I find it hard to believe that evolution is completely blind. The search space that it can explore via mutations is astronomically large. Considering that the experiment is run at planet-scale over billions of years doesn't really save the argument as it takes some specimen years to develop and get feedback on their fitness. It's hard to believe that it's truly just random "bit-flips".
I'm not trying to suggest woo here, but there has to be some mechanisms to constrain the search space somewhat.
The search space is highly constrained. All life on this planet is based on hydrocarbon chemistry, more or less, and must operate in the face of high rates of oxidation and water as pretty much the only available solvent. Even with such constraints, the differences between what has evolved (bacteria to blue whales! viruses to polar bears! algae to orchids!) are staggering.
The fact that you find something hard to believe doesn't say much at all. Humans have all kinds of things that we find hard to believe - for example, I find it almost impossible to believe that there is only one object I can see in the night sky with my own eyes that is outside of our galaxy - but that doesn't make them any more or less true.
Let's take human DNA as an example. It contains 3.2B GTCA base pairs. This gives rise to 4^3.2B possible combos. It's just not possible to navigate this space blindly. There is not enough atoms in the universe to do that. It is known that there is bias in what mutations are favoured.
Only a tiny percentage (around 1%) of the DNA in chromosomes codes for proteins.
And yes, certain mutations are favored precisely because of the chemistry constraints (an extremely basic one is which base pair changes actually alter the resulting protein; a more sophisticated one is which amino acid changes alter the physical functionality of the protein).
Of course there is bias, the bias is provided by the natural environment where the organisms coded by the genome must thrive or die. The bias is applied after the mutation occurs, but the mutations themselves are random, or nearly so. Probably there is some differential rate between the likelihood of each of the four base pairs to mutate into each of the others, but I would guess its nearly parity, because that would probably be close to optimal (though that depends on the details of the genetic coding scheme, ie the triplet code that translates nucleotide triples into amino acid codons).
There are multiple constraints that I can immediately identify. Maximal temperature extremes, barometric pressure, atmospheric/substrate compositions, etc. The bias is inherent to the history of the planet Earth and the gradients present across that time and space. I'd say it's highly constrained.
But is the diversity really that staggering?
I mean most animals including possibly dinosaurs that have ever existed share a lot of internal organs, in the same place. They have eyes, brain (with a lot of the same brain areas, even birds have something like a prefrontal cortex but it's called something different).
They all have legs, torso, head.
I would say there is a lot more commonality than difference. The differences come from slight variations on a basic template that works, and then the body looks different and so on.
I'm not sure how to think about the diversity that evolution creates and how diverse it actually is. I would say there are _a lot_ of repeating patterns all across history, with variations on those repeating patterns always changing.
You're choice of samples is rather skewed towards ones sharing a relatively recent common ancestor. Octopus and Sea Squirts are also animals, and they don't have legs or torsos or, in the later case, heads or eyes. Octopus brains are also rather different from those of vertebrates, and they have 8 mini-brains for more distributed/localized control of each major limb.
That said, I agree with you that there is a lot of commonality in life. Even in the case of Octopus we share a lot of DNA. I just mostly think that is due to common ancestor and common environmental pressures, not to some fundamental limit in the breadth of evolutionary potential itself. Its probably worthwhile to wonder at how that actually works though. Maybe evolutionary potential could be improved.
> some mechanisms to constrain the search space somewhat.
Your perspective has the unfortunate bias of being posed at the end of a long stream of evolution that happened to emerge with an intelligence far superior from other living things.
> Considering that the experiment is run at planet-scale over billions of years
It's not just planet-scale, it's universe-scale. Lots of planets conduct the experiment, ours just happens to have resulted in intelligence.
> It's hard to believe that it's truly just random "bit-flips".
Mutations introduce randomness but beneficial traits can be selected for artificially, compounding the benefits.
> It's not just planet-scale, it's universe-scale. Lots of planets conduct the experiment, ours just happens to have resulted in intelligence.
My argument doesn't depend on the existence of an intelligent species on the planet. The problem already arises when there are multiple species on ONE planet. If you calculate the pure combinatorial distance between the DNA of 2 species, you must find that you can't just brute force your way from one to the other before the heat-death of the universe. This is why mutation bias exists: not all mutations are equally likely, evolution favours some kinds over others.
> If you calculate the pure combinatorial distance between the DNA of 2 species, you must find that you can't just brute force your way from one to the other before the heat-death of the universe.
Can you expand on this? I'm not seeing why it is implausible for one genome to mutate into another, that seems like it could be accomplished in reasonable time with a small, finite number of mutations performed sequentially or in parallel. After all the largest genome is only about 160 billion base pairs, and the average is much smaller (humans are 3 billion base pairs). So what's the difficulty in imagining one mutating into another?
Your maths doesn't seem right. You can estimate mutation rates very easily, and you don't end up at crazy numbers. The sequence space explored by evolution is tiny compared to the possibilities and closely interlinked. A simple example is comparing haemoglobin sequences from different animals.
The constraint is a life-forms' existing form. A given genetic sequence can only move (in general) a small distance from the existing sequence.
Since you're already starting with a successful sequence, the odds are that a small variant on that sequence is also going to be only marginally more or less successful than the original sequence.
Epigenetics can arguably be an example of what the comment means by narrowing the search space. You can have heritable changes to gene expression that are not part of your genome, but are a result of feedback from the environment (and not random mutations, viability of which natural selection will judge over future generations)
Look at software fuzzing, particularly the coverage guided mutators (basically a simple “genetic algorithm”.
It’s amazing what a few random bit flips combined with a crude measurement can do.
To me, evolution at first seem implausible. Monkeys banging on a typewriter aren’t going to write Shakespeare. But add a crude feedback loop to them, and soon they’ll be dishing out Charles Dickens too!
truth = claim.replace(/I'm not (.*?), but (.*)/, "I'm $1.");
Then again this is a discussion about "Experts explore new mushroom which causes fairytale-like hallucinations" so maybe woo is appropriate, and you should embrace it.
Is there a way their question could have been phrased that would have not drawn you to make that assumption, which seems to be an ethos attack, or are you predisposed to reply in such a way about any philosophical evolution question?
When people say /I'm not (.*?), but (.*)/, they invariably are what they're claiming they aren't. That's what that phrase means. For example, we've all heard it a million times from people defending their vote for Donald Trump. There's even a wikipedia page about it:
If you really mean $2, then just say $2, you don't have to preface it with "I'm not $1, but". That's a waste of words, beating around the bush, a rhetorical shield, that reveals that you really are $1 and you feel the need to be defensive about it.
The word "but" in that context means the thing before it is false, just air escaping from the folds of your fat, and you can ignore everything before the "but".
"But" is a contrastive conjunction, signaling the clause before "but" is expected, socially required, or reputationally protective, and the clause after "but" is the actual communicative payload. It means to discount or ignore $1 and evaluate the speaker by $2. Saying “I’m not $1, but $2” does not strengthen $2, it does't make $2 safer or clearer, it just signals defensiveness, and undermines credibility.
Again, this is a discussion about psychedelic mushrooms, fairytale-like hallucinations, and machine elves, so woo away all you want!
I have little patience for intelligent-design and the likes, if that's what you are getting at.
All I'm saying is that blind enumeration of mutations seems combinatorially infeasible due to the vastness of the search space. It is already known that mutation bias exists, so what I'm saying shouldn't be that controversial.
In stark contrast to what you're claiming, I have absolutely zero patience for intelligent design and the likes -- that’s exactly my point.
All I'm saying is that the whole point of the theory of evolution is that blind enumeration of mutations is not required, and that combinatorial feasibility emerges in spite of the vastness of the search space. It is already well known that mutation bias exists, so none of this is controversial.
Multiple commenters here have already explained this from different angles, including chemical and environmental constraints (PaulDavisThe1st), developmental and functional constraints (Supermancho), and even software analogies like coverage-guided fuzzing and genetic algorithms (BobbyTables2). These are not fringe ideas; they are standard ways of explaining why your "astronomical search space" framing is a strawman.
You are hedging; I am not trying to weasel word or distance myself from evolution, or use red-flag rhetorical "I'm not $1, but $2" devices. I have read, agree with, and acknowledge the other replies to your message, because I understand that evolutionary theory already fully explains the concern you're raising.
Your claim that "blind enumeration of mutations seems combinatorially infeasible due to the vastness of the search space" flatly contradicts the theory of evolution.
This has also been directly challenged by other commenters asking you to justify the alleged combinatorial barrier in concrete terms (uplifter), and by others pointing out that genomes do not need to traverse all possible combinations to move between viable states.
The entire point of evolutionary theory is that blind enumeration is not required, and that combinatorial feasibility emerges from selection, heredity, population dynamics, and cumulative retention of partial solutions. No "woo" is required.
Evolution is blind with respect to foresight, but not blind with respect to feedback, structure, or retention.
Mutation bias, developmental constraints, and non-uniform genotype–phenotype mappings are foundational components of modern evolutionary biology, not ad-hoc patches.
People who doubt evolution tend to rephrase it into a strawman -- "random bit flips over an astronomical search space" -- and then declare that strawman implausible.
Several replies here explicitly reject your framing. For example, thrw045 points out the massive reuse of structural templates across species, and PaulDavisThe1st notes that only a small fraction of DNA even codes for proteins, further undermining the idea of a uniform, unconstrained search.
Your "I'm not pushing intelligent design, but evolution seems combinatorially infeasible" move closely mirrors the Discovery Institute / "teach the controversy" pattern: disclaim ID, then introduce a doubt-claim based on a strawman of evolution as uniform random search, then retreat to "just asking questions." That strategy is explicitly, insincerely, and unintelligently designed to manufacture doubt about evolution while insisting it is not religious.
We can see the sealioning pattern play out here in real time: repeated insistence that ID is rejected, followed by reiteration of the same mischaracterized impossibility claim, even after multiple substantive explanations have already been given.
I’m not hedging like you are here: evolutionary theory does not claim "blind enumeration over an astronomical space," and treating it that way is simply a misstatement of the theory.
I think I and other people recognize your rhetorical patterns and misunderstandings, even if you don't, thus the downvotes. Other commenters have fully addressed your doubts about evolution. To me, the big give-away was your "I'm not $1, but $2" wording.
In any case, this is a thread about psychedelic mushrooms and hallucinations, so if some machine elves want to weigh in with some woo about population genetics, I suppose that’s fair game.
great, but we still cannot say anything beyond "what survives, survives". fitness is a central concept to natural selection and ultimately evolution, but it seems to bother nobody that its an empty concept, a tautology. its a nice observation but doesnt actually explain anything, and I expect science to explain the world.
Saying "what survives, survives" may bother Creationists, the Discovery Institute, and Intelligent Design pushers, but not actual scientists who don't have an ideological agenda to discredit evolution as revenge because it discredits their religious dogma.
Saying "what survives, survives" is like saying physics explains motion as "things that move, move." That’s not what the theory actually claims; it’s a caricature.
Evolutionary theory explains mechanisms, not slogans. "Fitness" is not the explanation, it’s a measurable consequence of those mechanisms.
If you want teleology or ultimate purpose, science won’t give you that, so take some shrooms and ask the machine elves. But evolution absolutely explains how structured complexity accumulates without foresight, and it does so with predictive, testable models.
Please inoculate yourself against believing and parroting anti-science Intelligent Design / Creationist talking points by understanding where they come from, and what they led to.
Intelligent Design is a religious ideology, not a scientific theory. The Discovery Institute is the evangelical advocacy organization that systematized and promoted it.
After Intelligent Design failed legally and scientifically, the Institute pivoted to the "Teach the Controversy" strategy -- not to advance new science, but to manufacture doubt about evolution in public education.
That approach -- revealingly effective -- became a template for later efforts to reintroduce religious ideology into secular institutions. Project 2025 represents the political continuation of that same strategy at a much larger scale: shifting from attacking a single scientific theory to reshaping education, governance, and public policy along explicitly religious lines.
>Have you heard of the Discovery Institute? Have you fallen under the impression that they know what they are talking about, or can be considered an even remotely legitimate source of information?
>Well, you've come to right place. They aren't. They're a propaganda mill, and all of their content is full of lies.
>They hide behind a paper-thin roster of scientists who have deluded themselves into dishonestly preaching outside of their expertise, and they blatantly misrepresent any scientific research or scientists they are referring to. Constantly. Sometimes they even commit slander.
>That's what this video is about, and it is the first installment in a series where I will expose the fraudulent activity of all the major contributors at the Discovery Institute, one clown at a time.
>Part 1 addresses Casey Luskin, and it is centered around some very serious slander he committed against an esteemed anthropologist.
>But don't worry, I cover lots of other lies and stupidity that come out of his mouth as well.
>If you're a fan of the DI, do please find the courage to watch this rather than running to the comments section to yell at me.
>It's not all that long, and I promise that I make it extremely clear and undeniable that Casey is a liar. If you have a shred of honesty within you, you will quickly see that this is the case. Enjoy!
This dives into the "Junk DNA" cannard:
Exposing Discovery Institute Part 10: Casey Luskin Again (Because He's Such a Loser Fraud)
I think you are psychoanalyzing me a little bit too much. Am I allowed to say that I'm an atheist and I don't believe in intelligent design, or are you going to explain to me that I'm confused about my own beliefs?
It's exactly because you’re rationally reachable that this matters.
I’m not questioning your beliefs, and I’m not saying you secretly believe in Intelligent Design. The issue is that some of the arguments you’re making didn’t originate organically or scientifically -- they were deliberately promoted through deceptive education policy and textbook standards, especially in large markets like Texas, precisely because that influence scaled nationally. People often absorb them without realizing their origin.
After Intelligent Design spectacularly failed in court, its proponents pivoted to influencing education standards rather than arguing science directly. In Kitzmiller v. Dover Area School District (2005), a U.S. federal district court ruled that Intelligent Design is not science and cannot be taught in public school biology classes because it is religious in nature.
>[Creationist defense witness] Fuller memorably called for an "affirmative action" program for intelligent design, which did not win much favor with [Judge] Jones in his final decision.
That was one of the most jaw-dropping moments in the entire Dover trial: an unintentional confession that Intelligent Design cannot meet the standards of science and therefore must be smuggled into classrooms under a quota system. "Teach the Controversy" is affirmative action for bad ideas: a grievance policy masquerading as pedagogy.
>"Witnesses either testified inconsistently, or lied outright under oath on several occasions," [Judge] Jones wrote. "The inescapable truth is that both [Alan] Bonsell and [William] Buckingham lied at their January 3, 2005 depositions. ... Bonsell repeatedly failed to testify in a truthful manner. ... Defendants have unceasingly attempted in vain to distance themselves from their own actions and statements, which culminated in repetitious, untruthful testimony." An editorial in the York Daily Record described their behavior as both ironic and sinful, saying that the "unintelligent designers of this fiasco should not walk away unscathed." Judge Jones recommended to the US Attorney's office that the school board members be investigated for perjury.
So the bald faced liars and $1,000,011 judgement losers pivoted to "Teach the Controversy", and states like Texas were their key targets, because of their centralized textbook approval process and market size, which historically shaped textbooks used nationwide.
The "Teach the Controversy" framing was designed to insert doubt about evolution without explicitly promoting religion, and its language appeared repeatedly in state curriculum debates.
As a result, many people encountered these arguments against evolution in school without ever being told where they came from or what they were designed to accomplish -- and repeat them without realizing how the same strategy continues today in much broader political efforts, like Project 2025.
Maybe won't be viewed favorably by the HN crowd, but I enjoyed the most recent Bret Weinstein interview on Joe Rogan [0] where Bret talks about his pet theory on natural selection / evolution (maybe 2/3 way through the interview).
Basically, the "junk" DNA we have may be "variables" that influence form and morphology, thus giving natural selection a vastly reduced design space to search for viable mutations. E.g. not much chemical difference between a bat wing and another mammals hands - mostly a difference of morphology. Allowing for more efficient search of evolutionary parameters instead of pure random walk.
1) No one asked why it's being down voted (to... -1, the horror). I'm not here for internet points.
2) This isn't my field - I am not making any claims, merely relaying what I thought was an interesting concept/mechanism I hadn't heard of before, that I thought other curious individuals here might also think was interesting. Isn't that the entire point of HN? I would have very much appreciated links or something to Google over this bizarre analysis of why my comment is downvoted. I didn't know this wasn't novel and was accepted science.
3) I understand Bret/Joe aren't looked upon favorably by certain crowds, particularly on this forum. I tried to get ahead of the "but didn't you know they can't be trusted!" comments and attempt to focus on the substance. If the substance is wrong, great! Let's talk about that.
4) You are assuming malice where there is none, and calling me disrespectful and insisting I must know things. I find that quite disrespectful and uncalled for. Not everyone has your opinions or knows what you know. 10k a day and all that https://xkcd.com/1053/
HN guidelines: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith."
You obviously know they can't be trusted, you just said so.
Why not just say that as a disclaimer to the video, instead of attempting to "get ahead" of other people who also know that, and will call you on omitting it.
Don't dismiss HN users as "certain crowds" and preemptively try to head them off at the pass for pointing out what you chose to omit. What "certain crowds" would that be, people who don't tolerate bullshit?
It's not about "certain crowds" who know Joe Rogan and Bret Weinstein are full of shit, it's about Joe Rogan and Bret Weinstein being full of shit, and you knowingly repeating and recommending their shit.
FIFY:
>Joe Rogan and Bret Weinstein are full of shit, but I enjoyed the most recent Bret Weinstein interview on Joe Rogan [0] where Bret talks about his pet theory on natural selection / evolution (maybe 2/3 way through the interview), which is bullshit.
Then don't just repeat their bullshit without question. You could have even gone as far to explain WHY they're full of shit, and who they really are, and what other malicious bullshit they spew, instead of just propagating their bullshit without warning, as if "certain crowds" are trying to suppress that vital information.
When you uncritically recommend and parrot bullshit, and try to preempt comments from "certain crowds" who you know rightfully disagree, it sure comes off looking like you believe it, which is not a good look. The strongest plausible interpretation is that you enjoy listing to deceptive idiots make fools of themselves and spread misinformation, and I'll give you that.
"Junk DNA as variables that reduce search space" is a very old idea, but it's routinely introduced in popular media as if it fixes a flaw in evolution — usually the "pure random walk" strawman. That framing is a huge tell, because evolutionary biology abandoned that view generations ago.
Weinstein and Rogan’s signature move is to take settled science, remove its context and literature, and rebrand it as contrarian revelation, implying experts missed something obvious or are hiding it. That move reliably revives strawmen, Intelligent-Design-adjacent language, and manufactured doubt, while producing zero new knowledge.
Joe Rogan and Bret Weinstein are notorious not because they're unpopular, but because they're dishonest and corrosive.
Whew boy, your comments are great examples of not being curious, understanding or promoting civil discussion.
> You obviously know they can't be trusted, you just said so.
A baseless accusation with no supporting evidence. I never asked anyone to trust anyone else. I didn't even assert the idea was true, just interesting to consider. I merely thought the idea was interesting as a layman with little knowledge of evolutionary biology.
> Why not just say that as a disclaimer to the video
Because you're putting words in my mouth that I don't believe.
> What "certain crowds" would that be
Curmudgeons like you.
> When you uncritically recommend and parrot bullshit
Really? I recommended something? I thought I said I enjoyed a video talking about a pet theory relevant to the topic at hand, which I had recently learned of.
This whole "framing is a huge tell", and "reliably revives strawmen, Intelligent-Design-adjacent language, and manufactured doubt, while producing zero new knowledge" shtick is boring and wrong. I've been atheist since I could critically think for myself and it's silly how off-base you are.
Since you seem to be so knowledgeable on the subject and confident in your position, can you point me to something I can read instead of just taking your word for it? Otherwise you're no better than them. I skimmed a few wikipedia pages [0][1] and didn't find the morphology "variables" Brett was discussing.
Then again, I wouldn't be surprised if you just pattern matched on Rogan/Weinstein, typed out your reply and don't actually know what was being discussed. The tone and tenor of your comments so far would seem to indicate so. Your entire objection boils down to "I don't like them and no one should listen to them". Light on substance, heavy on the ad hominem - not exactly persuasive.
I’m not objecting because I "don't like" Rogan or Weinstein, and I'm not saying you're religious or wittingly pushing ID. I’m objecting because the specific framing you're repeating is decades old, well studied, and routinely mis-presented in popular media as if it repairs a flaw in evolutionary theory that doesn't actually exist.
The idea that non-coding DNA, developmental constraints, or regulatory structure "reduce the search space" is not controversial, it’s foundational. What is misleading is presenting this as a fix for a "pure random walk" model of evolution. That model was abandoned generations ago and is mainly kept alive in popular discourse by critics of evolution (Creationists, the Discovery Institute, Intelligent Design proponents, Teach the Controversy perpetrators, anti-science podcasters, etc).
Weinstein calling this a "pet theory" is itself revealing. What he’s describing is not a theory in the scientific sense at all, and certainly not his. It’s a loose, personalized retelling of ideas that have been standard in evolutionary biology for decades -- regulatory architectures, developmental constraints, biased variation, and genotype–phenotype structure.
Labeling it a "pet theory" performs two rhetorical tricks at once: it makes old, well-established work sound novel and contrarian, and it subtly implies the field has overlooked something obvious that only an outsider is willing to say. That framing flatters the audience, but it misrepresents the science.
His "pet theory" is a non-refundable Monty Python dead parrot: widely known, long settled in the literature, yet periodically propped up and insisted to be alive as if it just said something profound.
Nothing here is hidden, suppressed, or newly discovered. What is new is the podcast packaging: stripping away the literature, resurrecting a long-abandoned strawman ("pure random walk evolution"), and then presenting the correction as a unique "pet theory" of personal insight rather than as settled biology. That move reliably generates the impression of deep insight without adding any.
If you want solid, non-Rogan, non-Weinstein sources, here are places to start:
Sean B. Carroll -- Endless Forms Most Beautiful:
Classic introduction to evo-devo, gene regulatory networks, and why morphology is highly constrained and reusable.
Gerhart & Kirschner -- The Theory of Facilitated Variation:
Explicitly addresses how biological systems bias variation toward viable outcomes. This is probably the closest rigorous treatment of what Weinstein gestures at, minus the hype.
Wagner & Altenberg (1996) -- Complex Adaptations and the Evolution of Evolvability:
Shows how genotype–phenotype mappings are structured, non-uniform, and historically constrained.
Pigliucci & Müller -- Evolution: The Extended Synthesis:
Covers developmental bias, constraint, and non-coding DNA without implying evolution was ever a blind bit-flip search.
Lenski et al. (2003–2015) -- Long-term E. coli evolution experiments:
Direct experimental evidence of cumulative selection exploiting structured variation.
None of this is new, hidden, or suppressed, or invented by Weinstein. It’s in textbooks and review papers.
The reason I push back hard -- and people get downvoted for recommending Rogan/Weinstein (which you already knew, just not why) -- is that their signature move is to strip this literature of context, reintroduce a strawman ("random walk evolution"), present a well-known correction as contrarian revelation, and imply experts missed something obvious.
That pattern reliably manufactures doubt without producing new insight.
So no, my objection is not "don't listen to them".
It's: don't mistake and parrot repackaged, incomplete explanations for novel insight, especially when they're framed as fixing a problem experts allegedly ignored.
If you want to understand this topic deeply, the literature above will take you much farther than a podcast -- or Weinstein’s dead parrot -- ever will.
How is pardoning people like Fauci, or even Hunter, that Trump was clearly going to target as part of an "enemies" list, more "self-serving" than literally pardoning anyone that makes you/give you millions of dollars?
(Changpenh Zhao - made him billions; Trevor Milton - donated $1.8 million; Walczak - his mom donated millions)
You don't have to prove it to me that Trump is a lot more self-serving than Biden. This should be obvious to anyone with half a brain.
That said, this shouldn't be a competition of who is "more self-serving". Just because your neighbour murdered two people, doesn't mean that you get to murder one.
> This is eerily similar to the comment on the thread about LLMs that says software teams have to maintain a "theory" or model about how the software works and when they cant, they can no longer function beyond limited forms of maintenance.
How can you so confidently proclaim that? Hinton and Ilya Sutskever certainly seem to think that LLMs do think. I'm not saying that you should accept what they say blindly due to their authority in the field, but their opinions should give your confidence some pause at least.
Do you know why they're called 'models' by chance?
They're statistical, weighted models. They use statistical weights to predict the next token.
They don't think. They don't reason. Math, weights, and turtles all the way down. Calling anything an LLM does "thinking" or "reasoning" is incorrect. Calling any of this "AI" is even worse.
If you have an extremely simple theory that debunks the status quo, it is safer to assume there is something wrong with your theory, than to assume you are on to something that no one else figured out.
You are implicitly assuming that no statistical model acting on next-token prediction can, conditional on context, replicate all of the outputs that a human would give. This is a provably false claim, mathematically speaking, as human output under these assumptions would satisfy the conditions of Kolmogorov existence.
However, the status quo is that "AI" doesn't exist, computers only ever do exactly what they are programmed to do, and "thinking/reasoning" wasn't on the table.
I am not the one that needs to disprove the status quo.
No, the status quo is that we really do not know. You made a claim why it is impossible for LLMs to think on the grounds that they are statistical models, so I disproved your claim.
If it really was that simple to dismiss the possibility of "AI", no one would be worried about it.
You're forgetting the power of the divine ineffable human soul, which turns fatty bags of electrolytes from statistical predictors into the holy spirit.
An LLM is very much like a CPU. It takes inputs and performs processing on them based on its working memory and previous inputs and outputs, and then produces a new output and updates its working memory. It then loops back to do the same thing again and produce more outputs.
Sure, they were evolved using criteria based on next token prediction. But you were also evolved, only using critera for higher reproduction.
So are you really thinking, or just trying to reproduce?
Humans do much more than replicate, that is one function we have of many.
What does an LLM do, other than output a weighted prediction of tokens based on its training database? Everything you can use an LLM for is a manipulation of that functionality.
It cannot scale because it doesn't have a solution for reusable components. That's why I have abandoned it. Frameworks like React solve this in a much saner way.
Reusable components are prerogative of the templating system, such as React, or Vue, or server side templates that the framework of your choice uses. Htmx works with already rendered HTML fragments from the back end and doesn't do templating on its own, so there's simply no room for it to "solve" reusable components
well now we got Web Components baked in. People don't want to give up React cause it is where many paying job is coming from.
It is the age old problem between better tech or better pay? What is even worse younger dev come in wanting to learn React cause it is the one that most job post is looking for. We end up negative economic to tech quality.
class Counter does Component {
has Int $.count = 0;
method increment is controller {
$!count++;
self
}
method HTML {
input :id("counter-$.id"),
:name("counter"), :value($!count)
}
}
I'm not trying to suggest woo here, but there has to be some mechanisms to constrain the search space somewhat.