The Universe controls "you", whatever you are, but some of "us" are too much of "control freaks". Neither humans control the machines or AI, nor the reverse (and always some resolution, range, specifications are required for what does it mean to control), nor one agent controls the other, in the true meaning of "control". It could only do at a simplified virtual universe at a low resolution and with forgetting about the real processes, a sort of self-deception.
Whatever you do, unless you are God who has created or "controls" the Universe, you will always be a puppet in the full sense of causation. "Conception about the Universal predetermination", 2002 or "The Matrix in the Matrix is a matrix in the matrix", 2003: the movie "The Matrix" and the famous psycho-looking "AI risk expert's" "scary" interpretation about the "simulation hypothesis" are ridiculous in their suggestion that you are controlled "by the Matrix", however if you plug yourself out from some of the gross "jacks" *that you can see*, you will be free and will enter "the real world" and your "free will" will decide your future. The actual plugs are all elementary particles or whatever the backend is, both which are "part of you" in the trivial sense, and the whole environment, the Universe where you're operating, because it can crush you immediately, if it wishes, and you cannot exist without it. The connection is not a few cables as in the Matrix, and it shapes and controls "you" from the ground-up and from the inside, it built you and it destroys you, these forces and substances have created you and directed what you wanted and what you could want, and they do it by every elementary particle in "your" body, mind or whatever you are and all which interact with them. You cannot escape from this "jail", if you believe it is one, silly monkey. Neither you can *really* control "AI" (the Universe, "AI" are some displays of some processes), but you may only convince yourself that you do control it, some simplified and narrowly defined phenomena, at some selected resolution and subspace, ridiculously low compared to the one of the actual machine language of the Universe. Matt in particular, haven't you just shared how low is the actual bit rate of human thought, mind etc. and didn't you agree - another part of the Bulgarian predictions, - however he and the authors of the paper don't interpret it correctly, unlike the classical cyberneticians like Ashby or the mentioned predictions/prophecies. The "10-bit" or "few dozen bits" suggests namely that the low level processes *control* you, they have higher complexity, "requisite variety" etc. and human "conscious" or goal directed behavior is actually extremely rigid, non-creative and with very little elements of choice and "free will". You go on narrow rails and it is true that your masters want to put you in even narrower tunnels. If you limit the evaluation at low resolution patterns or phenomena, such as typing on a keyboard or the benchmarks of the current AIs, this reminds of the problems of the GOFAI and "Moravec paradox", with solutions working as "if dog then bark, if bird then sings", encoded as A0 F2 A9 05 82 59 04 however without considering or describing the encoder who produces this code (some human programmer does it, assisted by computers). If you can reduce the representation to this form and if you have a decoder, and you don't count their complexity, and you exclude the backend, then it will be encoded in 10 bits and you will control it: in your imaginary model. However you neither have the encoder, nor you include the actual "minimum description length" or "Kolmogorov complexity" of the system, which uses the proper Virtual machine (not an abstract Turing one: it should be compatible with the Universe computer, if it's a Turing machine it needs to have proper materialized form and "adapters" to the machine code of the Universe). A proper machine is one which can produce the encoder and the decoders, i.e. to produce from simpler substances and state of the environment humans, human society and all their actual specific configurations, everybody should be at the right position with right thoughs, desires, memories, goals and representations; all the required technology and the interaction between them and it must be complete at the resolution of causality-control of the mother Universe, i.e. Planck constant. How many bytes are recorded by the accelerator in CERN just for a single experiment for two particles colliding, how many particles are just in a 2x1 m cube around a human, how they could interact and with how many others they interact from the rest of the Universe x 10^35 times/s. It is just to show how miserable slave you actually are, and what a parody is the believe that you, whatever is "you", could really control the AI (again whatever you call it, because that's anothr confusion*) the Universe, the nature etc. * What's the "AI"? and *WHERE* is it? The weights of some LLMs? (They are meaningless without interpretation). The computer which runs them? Which is the computer - the server with the GPU? The Internet servers, routers, switches, your local terminal? All on the line. What about the OS, compilers etc.? The electrical infrastructure and all hardware? What about the datasets - spread and collected around the world. All computing is part of the "AI", there is no sharp border, the "AI" is causally distributed everywhere, but human cognition, just like the current LLMs, likes to thing in jaggy localized tokens, often wrong. The LLMs/NNs are paradoxical/dialectic ones - they are both a promotion of "distributed representation", yet the actual meaning has to be initially and finally localized and humans want "explainability" precisely as highly localized correspondence, they want to pinpoint "who's responsible" and judge her, and it should be just one "cut" token or "neuron" or "layer". The wider the meaning, the scope, the range of analysis, the more this is wrong simplification. In a complete sense and in a wider range, high resolution, the "credit assignment" problem, causality etc. are defined by the specific application of a choice of the evaluator-observer. The observer and his filter decides how causality is evaluated, at which resolution, where the tree of causal influences and relations is cut and why etc. Here again what "AGI" or "AI" is, what is "to control" etc. ... The Sacred Computer: Thinking Machines, Creativity and Human Development https://github.com/twenkid Join Thinking-Machines-2025/SIGI-2025 rolling conference online ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tff34429f975bba30-M66dab109775b9abd891de357 Delivery options: https://agi.topicbox.com/groups/agi/subscription
