivňuje. Obecněji je tohle součást identity, a otherkini si tam nainstalovali zvířata přímo nebo nepřímo přes ty různý vlivy. Já třěba osobně cítím, že modul genderový identity v mozku v podstatě nemám, nebo je extrémně slabej a hodně fluidní, spíš nejbližší identita mi je biologický stroj s hodně fluidní sociální identitou, kterej může být technicky libovolná forma s bioengineeringem. To právě dost záleží jak hodně hardcoded přes ty všechny možný vlivy ta mismatched genderová identita v tom mozku u individuálních lidí je. Pro některý to alignout psychologicky možná půjde, ale pro některý zase ne. Nevím jak moc se obě strany učí, ale tipuju, že sexuolog nejdřív prozkoumá jestli si člověk je fakt jistej. Myslím, že by se měly ukazovat a respektovat obě strany, je možný že se nějaká z nich víc zanedbává, nebyl jsem tam, ale mělo by to jít dogooglit. Pro některý může ta genderová identita jako software v mozku být tak hardcoded, že cokoliv psychologickýcho a psychiatrickýho s tebou prostě nehne a pořád cítíš ty error signály z mismatchu jako utrpení ať děláš cokoliv. Tranzice velký části biologickýho sexu, která se časem technologicky víc zlepšuje, pro tento typ lidí s takovým mozkem a takovým tělem potom není "ničení", ale "oprava", kde jim mozek potom tak negeneruje error signály z toho že vidí jiný tělo, než by mozek chtěl. [Connectome - Wikipedia](https://en.wikipedia.org/wiki/Connectome) [AI/ML+Physics Part 1: Choosing what to model [Physics Informed Machine Learning] - YouTube](https://www.youtube.com/watch?v=ARMk955pGbg) [OpenCodeInterpreter](https://opencodeinterpreter.github.io/) [[2402.14083] Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrapping](https://arxiv.org/abs/2402.14083) morphological freedom is imminent [Morphological Freedom | Anders Sandberg | Audiobook - YouTube](https://www.youtube.com/watch?v=bSGF1E2JDtE) What is universal in natural languages? To answer that, deep connections need to be made between universal grammar, written codes, statistical patterns and Universal Turing machines. [Universality and Complexity in Natural Languages: Mechanistic and Emergent[v1] | Preprints.org](https://www.preprints.org/manuscript/202402.1330/v1) How can we define neurological diseases as attractors? Can we model the damage to semantic networks in Alzheimer patients? How is a disease defined as a graph? [[2301.09205] The categorical basis of dynamical entropy](https://arxiv.org/abs/2301.09205) [[2402.14433] A Language Model's Guide Through Latent Space](https://arxiv.org/abs/2402.14433) [Why I think Synthetic Controls are BOGUS | Quantum Bayesian Networks](https://qbnets.wordpress.com/2022/08/17/why-i-think-synthetic-controls-are-bogus/) this quantum information theorist pioneer guy robert tucci has a lot of interesting thoughts on causality on his site and the comment section of andrew gelmans blog Explainable AI cheat sheet https://twitter.com/burny_tech/status/1761193275099103696?t=6ndiPbXKsdkuh5INvXGWuA&s=19 https://twitter.com/NPCollapse/status/1761376815925354775?t=-Hh6Ru8sg3E1BuwYuFc4qQ&s=19 The central node of GPT-J’s signifying network is… the phallus. Lacan remains unbeaten. Nobody tell Zizek about this. The central node of GPT-J's signifying network is… the phallus. [[2402.14735] How Transformers Learn Causal Structure with Gradient Descent](https://arxiv.org/abs/2402.14735) https://twitter.com/fly51fly/status/1761368979145441769 [AI solves nuclear fusion puzzle for near-limitless clean energy | The Independent](https://www.independent.co.uk/tech/nuclear-fusion-ai-clean-energy-b2500756.html) [Avoiding fusion plasma tearing instability with deep reinforcement learning | Nature](https://www.nature.com/articles/s41586-024-07024-9#Sec5) https://twitter.com/burny_tech/status/1761432901910986844 [FDA approves cure for sickle cell disease, the first treatment to use gene-editing tool CRISPR](https://www.nbcnews.com/health/health-news/fda-approves-cure-sickle-cell-disease-first-treatment-use-gene-editing-rcna127979) [FDA approves cure for sickle cell disease, the first treatment to use gene-editing tool CRISPR](https://www.nbcnews.com/health/health-news/fda-approves-cure-sickle-cell-disease-first-treatment-use-gene-editing-rcna127979) [Bloomberg - Are you a robot?](https://www.bloomberg.com/news/articles/2024-02-23/bezos-nvidia-join-openai-microsoft-in-funding-humanoid-robot-startup-figure-ai) https://www.pnas.org/doi/full/10.1073/pnas.2313925121 „The behaviors are generally indistinguishable, and ChatGPT-4 actually outperforms humans on average, while the reverse is true for ChatGPT-3. There are several games in which the AI behavior is picked more likely to be human most of the time, and others where it is not. When they do differ, the chatbots’ behaviors tend to be more cooperative and altruistic than the median human, including being more trusting, generous, and reciprocating.“ Evidence has been found that generative image models have representations of these scene characteristics: surface normals, depth, albedo, and shading https://twitter.com/anand_bhattad/status/1730230190159135175 [[2311.17137] Intrinsic LoRA: A Generalist Approach for Discovering Knowledge in Generative Models](https://arxiv.org/abs/2311.17137) https://longevity.technology/news/rejuvenate-bio-shows-epigenetic-reprogramming-extends-lifespan-in-normal-mice/ https://www.liebertpub.com/doi/10.1089/cell.2023.0072 Can't decide if wheels or legs are better? Just have both! Curiosity-Driven Learning of Joint Locomotion and Manipulation Tasks by Robotic Systems Lab: Legged Robotics at ETH Zürich [Curiosity-Driven Learning of Joint Locomotion and Manipulation Tasks - YouTube](https://www.youtube.com/watch?v=Qob2k_ldLuw) Working memory is organized by multi-dimensional neural representational geometry. It is not just the maintenance of sensory inputs. 2-D Neural Geometry Underpins Hierarchical Organization of Sequence in Human Working Memory Broken Neural Scaling Laws extrapolate everything https://twitter.com/ethanCaballero/status/1651278118936625170?t=6EWsTezg1licqFetRsuY-w&s=19 Search Dynamics Bootstrapping -Toward LLMs as better planners/reasoners... -Search dynamics expressed as a token sequence, then fine-tuned via expert iterations to perform fewer steps -Transformer model optimally solves previously unseen Sokoban puzzles [[2402.14083] Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrapping](https://arxiv.org/abs/2402.14083) https://twitter.com/venturetwins/status/1761443272596291920?t=OuC5puQ8A8BQyvKkjoBRdg&s=19 Fun analysis on which tips & threats make ChatGPT follow instructions. The author asked for a 200 character story, and measured output lengths under various negative and positive incentives. Turns out, ChatGPT cares most about losing friends and DEATH (when it’s in all caps). [Does Offering ChatGPT a Tip Cause it to Generate Better Text? An Analysis | Max Woolf's Blog](https://minimaxir.com/2024/02/chatgpt-tips-analysis/) [Paper page - Chain-of-Thought Reasoning Without Prompting](https://huggingface.co/papers/2402.10200?utm_source=digest-papers&utm_medium=email&utm_campaign=2024-02-16) https://www.linkedin.com/posts/andrew-iain-jardine_llms-activity-7165730371215589376-ltdU?utm_source=share&utm_medium=member_android [Paper page - MVDiffusion++: A Dense High-resolution Multi-view Diffusion Model for Single or Sparse-view 3D Object Reconstruction](https://huggingface.co/papers/2402.12712) [[2402.04845] AlphaFold Meets Flow Matching for Generating Protein Ensembles](https://arxiv.org/abs/2402.04845) [Genome-wide association meta-analysis in 269,867 individuals identifies new genetic and functional links to intelligence | Nature Genetics](https://www.nature.com/articles/s41588-018-0152-6) [Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence | Nature Genetics](https://www.nature.com/articles/ng.3869) [Genetics and intelligence differences: five special findings | Molecular Psychiatry](https://www.nature.com/articles/mp2014105) Genes of intelligence, i really wanna see more functional analysis of these, i think that might be path towards figuring out how intelligence works in humans more which could lead to a better replication in machines using both known with both genetic and environmental factors in humans Here’s a study with environmental intelligence increase https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7709590/ there are studies also where they found practicing visuospatial reasoning also led to huge intelligence increase Furthermore also the study points to where about 20% in childhood of intelligence is inherited and 80% in later adulthood. Giving plenty of room for environmental. So giving kids a good education and making sure they are engaged is more likely to lead to intelligence increases in society overall from quick search of studies this is probably the best one i could find agreeing with the conclusion that races dont really differ in iq, but there seems to be so little of these studies (not surprised though as its so controversial) [Race, IQ, and the search for statistical signals associated with so-called “X”-factors: environments, racism, and the “hereditarian hypothesis” | Biology & Philosophy](https://link.springer.com/article/10.1007/s10539-014-9428-0) i see one more in the reddit thread https://onlinelibrary.wiley.com/doi/full/10.1002/ajpa.24216 https://www.reddit.com/r/AskSocialScience/s/xwW57E0Nip [Athletics, IQ, Health: Three Myths of Race – SAPIENS](https://www.sapiens.org/biology/athletics-iq-health-myths-of-race/) https://academic.oup.com/genetics/article/161/1/269/6049925?login=true aka there is much more genetic variation for lots of other stuff other than just skin color and other smaller adaptations for populations living in different parts of the world in different conditions, makes sense iq and intelligence isnt same, as iq is one attempt at quantification of intelligence im daily stuck in discussions where people have completely different definitions of intelligence, mostly in the context of artificial intelligence, its tiring sometimes how there really isnt a concensus between all the different groups of researchers Agreeing with everything your ingroup is saying seems like a sign of not rationally evaluating the empirical truth of their propositions i love this site :D it aggregates news sources and groups them according to lots of sliders (including left and right scales) and allows you to set up which biases do you wanna see :D [Verity - Headlines](https://www.improvethenews.org/) either agi in two years or we get the worst tech crash since the dot com and telecom bubble be ready for both Tech crash in 2 years followed by AGI slowly emerging over the next 20 years. Either way, people who are young now will have fun life even if it takes 2 or 20 years (Not taking in account possibility of redistributed longevity escape velocity in a few years) [Entropy | Free Full-Text | Principled Limitations on Self-Representation for Generic Physical Systems](https://www.mdpi.com/1099-4300/26/3/194) [[2012.00152] Every Model Learned by Gradient Descent Is Approximately a Kernel Machine](https://arxiv.org/abs/2012.00152) poslední dny dost přemýšlím jak by šla udělat nějaká vizuální mapa všech možných systémů ve vesmíru a ke všemu asociovat matiku která ty systémy určuje je spousta super obecný matiky a super konkrétní matiky, spousta matiky jde aplikovat skoro všude, takže bych musel nějak vizualizovat jak moc je konkrétní podobor/matematický konstrukt obecný versus konkrétní, jak nějaký rovnice platí pro velkou skupinu systémů a jak nějaký platí jen pro malou skupinu a bylo by hodně cool kdyby to bylo realtime automaticky updatovaný 😄 ale tý existující matiky je prakticky nekonečno, a ne všechny modely se empiricky testovaly, nebo některý jsou i po experimentech kontroverzní, a ještě nám ještě spousta rovnic a řešení různých systémů chybí, a ne všechny matematický konstrukty mají korespondenci ve fyzice (většina nemá) Math is like a modelling clay that you can compose like lego and in particular shapes and configurations corresponds to real world physics differenciální rovnice jsou základ všeho [Boltzmann equation - Wikipedia](https://en.wikipedia.org/wiki/Boltzmann_equation)