PWR 2STA: Ethics and AI
Industry mentions of “Killer machines” often intend the adjective “killer” as marketing click-bait. Wow, this “killer app” -- an amazing algorithm or robot or concept that solves a previously unsolvable problem! “P = NP?" Solved!! Or, “Killer robots,” that might be taking over every aspect of our lives, which they already are. No need for click-bait here.
In the first iteration of this course in 1985, we looked at ethics and AI through the “killer machine”: Therac 25—a large, ungainly, and by today’s standards, low-tech radiation machine that actually killed people due to a software bug that caused Therac 25 to deliver lethal doses of radiation to patients. Accordingly, students explored tensions around problems of software design and asked hard ethical questions like, “who bears responsibility for killer machines’ failures and often unexpected outcomes?”
Today, more than three decades later, ethical questions around AI remain remarkably the same as during the time of Therac 25. Yet machine learning and artificial intelligence are also raising new challenges: data training sets, classifiers, ontologies that import many age-old problems and prejudices. Computer science cannot resolve these issues on its own. Rather the solution lies in interdisciplinary dialogue from many fields, including philosophy, ethics, social science, law, and medicine.
This PWR course invites students to spend a quarter researching, debating, writing, and presenting on the “killer” machines of our era that, thanks to the growth of machine learning, are both amazing and remarkably mundane: autonomous cars, medical devices, A.I. and quantum computing, computer vision, robotics, human-computer interaction, siri, alexa, nest and many more. Final research projects will argue for the ethical consequences around a specific aspect of AI and machine learning.
(5-minute live oral presentation; written text of 900-1200 words): You will propose a topic for an interdisciplinary research project that will address the ethical complexities of AI in any of its many manifestations. Proposals will identify relevant scholarship and primary sources, and will pose focused research questions. Potential topics include the autonomous cars, bot-fueled political campaigns; women and data science; machine learning; algorithms that track sex traffickers; cybersecurity and the grid; the civil rights of sentient artificial life forms; Brain-Computer Interfaces; algorithms and personal privacy.
Written Research-Based Argument
(3000-3600 words; 10-12 pages): After presenting your proposal, you will pursue research on your topic and craft a convincing and well-researched academic essay. Your essay will draw on analysis of primary sources and relevant secondary sources to support and contextualize its argument.
Delivery of Research
(10-minute live oral presentation with appropriate multimedia support): You will translate your research-based argument into an oral presentation, which you will deliver at the end of the quarter.
(5 minute podcast or videocast): For this project, you will translate your research project on Ethics and AI into a short podcast or videocast in MP3 or video file format, with an option to post to the Internet. If you love making short audio or video recordings you can develop an RSS feed. This is your chance to communicate your research to a wide audience in another genre with a more conversational style.