Quara: "The free lunch theorem in the context of machine learning states that it is not possible from available data to make predictions about the future that are better than random guessing. Data by itself only tells us the past and one cannot deduce the future from it."
Really? That's not true. The machine learns better probabilities, which are never perfect but more approximate. The whole world and beyond has patterns / is predictable. What I gathered from Wikipedia is that the theorem states a machine/algorithm is good at basically x and so is some other machine/algorithm good not at x but it is good at basically y. So all machines/algorithms possible in this universe are equally useful/valuable, meaning water bottles, 1 atom, serial killers, computers, GPT-2, a galaxy, law of physics, and a human are all equally useful/valuable and as capable. But this can't be :D. The world works by the laws of physics, we aren't spirits from heaven, we are machines, we work a CERTAIN way, and only that way. If particles different follow predictable laws, then they could only act randomly, and either way you pick there's no free will. We aren't lucky to have the physics we do have, we are stuck in it and are happy like water is stuck flowing down a river and has to be happy doing what it is doing. So yes AGI is possible and we are real examples. We are multiplying all over Earth our babies/knowledge/skills and are becoming closer to the best machine that can't be killed; Darwinian Survival. Earth will become nanobots and can instantly self-regenerate and body organs / knowledge / skills that it loses. All of Earth will be nanobot modules made of nanbot modules so that every nanomodule bot knows where, when, and what everything they want is, for best resource utilization. From chaos (no patterns) evolution/physics BRINGS a approx. full pattern fractal in a large enough system like Earth. During chaos, the particles themselves are the only patterns based on their laws. Their sorting/evolution is what evolution/intelligence is, sorting items (particles) to make patterns emerge. Of course it happens on its own, as all does, so it's self-organizing particles. You can't create/delete particles or computer bits, only sort them, such as the magnetic 1s and 0s on your HDD or the electrical pulses on your circuit board. Lossless Data Compression is an evaluation for AGI as Matt Mahoney and Marcus Hutter proves, the more data you have the better the compression and the better the extraction ratio from such little data. The compression is the Learning, the extraction/decompression is the insight extraction, which it regenerates and then swallows back to add it to its knowledge. Planets and atoms are unstable when too large because of all the combinational interactions in the core, and the compressed core explodes extractions giving back the star's energy/matter it robbed. Black holes too. Fire makes cold wood release energy, so do our stomachs and so does car gasoline, from nothing, a cold motionless battery, you get free energy, free extraction insights (related data from insight dataset to compress). Human agents are food finders, like cells, we burn fuel, our own goal could kill us like the Paperclip Effect, and global warming since nanobots will find so much food and energy. Humans find food better and breed and survive death faster because we are smarter, and we do this goal because it makes it replace the non-smart ones, Darwinian Survival of the fittest. So AGI is all about survival/fleeing/uncontrollable breeding of nanobots, using smart methods to replace/update the rest faster than competition by aligning magnet domains with the other neurons/friend connections to emerge a strong laser / faster brain wave propagation in the swarm. And the more context you have; swarm, the smarter you are, extra context helps, for every pattern you know you can answer/solve 10 unseen problems. Our unlighted at birth domains during chaos align and cooperate more near the end, exponentially faster as we see today the progress curve is shooting up. Our operation is a global effort, we affect each other as context, I am defined by all of you and vice versa, global context. Local context is more noticeable; non-subconscious context; high awareness attention. In a dictionary each word explains each other, like Word2Vec network and human friend network, in this small world network dictionary you can reach any node in just a few steps, hence local and global humans can affect each other despite far away ex. I affect the president and not just my mom the most. It's background noticeable, not prominent. So yes there is a universal intelligence and patterns. There is a way the universe/particles work, and this IS it, we are it and in it. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T353f2000d499d93b-Mb03fafb87ea4cc3dd4e32dc0 Delivery options: https://agi.topicbox.com/groups/agi/subscription
