For a given calculation on given hardware, the 100th digit of a floating point decimal can be replicated every time. But that digit is basically just noise, and has no influence on the 1st digit.
In other words: There can be multiple "layers" of linked states, but that doesn't necessarily mean the lower layers "create" the higher layers, or vice versa.
I don't think he's missing the point at all. A band saw is an immutable object with a fixed, deterministic capability--in other words, a tool.
An LLM is a slot machine. You can pull keep pulling the lever, but you'll get different results every time. A slot machine is technically a machine that can produce money, but nobody would ever say it's a tool for producing money.
People keep trotting this argument out. But a band saw is not deterministic either, it can snap in the middle of a cut and destroy what you're working on. The point is, we only treat it like it's deterministic, because most of the time it's reliable enough that it just does what we want. AI technology will definitely get to the same level eventually. Clinging on to the fact that it isn't yet at that level today, is just cope, not a principled argument.
I feel like we're both in similar minds of opposite sides, so perhaps you can answer me this: How is a deterministic AI any different from a search engine?
In other words, if you and me always get the same results back for the same prompt (definition of determinism,) isn't that just really, really power hungry Google?
I'm not sure pure determinism is actually a desirable goal. I mean, if you ask the best programmer in the world the same question every day, you're likely to eventually get a new answer at some point. But if you ask him, or I ask him, hopefully he gives the same good answer, to us both. In any case, he's not just a power hungry Google, because he can contextualize our question, and understand us when we ask in very obscured ways; maybe without us even understanding what we're actually looking for.
For every real valued function and every epsilon greater than zero, there’s a neural network (size unbounded) which approximates the function with precision epsilon.
It sounds impressive and, as I understand it, is the basis for the argument that algorithms based on NN’s such as LLM’s will be able to put perform humans at tasks such as programming.
But this theorem contains an ambiguous term that makes it less impressive when you remove it.
Which for me, makes such tools… interesting I guess for some applications but it’s not nearly as impressive as to remove the need for programmers or to replace their labour entirely with automation that we need to concern ourselves with writing markdown files and wasting tokens asking the algorithm to try again.
So this whole argument that, “you better learn to use them or be displaced in the labour market,” is a relying on a weak argument.
I think the distinction without a difference is a tool being deterministic or not. Fundamentally, its nature doesn't matter, if in actual practice it outperforms everything else.
Be that as it may, moving the goalpost aside. For me personally this fundamentally does matter. Programming is about giving instructions for a machine (or something mechanical) to follow. It matters a great deal to me that the machine reliably follows the instructions I give it. And compiler authors of the past have gone to great lengths to make their compilers produce robust (meaning deterministic) output, as have language authors tried to make their standards as rigorous (meaning minimize undefined behavior) as possible.
And for that matter, going back to the band saw analogy, a measure of a quality of a great band saw is, in fact, that the blade won’t snap in half in the middle of a cut. If a band saw manufacturer produces a band saw with a really low binomial p-value (meaning it is less deterministic/more stochastic) that is a pretty lousy band saw, and good carpenters will know to stay away from that brand of band saws.
To me this paints a picture of a distinction that does indeed have a difference. A pretty important difference for that matter.
I've never understood this train of thought. When working in teams and for clients, people always have questions about what we have created. "Why did you choose to implement it like this?" "How does this work?" "Is X possible to do within our timeframe/budget?"
If you become just a manager, you don't have answers to these questions. You can just ask the AI agent for the answer, but at that point, what value are you actually providing to the whole process?
And what happens when, inevitably, the agent responds to your question with "You're absolutely right, I didn't consider that possibility! Let's redo the entire project to account for this?" How do you communicate that to your peers or clients?
It would not be shocking at all if in 10 years, "Let's redo the entire project to account for this" is exactly how things work.
Or lets make 3 or 4 versions of the project and see what one the customer likes best.
Or each decision point of the customer becomes multiple iterations of the project, with each time the project starting from scratch.
Of course, at some point there might not be a customer in this context. The "customer" that can't handle this internally might no longer be a viable business.
"You're absolutely right" feels so summer 2025 to me.
A key difference is that each of the mediums you mentioned are deterministic and unbiased (to a certain degree.) The the work created can therefore be inferred to be a "pure" expression of the artists intent. A pro photographer and my mom will get wildly different results even with the same equipment. Not so with AI, which very much has it's own bias and is eager to inject it.
The other question is, is AI a tool or a medium? I often hear people say "Well EDM was looked down on when it first came out," but EDM is not a tool, it's a genre. I think most artists wouldn't really care about "AI" becoming a genre of art, but it's silly to think that all future art will be AI just as it would have been silly to think EDM would have replaced all future music.
> A pro photographer and my mom will get wildly different results even with the same equipment. Not so with AI, which very much has it's own bias and is eager to inject it.
That particular AI models have their own bias and are eager to inject it is among the reasons why a skilled user and an unskilled user will have very different results, not a reason why that isn’t true.
> The other question is, is AI a tool or a medium?
Is oil paint on canvas a set of tools or a medium? In art, a tool ot set of tools often characterizes, or even defines, a medium; they are different but not orthogonal concerns. (And the cultural phenomenon of identification of a regularly-used tool or combination of tools as defining a medium generally only happens well after that tool or combination has been in significant use for a while.)
AI is a broad category of tools. Particular combinations of those (either with eachother or with other tools) may also come to be be understood as particular media.
>That particular AI models have their own bias and are eager to inject it is among the reasons why a skilled user and an unskilled user will have very different results, not a reason why that isn’t true.
Not quite what I mean. If you and I both take a photo of the same controlled scene with the same camera, the result will be essentially identical. If you and I both type the exact same prompt into Nano Banana, we will both get very different images. So, how is one supposed to know what parts of the AI image are intentional or incidental? If the AI image is "good," is it good because of or despite the prompter?
>Is oil paint on canvas a set of tools or a medium? In art, a tool ot set of tools often characterizes, or even defines, a medium
Agreed, and this is basically what I'm saying. I'm fine with siloing AI art into it's own category and I'm sure some cool work can be done there. But it's fundamentally odd to think that AI will, for some reason, replace or displace other art.
All due respect to your mother, but a pro photographer would certainly achieve better results. Your mom may recognize something is not right but be unable to articulate it clearly to the tool. Same problem that's always been. The bar has been lowered, not removed.
People will argue that if a shortform video is human made or AI generated it doesn't matter, it's domamine-triggering filler either way.
But I do think that the parasocial relationships and discovering new influencers is a big part of the hook for many people, and taking that away may cause many to have a "what the hell am I even watching" moment.
It's easier to justify the addiction when it feels like you're "hanging out with a friend." When content is AI generated from concept to production, it's just...talking pixel soup.
While I appreciate your positive experience, if you were unable to write, storyboard, and communicate your idea before AI--than AI is not going to be "the thing" that suddenly turns you into a filmmaker.
You're writing anyway, you just call it prompting. If you feel that the AI is giving you back more than what you're putting in with that writing, you have to ask yourself who's vision the result really is.
>if you were unable to write, storyboard, and communicate your idea before AI--than AI is not going to be "the thing" that suddenly turns you into a filmmaker.
The person you are replying to doesn’t say that they couldn’t do these aspects before. In fact they mention that they are able to communicate EXACTLY what they want to AI. The only problem was that their idea is high concept.
Getting a high concept film made is one of the most privileged positions in the world. Practically no one can actually just do this. Even people in the film industry with family connections to producers struggle to get their films made.
You have no idea what you’re asking from people when you tell them they should have just made it.
> You're writing anyway, you just call it prompting. If you feel that the AI is giving you back more than what you're putting in with that writing, you have to ask yourself who's vision the result really is.
They didn’t say that, they said they were getting exactly what they were seeing in their head.
Animating is a ton of work. Animation decisions can give a great deal of "feel" to a cartoon. Studio Ghibli works are distinct from Disney's old animation and each conveys a distinct feeling. Those decisions are often made by the production team. Its why One Punch Man season 3 is so derided this year after Season 1 was heralded as the greatest season of Anime in a very long time. You'd think following the literal story boards provided by the already published Manga would be simple! but somehow it lost all of its charm.
Animation, or shooting video is the result of many, many pieces. If you're not putting a ton of work into the decisions that result in the final "shot" the video camera makes then the video that is shot isn't magically going to show an enormous amount of care in its construction.
A video of an ocean (or anything) is not, inherently, art, or creative. But also...if you're taking a video of the ocean, it's probably because you want to capture/share a video of the ocean, which an AI generated video is not.
Unless you're making a very particular type of video, you likely want to capture aspects of the water's movement, color, sound, or interactions without much care about faithfully capturing a video of the literal ocean. The former part is a big portion of what turns a video of the ocean from not inherently being art to being as much of art as you make it so. Saying the only reason would be to share the literal ocean forces the art out, not in.
In other words: There can be multiple "layers" of linked states, but that doesn't necessarily mean the lower layers "create" the higher layers, or vice versa.
reply