nipah

Okay, doubt. What level of complexity you believe this project has? Including the changes that required changing burn-cubecl.

https://github.com/mii-nipah/voxcpm-rs

--- Just to be clear, I'm not saying they don't make mistakes. In fact I constantly scream into the void with the sheer amount of absolute stupidity of those models, however I would never say, using them for what I use, that they can only be used for simple and small use cases.

"rewrite 50% of what it generates" See, I'll not claim they write good code. But have you considered maybe your standards are a little bit too high for the tool? I made like 15 tools already using AI for my use, most of them I barely needed to touch in the code. The code is not great, no, but it's not useless and that's what matter for me. You try and iteratively ask for the AI to do things. If you want to ensure a higher degree of quality you can ask for tests and use techniques such as mutation testing to increase coverage, etc.

If you expect the same level of quality as you would write by hand, then you probably is better off... not using those tools. I mean if I was rewriting 50% of the generations I get I would not be using them at all.

The second test scores 60%, the first was way higher. And I specifically said ""unless you are saying "I could not grasp it immediately but later I was able to after understanding the point" I think you and your friends should see a neurologist"", to which this person did not responded. I saw the tests, solved some, I suspect the variability here is more a question of methodology than an inherent problem for those people. I also never stated that my point depended on those people scoring 100% specifically on the tests, even if it is in fact extremely easy (and it is, the objective of this test is to literally make tests that most humans could easily beat but that would be hard for an AI) variability will still exist and people with different perceptions would skew the results, this is expected. "Significantly misrepresenting the numbers" is also a stretch, I only mentioned the numbers ONE time in my point, most of it was about that inherent nature (or at least, the intended nature) of the tests.

So on the edge, if he was not able to understand them at all, and this was not just a problem of grasping the problem, my point was that this would possibly indicate a neurological problem, or developmental, due to the nature of them. It's not a question of "you need to get all of them right", his point was that he was unable to understand them at all, that it confused them to an understanding level.

Your view is incomplete. It did lowered lifespans, but also increased them MUCH MORE than before. It was not "oh, it got worse, now it is recovering". It was "it got worse, now it is much better than before". The amount of infant deaths it lowered, for example, is massive.

And it is true, those people did not got an iPhone and died, but this is also you saying this for them. You don't know all the specifics of history or all their motivations, the industrial revolution had a bloody story, but it's origins were also organic, it also had aspects of improvement. The world population grew almost 10x.

I don't think we are in a position to judge those past events to the lens you are posing.

If you should never say never, you are already contradicting yourself.

You don't validate an hypothesis without testing counterfactuals, tho.

Did you wrote it from scratch to compare? There's an old motto devs use all the time, you know. Measure first, don't guess. How do you know it would not have took you the same time or less to write the program if it was you? Or if for example, if you were using the AI to write the boilerplate for you while you focused on the core of coding? Or using it as a tab completor assistant instead of it being an agent? Saying it saved you time is easy when you don't have the data to back it up, it's easier than thinking that maybe, maybe this was not that good of an use of your time.

700 lines of code is 2 weeks of work for a good developer? My friend, I wrote 350 lines of executable code (excluding boilerplate) in a morning (4AM to like 9AM, maybe a bit more) to make a test with voxel octrees like yesterday. There's no reason it would take "2 weeks of work for a good developer" to write 700. What takes times in those projects is the research, if you already have this fresh in your head it should not take more than 3 days to make something very simple but reasonable, and a week at max to make something good (not perfect, but good).

This will only get better with time.

Prove it.

Your estimation maybe right, but maybe also there is a point on why it is right: https://neilmadden.blog/2025/06/06/a-look-at-cloudflares-ai-...

Maybe because (and I'm quoting that article) it is still lacking in what it should have that you managed to accomplish this task in "few days" instead of "a few weeks, maybe months".

Maybe the bottleneck was not your typing speed, but the [specific knowledge] to build that system. Because if you know something well enough, you can build it way faster, like rebuilding something from scratch, you will be faster as you already know the paths. In which case, my question would be: would not be writing this as fast, or maybe at least more secure and reasonable, if you had the complete knowledge of the system first.

Because contrary to LLMs, humans can actually improve and learn when they do things, and they don't whey they don't do things. Not knowing the code to the full extent is worth the time "gained" by using the LLM to write it?

I think it's very hard to estimate those other aspects of the thing.

Novelness is not a characteristic of interpolation, tho, it's about extrapolation. If you have plenty of clients and plenty of related stuff to the provider side, even if on on auth, then it could be considerably trivial for the LLM to interpolate on that field.

quoting my own previous response: > Also, I mentioned mammals because those problems are of an order that mammals and even other animals would need to solve in reality for a diversity of cases. I'm not saying that they would literally be able to take the test and solve it, nor to understand this is a test, but that they would need to solve problems of similar nature in reality. Naturally this point has it's own limits, but it's not easily discarded as you tried to do.

---

Have you seen them or shown them to average people? I’m sure the people who write them understand them but if you show these problems to average people in the street they are completely clueless.

I can show them to people on my family, I'll do it today and come back with the answer, it's the best way of testing that out.

No, I think I saw the graphs on someone's channel, but maybe I misinterpreted the results. But to be fair, my point never depended on 100% of the participants being right 100% of the questions, there are innumerous factors that could affect your performance on those tests, including the pressure. The AI also had access to lenient conventions, so it should be "fair" in this sense.

Either way, there's something fishy about this presentation, it says: "ARC-AGI-1 WAS EASILY BRUTE-FORCIBLE", but when o3 initially "solved" most of it the co-founder or ARC-PRIZE said: "Despite the significant cost per task, these numbers aren't just the result of applying brute force compute to the benchmark. OpenAI's new o3 model represents a significant leap forward in AI's ability to adapt to novel tasks. This is not merely incremental improvement, but a genuine breakthrough, marking a qualitative shift in AI capabilities compared to the prior limitations of LLMs. o3 is a system capable of adapting to tasks it has never encountered before, arguably approaching human-level performance in the ARC-AGI domain.", he was saying confidently that it would not be a result of brute-forcing the problems. And it was not the first time, "ARC-AGI-1 consists of 800 puzzle-like tasks, designed as grid-based visual reasoning problems. These tasks, trivial for humans but challenging for machines, typically provide only a small number of example input-output pairs (usually around three). This requires the test taker (human or AI) to deduce underlying rules through abstraction, inference, and prior knowledge rather than brute-force or extensive training."

Now they are saying ARC-AGI-2 is not bruteforcible, what is happening there? They didn't provided any reasoning for why one was bruteforcible and the other not, nor how they are so sure about that. They "recognized" that it could be brute-forced before, but in a way less expressive manner, by explicitly stating it would need "unlimited resources and time" to solve. And they are using the non-bruteforceability in this presentation as a point for it.

--- Also, I mentioned mammals because those problems are of an order that mammals and even other animals would need to solve in reality for a diversity of cases. I'm not saying that they would literally be able to take the test and solve it, nor to understand this is a test, but that they would need to solve problems of similar nature in reality. Naturally this point has it's own limits, but it's not easily discarded as you tried to do.

You read at the same speed line-by-line your code when you are in your git client?

You are doing something wrong. I go line-by-line through my code like 7x faster than I would do it for someone's else code, because I know what I wrote, my own intentions, my flow of coding and all of those details. I can just look at it en passant, while with AI code I need to carefully review every single detail and the connection between them to approve it.

It's not just time spent typing. Figuring out what needs to be typed can be both draining and time consuming. It's often (but not always) much easier to review someone else's solution to the problem than it is to solve it from scratch on your own.

This is EXTREMELY false. When you write the code you [remember] it, it's fresh in your head, you [know] what it is doing and exactly what it's supposed to do. This is why debugging a codebase you didn't wrote is harder than one you wrote, if a bug happens you know exactly the spots it could be happening at and you can easily go and check them.

Now you are just being silly with your comparisons. There is no analogy between those things: * the difference between handwritting a book and typing is the extreme pain you would feel in your hands versus being able to write more in the same time without it * the difference between typing and using your voice could be of a similar magnitude for someone with problems in their hands * the difference between any of those writing methods and using an AI to do it for you, is that you are abstracting YOURSELF from the equation, not the method of writing. It's not analogous, not even from a mountain of distance far. You are not less "bottlenecked" because you don't need to write the thing yourself, you are just not producing it at all, it's more analogous to you guiding the hands of another person with vague instructions, using of their own expressivity to make your book for you, then claiming it was you who wrote it. It's not a bottleneck question, it was never a bottleneck question, and this is the case because code IS the writing, it IS the problem solving area where you need to put your mind to work, not writing a prompt, but coding in a specific and well defined formal syntax.

Amazing, because I realized I procrastinate MORE when using LLM to write code which I know I could write. And not only that, I feel I'm losing the ability to do the coding myself and solve the problems myself when delegating this to the AI. Which is why no one should base their own decisions for life, like using or not using an LLM, on some random story from the internet.

I have a very simple question with like, 5 lines at best, that basically no model, neither reasoning or simpler could grasp. For obvious reasons I'm not disclosing it here (because I fear data contamination in the long run), but it basically breaks the "reasoning" of those things. Unfortunately, I still can't try the o3-pro because the API version is not easily available, and I'm certainly not willing to pay for it in pro mode, but when it comes to the plus version (if it comes) I'll try. To this date, because of this question (and similar ones) I stand very unimpressed with those models, the marketing is a thousand times larger than reality, and I suspect people in general are surprisingly less capable of detecting intelligence than they think.

The normal o3 also managed to break 3 isolated installations of linux I was trying it with, a few days ago. The task was very simple, simply setup ubuntu with btrfs, timeshift and grub-btrfs and it managed to fail every single time (even when searching the web), so it was not impressive either.

"most people I show them too have issues understanding them, and in fact I had issues understanding them" ??? those benchmarks are so extremely simple they have basically 100% human approval rates, unless you are saying "I could not grasp it immediately but later I was able to after understanding the point" I think you and your friends should see a neurologist. And I'm not mocking you, I mean seriously, those are tasks extremely basic for any human brain and even some other mammals to do.

I don't think two wrongs make a right, tho. Looking down at formal education is not the same as embracing anti-intellectualism. And while I admit the software industry is absolutely full of bullshit anti-intellectualism, I also don't believe formal education is or should be the standard for the education, formalism is not automatically better than alternative means of learning.

The problem of anti-intellectualism in SE is just the consequence of the field being more "democratized". Or, to put in other words, the mass is stupid and the mass-man is stupidier and primitive.

I'm sorry, 2000 USD per month is MUCH more costly than an engineer from a third world country, it can basically pay for a senior where I live. Even 200 USD is sufficient for an intern here. The problem with your point is that it's not counting on the fact that this work can be done all over the world.

What’s clear is this tech is here now, and complaining about it feels a bit like mourning the loss of punch cards when terminals showed up.

Just stop with this, it's bulshitty. There's nothing related between LLMs and the migration from punch cards to terminals, nor to photoshop compared to film theatre, literally [nothing]. This is a pretty underwelming way of trying to say people that are critiques of this are akin to nostalgic people that "miss the old good days", when there are more than enough pertinent reasons to disagree with this tech in this case. Basically calling opposing people irrational.

I'm not talking about doom or software dev dying or any bullshit like that, I'm just saying this kind of point you make in the end is not reasonable.

We have some hints to be skeptical on the prospect, I would say.

Like the fact we can't even explain an everyday phenomena all humans experience like consciousness. If you don't understand a common phenomena that (supposedly) occurs in the brain, you have a fundamental lack of knowledge on how the brain itself works, and thus simulating it would not be a thing possible to do.

Another hint is the fact that we can't currently even simulate a single cell because of how much complex they are, and the most advanced neuron models (like hodgkin-huxley) are still gross simplifications of how we think a real neuron works. We don't have any proof that this is a possible thing, it is something people believe it is the case, but pretty much could be a dead-end.

Other is that we don't have a way to reliably know the state of the alive brains without modifying them, so reproducing one with fidelity appears to be very hard or maybe even impossible.

And so on.

Or that it is, like most people working in projects like those believe without first proving.

Sabine believes in superdeterminism, so it is her own position, not an universal one, nor a very popular one amongst physicists. It is VERY hard to say quantum physics is deterministic. And the studies you are citing are famous for being extremely restrictive, always made in closed-bounds and drawing unclear conclusions, if this proved our decisions originated 7 seconds before our action action of consciousness it would be REVOLUTIONARY, but it was not, it was defended as a determinism enabling position by many philosophers*, and was scrutinized by many others. For example, by having 2 options only (the study mentions that) you already increase the probability of your machine having consonance with the persons previous decisions (you can, actually, decide for X and then just after decide for Y, even minutes before you actually do the thing, being consistent with your own internal decisions is not being "determined", and the fact that this is only 60% matching is a good indicator there's something more going on here).

Another plausible interpretation of the case is that our brain does have a predisposed intuition building about something, and so it is not a [surprise] that the person chose whichever his intuitions perceives as better or more desirable, it also explains why the 60% correctness and disproves this as being evidence for some "determinism". Nobody is saying, for example, that free will equates being unable to be influenced by something, it is not a surprise that someone that has a vice in crack is craving for crack and in "70% of the cases" (or whatever) decides for using more crack instead of not using. Or that, if I desire to eat X, and X is available, and I'm planning to do it, I will eventually do it, it is expected that my actions are in good relation with my previous intents, desires and knowledge. The same way, it would be impossible for those experiments to predict something before showing the subjects what is being tested, and people are known to acting very differently (like, trying to outsmart the scientists or show themselves as better than they are) when tested, so the own thought of "I need to chose X" could be in their minds way before they [state] their conscious decision (because, if I like chocolate ice cream, even if I stare to the menu for 30 seconds I will still probably chose chocolate ice cream, but sometimes I could chose mint, it is not a surprise that decisions follow some kind of pattern when it is reasonable to expect so). The question is always about to what extent this is a [determinant], and the free will defense merely needs to say "not 100%".

It is also extremely unrealistic and out of this world to think our decisions would be able to be "predicted seconds before they happen", if the decision was conditioned to specific reaction events this would be IMPOSSIBLE, your brain cannot decide 7 seconds ahead of time what is the correct decision for a problem that you have only 3 seconds to decide (like "press the green color" when it appears or "type the word being shown in screen when you see it"), no kind of predictability would realistically arise from this kind of behavior (because it is physically impossible) rather than "the brain is preparing to type".

Except that the brain is not predictable at all, this is one of the first things you would know. You can say the definition of free will is not "consistent", but my problem is deeper and more reasonable: People just assume for no reason they can talk about free will, while they can't talk about conscience. As free will is a property of being conscious, people should shut up about free will before they can explain both causally and ontologically the consciousness phenomena. But I'm afraid they will not.

Why there is nothing rather than nothing?

This is, I fear to tell you, IMPOSSIBLE to explain by other than methaphysics, it is an uncloseable "gap" that humanity will never solve. As far as I know, it's not about gaps, it's more about the unreasonable stance that everything is explainable by reductionist physics that corrupts the modern thinking, it is more safe for you to assume people's religious beliefs come from "gaps" rather from their own rationality. In reality, you just turned science into your own religion.

This is a strawman, but it also could be the case of it being something supernatural. You don't have the skill to enter in a serious discussion about that either, so calling it "magic" is the unique thing will help you here (as people are usually scared of things they can't explain).

But it would be the case that the universe is non-computable, and if this is the case, even without "magic", you would not be able to reproduce certain kinds of things (even not conscious ones) with a computer.

This is just partially a solution, because this assumes cognition as a whole is computable.

If you don't believe in God, actually you NEED to buy this, or buy into some other methaphysical eternalism that is as hard to prove as truly random events. The universe either started randomly or it was created, and the existence either existed eternally or it was created (both prospects are also problematic in some level, but at least it is something better than thinking you can not have a position on this and also assume that determinism is trivially true).