Michigan Law Mini-seminar Considers AI Doomerism and Technology Risk

Assistant Professor Salomé Viljoen’s mini-seminar, Does AI Pose an Existential Threat to Humanity?, opened with discussion about what AI technologists and philosophers call the alignment problem. This school of thought explores the idea that powerful artificial general intelligence (AGI)—loosely defined as AI that can complete complex tasks that reach or exceed human capability—could develop in a way that conflicts with the continued existence of humanity. 

Companies in the AI space see AGI as something of a lodestar, and OpenAI, Meta, Google, and others have invested billions of dollars into AGI research. They see the consumer-facing chatbots that have come to market in recent years as a stepping stone in the race for AGI. But a common point of discussion among alignment theorists is the risk of inadvertently designing an AGI that produces unintended results—to potentially devastating effect.

“You can imagine, for example, designing an algorithm with an objective function to eliminate traffic congestion in Ann Arbor,” Viljoen says. “But if you don’t appropriately specify or constrain the prompt, an artificial general intelligence could determine that the most complete way to eliminate traffic would be to kill everyone in Ann Arbor. But Ann Arbor is an attractive place, so maybe people from Ypsilanti would then move to Ann Arbor, so perhaps the AGI would choose to kill everyone in Southeast Michigan, and so on.”

While this scenario may seem far-fetched in 2025, some technologists and engineers are focused on aligning the incentive mechanisms of AI and future AGI systems with human flourishing. But even if AGI isn’t realized, Viljoen says there is value in exploring the different ways that opportunity and risk are being framed in the context of these technologies. 

“The alignment problem is one way to conceive of the risk of AI, and I want students to take that seriously. But AI is a real technology that exists right now, and there are a lot of ways to bring this topic down from the theoretical to the immediate and the material, and to think through the costs and benefits,” she says. “I want to provide other ways of articulating how we should think about trade-offs and how we should think about this technology from other perspectives."

From there, the discussions were more granular and examined how AI is already interacting with the environment and energy infrastructure, racial and sexual forms of discrimination, labor law, and other aspects of law and society. Viljoen assigned a variety of readings to inform the conversations, including news articles, proposed regulations in the US, and analysis of existing AI laws in the European Union.

Viljoen taught the mini-seminar in fall 2023, when ChatGPT and similar products were beginning to take off—a trend that has only accelerated since. In winter 2026, Viljoen will teach a course called Law, Technology, and Society and will offer a course on digital governance and privacy during a future term, both of which will involve discussion of AI.

“What I really want to get across in my classes is not how to use the technology but how to think about it as something that lawyers are going to be regulating,” she says. “We can choose the direction along which this technology unfolds and the ways it can impact our lives. There’s a role and an expertise for lawyers in setting the social rules that determine whether or not these technologies end up being socially valuable or socially harmful.”