The surprising link between political strategy and health policies
Generative artificial intelligence applications like Stable Diffusion have caused concern about the impact of AI on creative work and production. We have already witnessed the use of AI-generated images replacing artwork produced by humans.
Artists are up in arms, arguing that AI's capacity to produce such artwork is a question of theft, as AI relies on machine learning. This means that for AI to generate art, it relies upon being fed existing, human-made art. Therefore, it is, essentially, plagiarizing. It's a complex argument, but there is much to explore.
Click through the gallery to determine if AI art is a matter of creative theft or if artists are just resisting the future.
Artists are deeply concerned about the impact that AI art is having on their careers, production, and the perception of art more broadly.
Generative AI models, such as DALL-E, are cultivated on machine-based learning. What does this mean? They’re fed millions of images gathered across the internet.
Some artists are arguing that this fundamentally violates copyright, while the companies using these methods say imagery falls under fair use regulations.
So are AI models just copying artists? Well, here’s where the story gets complicated. When a person makes art inspired by another artist, it’s usually not an issue, unless it’s a direct replica.
But when AI art generates an image, it does so with very specific references to its “source material.” For example, AI art application Lensa (pictured), generates images that often have signatures in the corner, as a human artist would do to sign their work.
This imitation of artist signage is something that Lensa learned from the source images it references when generating images.
Copyright law doesn’t take AI into account as of yet. Large AI art generators are based on an the assembly of billions of images with a descriptive text accompanying them.
When this dataset is input into the generating platform, the images that are linked to the text are mixed together, which then generates a new image almost immediately.
For example, if you input “cat with a hat,” the platform will find ‘similarities’ of images of cats, hats, and cats wearing hats, and will fill in any gaps to produce an image that is ‘accurate.’
The sense of accuracy is something that AI art actually struggles with. For example, generative AI art has reportedly had consistent trouble with human anatomy.
Images of people often have more or less fingers than the average human being has. This can be considered quite trivial, but it does point to how AI art is not only different from human-produced art but has pitfalls.
Perhaps AI generated images of “cat with a hat” as a prompt most likely aren’t the key images that are in most risk of violating copyright laws.
Instead, input that specifically references an artist’s style and method is much more relevant to the discussion of whether AI art steals from human artists.
The argument driving the discussion revolves around the fact that artists do not give their consent for their art to be fed into machine learning.
A lawsuit filed by a group of artists against several large AI art companies reflects these gaps and concerns, seeking to pressure companies that are engaging in perceived theft.
The artists argue that AI art not only steals their art to generate faux-original pieces, but collects revenue from those pieces.
There is clearly an ethical question underlying these arguments, but the sustainability of artistic work is also being questioned.
Namely, why would companies or other institutions pay for artwork when they can just input data and immediately obtain a similar output inspired by the work of an artist that they appreciate?
Some might question why this is limited to AI and not extended to how humans who are inspired by other artists.
While that’s a fair question, copyright laws do address overlaps. However, these applications and platforms facilitate this process in a very direct way.
In this way, according to Miscellany News, these platforms create the condition in which artists are always competing with iterations of their former work.
This isn’t a point that’s often highlighted, but in conjunction with the issue of financial compensation, if this is adequately addressed, perhaps artists would be more comfortable with the medium.
Legal challenges have to be fought out in national courts, which offers another challenge to artists. Discussions in legislative spaces point to the lack of understanding that representatives have on this matter.
It’s clear that most legislative representatives have a difficult time grasping how a lot of new technology works to begin with. Pictured is a Senate Judiciary Subcommittee on unauthorized recreations from generative AI.
Coupled with a general environment in which artists and their work are not socially valued, the climate is favorable for exploitation.
The challenges that visual artists are facing are not so different from that of other professions that produce intellectual labor.
Writers, academics, and other professionals are arguing along similar lines. The mass extraction of data from the internet can be seen as a large casting net that picks up everything to feed into an artificial mind.
The issue for these professions is not just the ethical question regarding the extraction of data and knowledge without recognition, but the issue of who profits.
Artists and other creative laborers cannot advance their careers, or at the very least will face limitations, when competing with a much cheaper version of themselves.
Sources: (NPR) (Miscellany News) (Medium) (arXivLabs) (ladder) (PhilPapers)
See also: How to differentiate a real person from an AI-generated image
The exploitation of AI art
Artists say it's a matter of theft
LIFESTYLE Artists
Generative artificial intelligence applications like Stable Diffusion have caused concern about the impact of AI on creative work and production. We have already witnessed the use of AI-generated images replacing artwork produced by humans.
Artists are up in arms, arguing that AI's capacity to produce such artwork is a question of theft, as AI relies on machine learning. This means that for AI to generate art, it relies upon being fed existing, human-made art. Therefore, it is, essentially, plagiarizing. It's a complex argument, but there is much to explore.
Click through the gallery to determine if AI art is a matter of creative theft or if artists are just resisting the future.