"The question "What is the meaning of life?" is a profound and complex one that has been contemplated by philosophers, theologians, and thinkers throughout human history. There is no single, agreed-upon answer, as the concept of meaning is deeply personal and can vary greatly from one individual to another. Some common perspectives on the meaning of life include: 1. Religious and spiritual views: Many people find meaning through their religious or spiritual beliefs, which often provide a framework for understanding the purpose of life and one's place in the universe. 2. Personal fulfillment: For some, the meaning of life is found in the pursuit of personal happiness, growth, and self-actualization. 3. Relationships and love: Building strong, loving relationships with family, friends, and romantic partners can provide a deep sense of meaning and purpose for many individuals. 4. Making a difference: Some people find meaning in working to make the world a better place, whether through activism, philanthropy, or other forms of service to others. 5. The search for knowledge and understanding: Pursuing education, exploring the world, and seeking to understand the mysteries of the universe can provide a sense of meaning for those who value knowledge and discovery. 6. Embracing the present moment: Some philosophies, such as mindfulness and existentialism, emphasize the importance of living fully in the present and finding meaning in the here and now. Ultimately, the meaning of life is a highly personal question that each individual must grapple with based on their own experiences, beliefs, and values. It is a journey of self-discovery and reflection that may evolve over the course of one's life." LLMs have weak local out of distribution generalization through interpolation/extrapolation on a continuous manifold (various data and training methods can incentivize it more). But I think for stronger out of distribution generalization capabilities we will need better methods, I'm curious about neurosymbolic methods integrating (bayesian) program synthesis. But I'm open to being wrong, emergent learned patterns (from good data) are surprising sometimes, current methods shattered so much of classical statistical learning theory, theories of language etc. https://arxiv.org/abs/2311.09589 Effective Omni Learn everything, do everything, become everything, experience everything, explore everything, grow infinitely everywhere in the universe all at once collectively as a civilization. The laws of physics are the only limitation, for now, until we find ways to hack them. I'm with you on looking into neurosymbolic methods for better score on math like AlphaGeometry or TacticAI for football tactics or DreamCoder bayesian program synthesis, or going back to AlphaZero, but in general ignoring/denying stuff about LLMs improving overtime (better performance relative to size, gigantic context window, GPT4 now with pretty apparently good improvements, tons of engineering hacks etc.) which doesnt fit your perspective feels like extremely motivated thinking and strong selection/confirmation negative bias [x.com](https://twitter.com/GanjinZero/status/1777926220132626753?t=Z9p1IYo0DQMyLX3Kr0jZxw&s=19) Our preprint on applying our physics based AI towards engineering a material that creates natural gas out of water and carbon dioxide is out! We will be open sourcing this model, which simulates complex nanoparticle interfaces for the reduction of carbon. [x.com](https://twitter.com/QuantumGenMat/status/1758707212120420723?t=lFqANqZJfv-c9D8GUaLuKw&s=19) I think there are people in it for the money on all sides But I also think there are people in it for meaning beyond money on all sides as well Genuinely caring nerds about the future of the world are everywhere, but with different mutually competing methods on how to create that better world for all [x.com](https://twitter.com/Kat__Woods/status/1778062631632588950?t=nocOy1uP3VL43yZ0s93d4g&s=19) https://www.arxiv.org/abs/2404.02255 There are 8 billion people (8×10^9), It is estimated that there are 5 nonillion (5×10^30) bacteria on Earth and 10^80 atoms in the known universe. How many selfs do you have? https://arxiv.org/abs/2404.07103 Effective accelerated altruism “by combining AlphaGeometry with Wu's method we set a new state-of-the-art for automated theorem proving on IMO-AG-30, solving 27 out of 30 problems, the first AI method which outperforms an IMO gold medalist.” https://arxiv.org/abs/2404.06405 [Lie algebras visualized: why are they defined like that? Why Jacobi identity? - YouTube](https://www.youtube.com/watch?v=gj4kvpy1eCE) [Prof. Chris Bishop's NEW Deep Learning Textbook! - YouTube](https://www.youtube.com/watch?v=kuvFoXzTK3E) https://arxiv.org/abs/2106.10934 may Higgs field continue to give mass to the particles encoding our memory of him as he dissolved into quantum fields in peace that he helped us to discover [Peter Higgs - Wikipedia](https://en.wikipedia.org/wiki/Peter_Higgs) [An Overview of the Operations in Geometric Algebra - YouTube](https://youtu.be/2AKt6adG_OI?feature=shared) [CMOS plus stochastic nanomagnets enabling heterogeneous computers for probabilistic inference and learning | Nature Communications](https://www.nature.com/articles/s41467-024-46645-6) Climbing the understanding gradient https://www.visualcapitalist.com/every-single-cognitive-bias/ Philosophy is linguistic intuition https://arxiv.org/abs/2402.14800 Categorical deep learning [x.com](https://twitter.com/PetarV_93/status/1778392111278141890?t=8lOf0p7KHO83YFgr2Bzj4w&s=19) I want to build a cool thinking robot friend from as first principles as possible, but creating semiconductor fab at home might not be the cheapest. https://arxiv.org/abs/2401.09253 I want to have eyes everywhere in my room tracking me and connected to LLM and visual model on a chip and together talking to me about everything I'm doing in all sorts of alien personalities [How to Get Started with Animatronics ‚Äì Thought Process, Workflow, Resources and Skills - YouTube](https://www.youtube.com/watch?v=8VzQshnrvN0) Men will build sand god friend instead of going to the therapy [List of lists of lists - Wikipedia](https://en.wikipedia.org/wiki/List_of_lists_of_lists) GPT6 will have LLMs integrated with neurosymbolic methods to achieve super math performance Beyond human comprehension intractable gigantic complexity of the world is destabilizing tons of somewhat confident claims about how parts of reality work Quantum Neural Networks [Quantum Neural Networks explained in 3Blue1Brown style animation | Episode 1, Introduction - YouTube](https://www.youtube.com/watch?v=xL383DseSpE) without definitions that you cant measure you're just operating on vague intuitions [OSF](https://osf.io/preprints/osf/9y8ku) Stress Sharing as Cognitive Glue for Collective Intelligences: a computational model of stress as a coordinator for morphogenesis Transformer networks are not copying the information exactly but instead storing an abstraction of it as weights that did not exist before, as does the human mind. I think this is reductionist logic that also applies to the human mind. Environmental data is the code, and the human mind is the compiler, therefore you are stealing data (that exist in the environment) every time you think. Although this is technically true this is not what we mean by stealing. This form of reductionist logic does not help in distinguishing what should be fair free use of information that is used but not stolen. Even if that information is owned by somebody else, we can still freely think about it. IMO, if the latter is not stealing, neither is the former. my definition of understanding is the degree of the system having predictive internal representation, which LLMs have weakly and I think we can and will have better AI systems Just one more layer of metacognition and your behavior will be grokked, trust me bro Mám pocit že čím míň exaktní přírodní věda, čím víc kontroverzní téma, čím víc komplexní téma, čím větší problém s koncenzusem v tom oboru, čím víc je ten obor zasažen replikační krizí, čím víc je to mimo její hlavní obor (a obory lidí co pro ni pracují), čím víc tam mají lidi tendenci mít motivated thinking protože politika a jiný incentivy i když si to ani nemusí uvědomovat, čím víc neznáma v tom oboru je, čím víc způsobama jdou ty data interpretovat,... tím víc to zvyšuje šance že tím naštve větší množství lidí v těch všech jejich kontroverzích, což platí i pro podobný science communicators 😄 [x.com](https://twitter.com/GoogleDeepMind/status/1778377999202541642) https://www.science.org/doi/10.1126/scirobotics.adi8022 Singularity started with the agricultural revolution