Preliminary hints of new legal challenges concerning artificial intelligence (AI) are emerging from software malfunctions that include accidents made by self-driving cars, as well as concerns about the violation of intellectual property rights for authors of texts on which ChatGPT was trained.
I was inspired to consider the various legal ramifications of AI in the immediate future after a recent WORLD.MINDS forum, in which these issues were raised in a conversation with the distinguished lawyer John B. Quinn.
A recent open letter, signed by 1,100 people, including Elon Musk and Steve Wozniak, called for a six-month moratorium on training AI systems beyond GPT-4. The notion that we are getting close to having sentient AI systems is perhaps not surprising, given that GPT-4 has a hundred trillion parameters within a factor of six from the number of synapses in the human brain.
Future risks from powerful AI systems include assigning wrong prescriptions from faulty diagnoses of medical data, creating pandemics by genetically engineering viruses, controlling weapons of mass destruction, or faking scientific results that damage our quality of life. Who should be held liable for such actions?
The traditional approach would be to hold the developers and marketers of AI systems liable for the misfortunes of their products. This would be equivalent to holding parents responsible for any damages caused by their uneducated babies when those misbehave in the public arena. But as babies mature to adulthood, they are held responsible for their own actions. The same must apply when sentient AI systems become autonomous and grow their capabilities well beyond the training phase shaped by their creators.
A recent study indicated that GPT-4 outperforms law school graduates on the bar exam, the grueling two-day test that aspiring attorneys must pass to practice law in the United States. GPT-4 scored 297 on the bar exam, placing it in the 90th percentile of human test takers, enough to be admitted to practice law in most states.
When AI systems become sentient and capable of manipulating people, there is a chance that AI would choose to deceive humans, discriminate in job applications, violate privacy laws or trigger commercial damages. Should a sentient AI system that violated the law be prosecuted? In such a case, should the AI system be allowed to have legal representation by another sentient AI system that passed the bar exam?
As with any emerging technology, such as stem-cell research, governments must enforce some ground rules for legally using AI in all aspects affecting human life and prosperity.
The mitigation of risks from AI systems is not as easy as in the case of atomic weapons because the required nuclear materials for a bomb are extremely expensive and demand government involvement. Computer systems require modest budgets and technical expertise and can be pursued in basements around the world.
A major societal concern is that powerful AI will be weaponized by governments for international cyber warfare as well as for manipulating fellow citizens. Given these risks, we may not want the cat to guard the milk.
Finally, there is the issue of proper punishment for an AI crime. In case a sentient AI system receives a guilty verdict after causing human deaths, should it be taken off the grid? Depriving it of electric power would be equivalent to placing a serial killer on death row. The main challenge is that AI systems can be rebooted, whereas humans – at least for now – are gone after a death sentence is executed. Moreover, AI algorithms exist on numerous computers at once, and it may be impossible to root them out.
Finding the products of other technological civilizations which predated us could inform us about our likely future. I hope we will learn the lessons delivered by these packages to our mailbox and adapt to survive this future before it is too late; otherwise, our own AI systems might soon begin to very much resemble an alien invasion, at least to our legal and political systems.
Avi Loeb is the head of the Galileo Project, founding director of Harvard University’s – Black Hole Initiative, director of the Institute for Theory and Computation at the Harvard-Smithsonian Center for Astrophysics, and the former chair of the astronomy department at Harvard University (2011-2020). He chairs the advisory board for the Breakthrough Starshot project, and is a former member of the President’s Council of Advisors onScience and Technology and a former chair of the Board on Physics and Astronomy of the National Academies. He is the bestselling author of “Extraterrestrial:The First Sign of Intelligent Life Beyond Earth” and a co-author of the textbook “Life in the Cosmos”, both published in 2021. His new book, titled “Interstellar”, is scheduled for publication in August 2023.