Have you ever considered a river as an intelligent system? Like a learning algorithm
>>16799533no
>>16799534Why not?
>>16799547I'm not retarded
>>16799533If the problem is, "Find the pathway of steepest decent?" then the greedy local algorithm performs very well. Mid-wits call this emergent and measure it with qualtitative analogies. Can you belive it?
>>16799533Imagine being an Indian rivershitter and just not understanding at all how intelligence, free will, Christianity, etiquette, grace, the ant and the grasshopper, all sort of come together to synthesize modern humanity.
>>16799553It's a bit more complicated than that I'm sure you'll agree, does like image look like steepest descent of any parameter?
>>16799566NTA but yes, it does.
>>16799533What do you even mean by that? A river is a system and there's a lot of stuff that lives in and around them and in that sense it can be said to be a system and it contains life. But the river itself is just the lowest area where water collects and flows. It's not itself an intelligent agent.
>>16799573OP is an H1B welfare baby who thinks the path of least resistance is a form of intelligence.
>>16799577t seething llama engineer working on probabilistic transformer architectures that can't even capture river dynamics
>>16799582I know how to shit in a ceramic bowl. Visa denied.
>>16799566ATA, that is exactly what it looks like. The aggregate results of guadrillians of individual "steepest decent" decisions looks exactly like this.I bet we can find a really good recursive rule for all those tree branches too.
>>16799573>But the river itself is just the lowest area where water collects and flows.Correct, an aggregate of guadrillians.>It's not itself an intelligent agent.No shit.
>>16799582>lama engineerNO BIOLOGISTS!
>>16799533In fact, yesss.
>>16799599No anon, it doesn't. You'd have to constantly change the curve youre descending, the dimensionality is far too high to say it just has a bajillion parameters. you can't just compress it to a single gradient field
>>16799805And yet, water does exactly that.And never once worries the fuck about it.Wow.Chudda > Soilentboi
>>16799806No it doesn't change any parameters at all actually
>>16799807Not quite Mensa material, are you?
>>16799809>A river updates parameters Oh I wasn't aware it's algorithm was mutable
>>16799533How does the river apply what it has learned?
>>16799560>ChristianityOnly thing it created was 1 billion Africans
>>16799533>Have you ever considered a river as an intelligent system? Like a learning algorithmSure I have. As a teenager, in the shower. Then I realized that by the same token, almost anything can be construed as "intelligent", rendering the distinction meaningless. I wasn't the kind of pseud who finds pleasure or substance in destroying useful distinctions, so I dropped it and concentrated on washing my penis.
>>16800396>How does the river apply what it has learned?How do you?
>>16800423kill yourself fag
>>16800424I no longer reply to bai-
>>16800544>b-b-b-bai...C-c-c-concession. You embody what you have learned (in your case - nothing) and apply it by being.
>>16799533There’s no prediction going on there, it’s just settling into a local minimum energy state. That’s no different than dropping a ball and calling the ball smart for falling. It didn’t gauge the surrounding space time, find the optimal radial in geodesic, account for the effects of air resistance and then stick to some pre-computed path. It just fell. It was acted upon the whole way down, never taking any action of its own.QED, that image. If the water is “intelligent” and, say, “wants” to be in the ocean, why has it not yet carved a straight path? Why does it require millennia of storms and floods to erode away its banks and “find” some slightly more favorable route through erosion?
>>16800569People will write posts like this one and then use the term "AI" unironically.
>>16800570Explain please
>>16800575Optimization algorithms used for ML (e.g. gradient descent variants) do nothing more "intelligent" than water passively following the natural gradients of the terrain. Inference is even more trivial - the bulk of the processing follows some fixed, pre-established path.
>>16800582By that token there’s a pre-established path in everything from mathematics to your own brain. I’d still say an algorithm is smarter than a river because it abstracts the problem. It’s no longer physically brute-forcing the solution but doing it computationally. Smarter still would be some second order abstraction that achieves the same results with less computation.
>>16800618>By that token there’s a pre-established path in everything from mathematics to your own brainProof?>I’d still say an algorithm is smarter than a river because it abstracts the problemThe algorithm doesn't "abstract" anything. It's humans who abstracted the problem.> It’s no longer physically brute-forcing the solution but doing it computationallyPutting aside the fact that "computations" don't exist, the computations for fluid dynamics that simulate a river are in no way more or less "intelligent" than the ones for Gradient Descent.
>>16800636>proof?Unless you’re a substance dualist your brain is operating on stimulus-response. There’s a definitive structure for mathematics that one can’t just vibe-calc.>the algorithm doesn’t abstractI never said it thinks for itself. It itself is an abstraction of the problem. I’m well aware these things don’t just manifest ex nihilo.>computations don’t existI never said that any atomic operation was imbued with the divine. If your overall gripe is that people ascribe some kind of personhood or agency to Grok then I’m with you. Otherwise you seem to be bickering over the imprecision of language.
>>16800677>Unless you’re a substance dualist your brain is operating on stimulus-response. There’s a definitive structure for mathematics that one can’t just vibe-calc.You just said absolutely nothing.>It itself is an abstraction of the problemYou said the algorithm is intelligent because it abstracts the problem. Since you take that back, what does being an abstraction of a problem have to do with being intelligent?>I never said that any atomic...PUTTING ASIDE the issue of attributing intelligence to abstract non-entities, the computations for fluid dynamics that simulate a river are in no way more or less "intelligent" than the ones for Gradient Descent.
>>16800582Your brain does nothing more "intelligent" than responding to electrical signals with a simple threshold function.
>>16800696>Your brain does nothing more "intelligent" than responding to electrical signals with a simple threshold function.Evidence of this psychotic delusion?
>>16800569Does that image look like a local minimum?
>>16800700Path of least resistance.
>>16800704Wrong again
>>16800706I don't know what you mean by "again". You're very clearly mentally ill and in denial about basic reality.
>>16800709Ok, if it's least resistivity, what is the resistive medium?
>>16800721The environment it flows through, with its height variations and obstacles and a fucking riverbed where the flow of water has already cleared a path.
>>16799533is a coin sorting machine sentient?
>>16800761Since when was intelligence sentience?>>16800745Observe:OP image
>>16800788>Observe:OP imageCircle a specific part of that image that is somehow inconsistent with what I just told you.Protip: you will deflect instead, which I will accept as a full concession.
>>16800790You will argue the bends are due to surface inconsistencies, when they arent. That path is not least resistance, this is a well known problem, it's actually closer to maximum. Thanks for proselytizing your cultism, and have a good day
>>16800813>the bendsEven random gradient noise can easily produce such bends for water to flow through. Bends prove absolutely nothing.>That path is not least resistance, this is a well known problem,And yet instead of providing a source for your "well-known" take, you get anally blasted and immediately scamper from the thread, just as you should. :^)
>>16800824You have any idea the error on a model predicting even a single curve across tens of meters and a literal frozen time slice? You can randomly generated curvies all you want, this problem is currently intractable
>>16799577it kind of reminds me of that indian E = mc2 + AI guy
>>16800836I notice you're very eager to dump your word salad yet very hesitant to provide this elusive source which supports your "well-known" take about water preferring paths more resistant to its flow, when less resistant paths are available.
>>16799806>what is erosion
>>16801084How does that cause local steepest decent to fail? It literally reinforces the need for a locally knowledgeable and global ignorant algorithm.You have consumed way too much whizz for one being.
>>16800880>Not looking For it yourselfWhen something challenges your presuppositions the burden of verification is on you, unless ofc your greedy learner doesn't have an escape function
