Fourth Amendment scholar Professor Matthew Tokson has taught at Utah Law since 2017, focusing on law and technology and originally teaching a cyberlaw seminar. In January, he began teaching a new seminar about what has become a new research interest: artificial intelligence (AI).
“AI has been an interest for several years, and it was time to update the cyberlaw seminar. I wanted to create a new seminar that’s focused more concretely on artificial intelligence,” Tokson explains. “The class brought together my scholarship interests and my interest in teaching what is going to be a huge growth area in terms of regulation and legal issues.”
Launching a brand-new seminar about AI
He notes in the Law of Artificial Intelligence class syllabus that he will approach AI from a lawyer’s perspective, not a technical perspective, with the goal of understanding regulatory issues surrounding AI. Students will cover the following topics:
- potential tort frameworks applicable to autonomous AI agents and devices
- risks of bias and discrimination from algorithmic decision-making
- implications of AI development and production for intellectual property law
- regulating autonomous weapons systems
- addressing catastrophic and existential risks
To touch on one aspect of the class—discrimination in algorithmic decision-making—Tokson points out that algorithms use data from the past that is inflected with bias.
“The algorithms can then apply that forward to make financial and health decisions. That’s been a big concern in the scholarship of AI and the law,” he says. “I want students to understand some of the ways AI is changing the law or raising unique—and in many cases, complex—legal issues and be a step ahead of the rest of the legal profession in being familiar with some of those issues.”
Helping students hone their writing skills and learn about AI
A significant portion of a student’s grade (70%) in Tokson’s class is based on a substantial research paper of at least 7,000 words about an area of law affected by artificial intelligence—and as in other classes at the University of Utah, students cannot use AI tools to write it.
“At one point, I was grappling with what extent to allow students to use AI in crafting a paper, but the University of Utah’s policy does not allow it and the decision was made for me,” Tokson recalls. “I tell students that I know it’s ironic that you can’t use AI in this class, even though we’re going to learn about the various chat bots. We rely on the honor system, because the systems are advanced enough now that they don’t leave obvious signs of AI authorship.”
Students have already submitted paper proposals with a thesis and brief summary, and Tokson is happy with the breadth of topics.
“One student is working on military applications and the laws of war and how those apply to artificial intelligence. Another is doing anti-discrimination law and algorithmic decision-making and how those two interact, and another is writing about surveillance. We have a nice variety,” he says.
Nico Micheletti, a 3L currently enrolled in Tokson’s seminar, says the most exciting part of the class is exploring a rapidly evolving field with profound implications.
“The capabilities of modern (and coming) AI systems are truly astonishing, and spending two hours each week discussing how this power will shape the future—especially from a legal perspective—is a damn good time,” Micheletti says. “For my paper, I’m examining how First Amendment protections intersect with deepfake images, particularly in cases where the subject is falsely portrayed in an offensive manner.”
Zach Berger, a visiting 3L from the University of Texas at Austin School of Law, says he has become really interested in AI law and was excited to see that Utah Law offered this course.
“I’ve done directed research on AI and copyright before, but Professor Tokson’s class provides a rare opportunity to engage more comprehensively with a really sprawling, nascent, and important area of the law. It’s such a new technology with such massive potential implications for society, so I love being able to think about its legal implications, particularly when the state of AI law is so unsettled and indeterminate,” he says. “Professor Tokson is engaging and funny, and he facilitates class discussions so well. I feel totally out of my depth when it comes to things like the intersection between AI and the law of war, but Professor Tokson and the other students have created an environment where you can feel completely comfortable contributing and wading into unfamiliar legal territory.”
Berger is writing a survey of copyright cases involving generative AI for his seminar paper.
“Generative AI models are often trained on billions of copyrighted works, and sometimes, their outputs include reproductions of copyrighted materials. Those two facts, along with the possibility that generative AI could economically displace scores of artists and creators, have led to dozens of copyright lawsuits against generative AI companies,” Berger explains. “The way courts deal with these early generative AI cases will have massive impacts on the development of copyright law and generative AI technology more broadly. And historically, technologically complex cases like these have been challenging for courts to deal with. I’m really excited to be able to dive deeper into this issue through Professor Tokson’s course.”
Focusing on big-picture AI concerns
Along with Professor Yonathan Arbel of the University of Alabama School of Law and Professor Albert Lin of the UC Davis School of Law, Tokson recently authored a paper about the systemic regulation of artificial intelligence, which is “among the first works of legal scholarship to address the AI alignment problem and the global risks of failing to ensure that AIs are aligned with broad social interests” according to the abstract.
“I worry about artificial intelligence as it gets capable of doing things beyond human capability. When you are competing for resources with a more powerful, capable, intelligent ‘species,’ that traditionally hasn’t worked out well for other species,” Tokson says. “Humanity is a great example of that, where we dominate Earth’s ecosystems and there’s no one who can push us around and tell us not to. Other species, especially those competing for the same ecosystems, do really poorly. That is the future concern.”
While intellectual property lawyers are most likely to begin specializing in artificial intelligence cases, Tokson says that AI will affect almost every area of law.
“Self-driving cars are already available in a select group of cities, and while I think they will decrease car accidents, there will still be accidents—so a personal injury attorney, for example, might find AI law relevant,” he explains. “Classic areas of law are going to be affected by corporate law and liability. There are algorithmic contracts, so any kind of lawyer can benefit from knowing about the issues AI creates.”
Preparing further AI research
A new paper about AI surveillance and the Fourth Amendment is currently in the works, Tokson says, outlining how the Fourth Amendment might deal with the pro-authoritarian effects of automated surveillance systems.
“I’m interested in the privacy and surveillance aspect of AI: facial recognition, AI analytics of surveillance cameras, and automated law enforcement technologies. These are surveillance systems that only need a few people to manage but that can allow you to exert power over a large jurisdiction. I’m writing about how the Fourth Amendment might adjust to prevent some of the effects of those kinds of systems. That’s been really fun to work on,” he says.