Info Blog #2: Policies for Creative Works

Some thoughts on Generative AI and the compensation of artists

Case Study
ARTificial: Why Copyright Is Not the Right Policy Tool to Deal with Generative AI

With the increase of generative Aartificial intelligence usage problems arise around the compensation of artists and how to protect artists work. So what is the solution? Michaela Mantegna argues that copyright may not be the right solution for generative artificial intelligence (GAI). She looks at the complicated relationship between generative AI and copyright law, pointing out that existing copyright frameworks are not written to handle challenges that AI-generated content pose.

These challenges include issues of authorship, originality, and the use of copyrighted materials for training AI models. The case study goes further, exploring viewpoints of both artists and society. Mantegna argues that expanding copyright law might fail to compensate artists, hurt the creative process, and overall hurt the development of AI. Mantegna instead offers a holistic solution, policy that addresses the larger ethical, labor, and economic concerns.

One concern for copyright law is its effect on the compensation of authors. With this concern comes questions such as: How should authors be compensated, and under which circumstances? This is both an ethical and legal question. Mantegna offers up the idea that if the labor of training of these AI machines were uncompensated, then the threat to creative workers and their livelihood would be less. If labor were not unethically used to train these machines, then their work cannot be mimicked. Their styles, voices, performances would remain theirs, unable to be truly replicated. Going through the legal process is expensive, complex, time-consuming, and often has unpredictable outcomes. It can hurt individual artists more. The solution of policy is much better and might look like alternative compensation or redistribution methods.

I think that a universal basic income would also be a great solution. This would not only benefit artists but the general public. In an age where so many are concerned about AI taking jobs as a whole, a universal basic income could provide stability and reassurance to those that need it most. Other solutions might look like taxation or government incentives. In a capitalistic society, there are large companies profiting off of the suffering of others. This isn’t an easy problem to solve, but shifting society’s perceptions and putting pressures on companies could also be majorly beneficial. Taxation would be a great solution to achieve that.

As Mantegna mentions, the problem begins with the training of AI. So what might it look like to prevent the problem from the beginning. This looks like asking questions about what is and isn’t ethical when it comes to the training of these machines. “Is training with unlicensed works a fair use, or an infringing one?” To answer this, the article brings up precedents like Authors Guild v. Google. In Authors Guild v. Google, it was found that if a work’s purpose is transformative, then it is fair use to create a copy. The concern here arises with the idea that generative AI might be “transforming” the original work, in which it would be legal to do so. To me, the answer to the question seems simple. Training with unlicensed works should be infringement, using work without giving proper credit. However, under law it may be considered fair game.

A follow up question that is just as important to answer, “are the outputs generated by GAI original or derivative works?” Again, under law this is open to interpretation. To me, the outputs are derivative works, simply because the only reason a machine is capable of producing the work is because of another person. If a person were to take inspiration from another person, it might be different. For me, the difference is strictly on the premise that if the machine was not trained by a certain work, it would not be able to create a piece. However, a person may be able to find inspiration from a multitude of sources. I understand the counterargument, why an output generated by Generative AI might be an original, as it has not been created previously. With that, I still understand it to be derivative strictly on the idea that it used another person’s work. I can see both sides but still don’t think it should be considered on a case-by-case basis because I think that might let too many incidents of people not getting credit when they deserve it, to fall through the cracks.

This then made me wonder, “should the outputs be entitled to some form of copyright protection? If so, how should we deal with AI authorship?” This question again does not have a definitive answer. Copyright protection requires work to be “original.” This means that if the work is not original, then it is not entitled to protection. I think that because it is not necessarily an original work, created by a real person, then it does not deserve copyright protection. While it could be argued that it is an original work, I believe there should be a distinction between artificial intelligence and human artwork. Making there a clear difference between the two allows for humans to have their work protected and that should be our priority.

I think it is incredibly important to think about how generative AI is impacting real human artists and their work. As the use of generative AI increases and becomes more ingrained into daily society, this problem will only be on the rise. Addressing the problem now, finding solutions, and making real artists the priority of our solution ensures that the hurt is limited and minimized before it spreads too far where there isn’t an available solution.