As companies like OpenAI, Stability AI, and Midjourney release AI image generators that can create original images based off of simple text prompts, artists have had to face a number of both specific and existential questions: was their copyright violated when their images were mined and used to train these AI models? As the technology becomes more accessible and more sophisticated, are they at risk of losing work?
There are few clear answers at this point, at least in the US. While the UK and the European Union have released more specific guidelines around AI development, such as the Digital Single Market Directive and the proposed Artificial Intelligence Act, the US is currently lacking regulations or legislation around what is already proving to be a disruptive technology.
Listen beautiful relax classics on our Youtube channel.
Last October, the White House released the “Blueprint for an AI Bills of Rights,” a “non-binding” white paper meant to “support the development of policies and practices that protect civil rights and promote democratic values in the building, deployment, and governance of automated systems,” according to the White House Office of Science and Technology Policy (OSTP).
To draft the Blueprint, OSTP brought on numerous scholars and subject-matter experts, like Suresh Venkatasubramanian, a professor of computer science and data science at Brown University.
“I’m used to thinking about AI being used in some decision-making capacity or some decision support capacity,” Venkatasubramanian told ARTnews. “In that context, harms accrue either in the form of discrimination, bad quality decision making or lack of transparency and accountability.”
Venkatasubramanian and his colleagues focused on cases where automated systems, supported by AI technology, contributed to discrimination, such as a case in which three black men were wrongfully arrested after AI supported facial recognition software by the French company Idemia misidentified them as criminals.
The Blueprint team ultimately put forward five principles that provide guidance on the use of these automated systems. While the group was focused on decision-making AI, rather than image-generating AI like DALL-E or Stable Diffusion, their proposed principle on data-mining appears relevant to artists, many of whom have decried the use of their images to train AI models.
“You should take as little data for people as possible, you should anonymize it as much as possible. Don’t draw information from people in a way that could reveal personal things about them,” Venkatasubramanian said about the guidelines on data usage, which primarily focus on personal data, rather than commercial.
Last week, a group of artists filed a class-action lawsuit against Stability AI, Midjourney, and DeviantArt, claiming that those companies violated copyright laws by using their images, along with tens of thousands of other artists, to train their image generators and produce derivative works.
When asked about the use of images, or other potentially copyrighted materials, Venkatasubramanian sympathized with artists and speculated about the potential harms and solutions.
“If the concern is that artists are not getting credit for their work, perhaps they should be compensated for using their art,” Venkatasubramanian said. “When you produce a particular AI generated art, could it be possible to track the influences that caused it and where those influences come from and to what degree?”
Artists’ key concern — that AI systems will automate away their jobs — is not addressed in the Blueprint. Venkatasubramanian said this is a situation that likely cannot be prevented, only coped with.
“[This is a situation] that would figure into how the government thinks about worker retraining,” said Venkatasubramanian. “How do you get more people into thinking about working in STEM, how can their skills be repurposed and reused as this sort of labor transformation happens?”
This response is not surprising. A 2020 World Economic Forum report projected that 85 million jobs will be lost to automation and A.I. by 2025. However, it also projected that the technologies will generate 97 million new jobs. Similarly, last August, the Government Accountability Office said that researchers project somewhere between 9 percent and 47 percent of jobs may be automated in the future, with the hardest hit expected to be those that do “routine tasks” or have lower levels of education. Meanwhile, arguments for a universal basic income are directly linked to the knowledge that automation will mean less work will be available.
GAO’s solution? “Workers impacted by automation may need new skills to adapt to changing job requirements or get a new job,” the organization said.
It appears that those in the creative sector may surprisingly be at risk of suffering from automation due to the advent of DALL-E and other image-generating AI.
Karla Ortiz, a Puerto Rican-born illustrator whose clients include Marvel and Disney, is one of the artists in the class action lawsuit against Stability AI, Midjourney, and DeviantArt,
“The promise of technology was always that we’re going to automate menial things so you have more time to be creative, but what does this mean when you automate creativity itself? What does that do for the humans who live out of being creative?” Ortiz told ARTnews in a recent phone call.
As artists and aspiring artists face the threat of automation, creatives have begun to argue that even “repetitive” or “low skilled” creative work is worth protecting as those entry-level jobs are often how people enter the industry or earn money while they develop other skills.
“I can’t tell you how many conversations I’ve had with students who are like, ‘Should I quit? Should I do something else?’ For people who are just starting out and looking for entry level jobs–well, those are the [jobs] that are most at risk,” said Ortiz.
Listen beautiful relax classics on our Youtube channel.
Another strategy being pursued by the Concept Artists Association, which Ortiz belongs to, includes fundraising to hire a full time lobbyist to represent the interests of commercial artists in D.C.
Despite the existential questions and major economic disruptions that AI could bring, governments are hesitant to enact regulation of the sector, according to Dr. Andres Guadamuz, an intellectual property lawyer who has appeared before the UK government in hearings concerning AI.
Guadamuz characterized the development of technology as a kind of global intellectual arms race.
“I’ve talked to policymakers about this and, specifically, they are in an AI race against China,” Guadamuz told ARTnews. “They’re afraid China is going to come and eat their lunch. So everyone wants to be as AI-friendly as possible.”
Source: artnews.com