somniatorsound.bsky.social
Film composer, audio geek and hobbyist writer. Director of "A Real Composer". PhD researcher.
89 posts
52 followers
91 following
Getting Started
Active Commenter
comment in response to
post
So the real april fools would be that they *dont* come?
comment in response to
post
To be fair, its good when it provides links to sources and you ignore the output text.
comment in response to
post
So the dataset could then be interpreted as the art world and the AI the creator.
So according to him, an AI is capable of being an artist, or possessing artistry.
comment in response to
post
For Becker, creator is the one who "made" the art piece and the art world is everything that influenced it. Is the AI the creator then?
In short, yes. For Becker, an art creator is someone, who creates art within the constraints of an art world - or in this case, the dataset.
comment in response to
post
infinite splitting of tasks - just within what is possible for it. Imagine humans collaborating on a painting, where every painter on the planet participates.
According to Becker, this is all artistry in the same way - there isnt much difference. Except the aspect of a creator.
comment in response to
post
Surely there must be limits somewhere?
It seems to be random. For a commercial AI, we do not know which art pieces the AI uses specifically to create your output, but there is a finite amount of them as an infinite dataset is impossible.
But the AI system gets close to what couldnbe defined as
comment in response to
post
I would argue that the dataset is already the influence, meaning "no influence" is also impossible for an AI.
And the opposite? This is where it gets interesting.
An AI will create art, pixel by pixel, by reading the dataset. Is it therefore splitting each possible task infinitely?
comment in response to
post
If we close this dataset at a point and create a closed system, is the AI then outside the definition of external influence? Like a human brain in a jar without any outside interference.
According to Becker, this would be impossible for a human, but an AI is not human. Does the theory apply here?
comment in response to
post
performes by a separate entity. Work cannot be defined by infinitely splitting it into smaller tasks. Therefore, artistry must be somewhere between these two.
But how does AI come into play?
Art generated by an AI is always based on a dataset - for example existing art.
comment in response to
post
And that art cannot be created in a vacuum.
The tools you use, the ideas you have, all come from somewhere. And no art piece is truly "singular", but takes an "art world" to create it.
This however translates to the opposite too. No art can be created so that each possible different task is
comment in response to
post
AI's also lack continuity. When writing page 3, it has already forgotten what was on page 1. A fully generated novel would be a weird jamble with the current tech (still).
comment in response to
post
Most AI's dont want to allocate enough memory for a consumer to create lenghty things. The longer the text is, the larger computing power it needs, and if a million people would generate novels at the same time, their servers would melt.
comment in response to
post
Length is kind of directly tied to the "risk" of the film. The longer it is, the more confident the studio is of it.
comment in response to
post
Public pages still own the copyright of any content posted on them and need to agree to the scraping. OpenAI scraping twitter requires twitter to agree to it.
There is no "free for all" content, especially in Europe.
comment in response to
post
Those are all very real issues that cannot be overlooked. Training data is currently illegal in many countries and there is no excuse for it. As much as humans have to adapt to AI, AI companies have to adapt to buying licenses for their data.
Job loss is inevitable, but most of it temporary.
comment in response to
post
What if this happens with films? A new unknown composer who scores the next blockbuster. No social media. No contact information. Just a soundtrack with a name.
comment in response to
post
This will become increasingly prominent in "low art" music, as I defined in an earlier thread. People are not really interested who is behind some stock music, so it might as well be AI, and the music user never finds out.
We are living at an i teresting break, where anything could be AI made.
comment in response to
post
Imagine the next Hollywood blockbuster credits a completely new name as the composer. An unknown person, with social media profiles and everything. How can you tell, if this composer really exists, or is just a front for an AI?
The film studio will want to actively hide that the composer was an AI.
comment in response to
post
Since AI is seemingly here to stay, in a business sense, it is in the companies' interest to hide the fact that they use AI to avoid PR nightmares and people boycotting them.
So the future holds even more fake artists, who are actually machines.
comment in response to
post
The story: harpers.org/archive/2025...
This raised the question, how does AI play a psrt in this?
According to research, people don't like AI created music. It creates controversy and negative views, and causes major problems for companies doing it.
comment in response to
post
Yea, it has software, recording room, instruments, mics, sound systems and everything. Anyone can book it for whatever they want.
comment in response to
post
In Finland, libraries are quite a lot more than just books. The Helsinki main library has recording studios, video haming rooms and 3D printers for anyone to use.
comment in response to
post
Tekoälyllä tuotettu sisältö aiheuttaa niin paljon negatiivisia vastareaktioita, että se tuskin on kestävä bisnesmalli.
comment in response to
post
Different issues are prioritized differently depending on people and ideologies.
"Overall, the combination of different perspectives is important because AI, unlike most other technologies, can affect people in different ways, roles, and constellations"
doi.org/10.1007/s001...
comment in response to
post
There are no conclusive or comprehensive studies made on how around the world different political views align with the use of AI. These are usually split among various different single factors, such as environment, accessibility, new tools, financial opportunities and so on.
comment in response to
post
"...there were nearly equal levels of acceptance between both Republicans and Democrats..."
escholarship.org/uc/item/1tn9...
comment in response to
post
First finding is that based on a study made in the US, there really is no difference on how different political views (left-right) like generative AI.
There are differences, but they are often related to how the AI is used and regulated, but the general attitude is very similar among everyone.
comment in response to
post
And a lighter note to the end, the poor quality "AI slop" content has an official Wikipedia page now.
en.m.wikipedia.org/wiki/AI_slop
comment in response to
post
On a final note: "the rise of generative AI may mean that artists will have to deliberately rely on traditional, artisanal processes to differentiate themselves from low art producers."
Computers in Human Behavior: Artificial Humans
Uwe Messer
doi.org/10.1016/j.ch...
comment in response to
post
A keyword here is the perceived effort. Especially in high art, the effort that a viewer perceives that has gone into the creation of the art plays an important role. (The more effort an art has, the more highly it is valued).
This applies to all art in general, but the bias is more visible with AI
comment in response to
post
"Those who create illustrations and stock images (produce art-like craft or low art) suffer less from the use of AI during creation than those who intend to create high art. Creative control and intentions interact to influence the evaluation of the product through authenticity and perceived effort"
comment in response to
post
Art can be classified as "high art" and "low art", where high art refers art as something you do for art's sake, and low art being something that is designed to serve a specific purpose.
According to the same study, the use of AI in low art faces a lot smaller backlash than high art.
comment in response to
post
AI art is also seen more authentic, if the art is aware of the AI involvement.
"...communicating a vision (e.g., to see nature through the eyes of the machine) is perceived by viewers as even more authentic."
comment in response to
post
But are these results universally true? Most of these studies are conducted within the early days of generative AI (on a larger scale). These results do not give an answer on how these perceptions change once AI becomes more normalized in art creation. However, at the moment, people don't like AI.
comment in response to
post
Any involvement of AI in an art creation process creates a more negative perception of the art. But using AI for ideating and other "non-visible" processes, the negative reaction is smaller
"..Using AI during ideation ... has a less negative impact on evaluation than using AI during implementation"
comment in response to
post
The negative perception does not stem from art quality, but rather from the fact that it is made with/by an AI.
"...when viewers are informed that a visual artwork was created with the assistance of AI, they tend to evaluate it less favorably... – even when the artifact is objectively identical."
comment in response to
post
"This upgrade not only enhances the capabilities of AI in video production but also underscores the potential of AI to transform creative workflows, making AI an integral part of the creative process rather than just a tool for execution"
medium.com/@macaipiotr/...
comment in response to
post
"Each image is essentially a stop on a journey, and users can connect these stops to create smooth transitions between them. In other words, it’s like telling a story where each part flows into the next."
www.tomsguide.com/ai/ai-image-...
comment in response to
post
This implies a completely new approach to editing, but lacks any kind of connection to real footage and remains very gimmicky as a technology.
For work roles, this means replacing the entire production pipeline to a single person who has full artistic control over the product.
comment in response to
post
This means adjusting and editing the video content while simultaneously designing the story direction.
So goodbye screenwriters?
Nonlinear editing is currently based entirely on generating video, and does not work with existing/real footage.
comment in response to
post
How does it work?
In this editing, you have two pictures and the AI generates the video between them. A video between these pictures can be referred to as a timeline.
You can create several pictures and make timelines between any two images, and advance the creation towards any image nonlinearily