The Digital Law Innovation Society is a student-led group at the University of Alberta that plays at the intersection of technology and the law. This post is the first of two parts in our highlights from their recent conference - Myths & Reality: AI and the Law.
AI as author: A copyright law perspective
Professor James Grimmelman kicked off the DLIS conference with a question that many of us non-lawyers think about when it comes to artificial intelligence - can an AI legally author a work? We’ve all heard about super-powerful technologies like GPT-3 which are being used to write articles, ad copy, stories and other written work. There are programs that morph photographs into new digital creations. Who legally owns the finished product?
For example, I recently attended a workshop where we engaged with some AI tools to create an artwork. I used a natural language processing program called Sudowrite and gave it a prompt about FOMO to create this poem:
On FOMO Island On FOMO Island, we all know that we are missing out. And that’s OK because missing out is normal. To commit to one thing means not doing something else. To spend time with one person means not spending time with someone else. To go to one place means not going to another. And that’s OK because we all have different wants and needs and we can’t be everywhere at once.
According to Professor Grimmelman existing copyright law is clear - that (really bad) poem is 100% mine. He referenced a paper, The Death of the AI Author, which underpins the argument. It’s not about the tools used in creation because tools can’t be a creator. Only people can be creators under current copyright laws. During the Q & A one participant noted the case of a monkey taking a selfie using a digital camera. This participant astutely pointed out that people argued over whether or not the monkey might be the “author” but nobody argued for the camera. The camera was merely a tool. But is AI in the same category, just a tool? For now it seems like that answer is yes - it’s not special. The one line that stood out for me was when Professor Grimmelman said “what machines do doesn’t count”. He meant this in the context of copyright. Yet, I have to wonder if that’s true in the broader sense. Much of the work I’m doing is about the ethics of what machines do mattering or counting even if accountability is ultimately attached to humans or groups of humans in the legally enshrined entity of a corporation.
Practically speaking, when copyright is called into question there are usually economic interests at play. While I think there is no risk here that any economic or literary gains might be made for On FOMO Island, or much of what we generate with today’s level of AI systems, I do wonder what might happen if the stakes get higher. Would there be a push to reexamine, revisit or reinterpret copyright law if millions or billions are at stake? If those who owned the AI algorithm that led to a breakthrough work would so easily let that ownership go? If we as humans contributed a one word prompt and then hit a button on our keyboard and the GPT-3 backed machine wrote an entire bestselling novel, an opera, a symphony, an academic paper - is the machine still just a tool, or something more? Conversely, if we are all using powerful AI algorithms and we all have this potential, do we need copyright laws at all? Is there authorship that requires protection? Perhaps the idea of copyright itself becomes unnecessary.
Read part 2 of this article HERE >
By Katrina Ingram, CEO, Ethically Aligned AI _______
Sign up for our newsletter to have new blog posts and other updates delivered to you each month!
Ethically Aligned AI is a social enterprise aimed at helping organizations make better choices about designing and deploying technology. Find out more at ethicallyalignedai.com © 2022 Ethically Aligned AI Inc. All right reserved.