We explored the reasons why some universities have recently created courses that combine AI and computer science with ethics in our last post. More than ever, ensuring that artificial intelligence (AI) is fair, accountable, and transparent is critical.

AI and Ethics Coursework: Analyzing the Algorithm

It’s important to think out the design of an AI on the front end, because once the AI is created and begins to learn as it processes more and more data, opening the “black box” and figuring out how an AI arrived at its response to a particular issue may not be possible. Because AI like this can be involved in crucial processes like diagnosing medical conditions or driving a car in the very near future, some researchers are designing software to analyze AI.

For example, Klaus-Robert Müller, professor of machine learning at the Technical University of Berlin, and his team developed an inspection program known as Layerwise Relevance Propagation, or LRP. This program can work backwards through an AI’s neural network to pinpoint how a response was formulated.

Another approach has been formulated by Sandra Wachter, a lawyer and researcher in data ethics and algorithms at the Oxford Internet Institute and the Alan Turing Institute, and her colleagues Brent Mittelstadt and Chris Russell. In their 2018 paper, these researchers’ approach is to work out what’s needed to change the AI’s conclusion to a desired result.

Approaches like these are not necessary, of course, if an AI has been thought out, with the ethical implications of its algorithms’ conclusions thoroughly established. That’s why universities right now are so concerned about their students being thoroughly educated on the effects AI can have on people.

Mary Gray, a senior researcher at Microsoft Research, notes, “Our challenge in industry is to help researchers and practitioners not see ethics as a box that has to be checked at the end, but rather to think about these things from the very beginning of a project.”

AI and Ethics Coursework: Harvard University’s John A. Paulson School of Engineering and Applied Sciences

Higgins Professor of Natural Sciences Barbara Grosz designed a course in 2015 called “Intelligent Systems: Design and Ethical Challenges” and invited philosophy professors to co-teach the course. The response was dramatic: by the second year, 140 students vied for 30 spots, and faculty expressed great interest as well.

So Grosz and Alison Simmons, the Samuel H. Wolcott Professor of Philosophy, developed an initiative called Embedded EthiCS. This model integrates computer science courses with philosophy: philosophy graduate students teach side by side with computer science professors, weaving activities and assignments into the curriculum that reinforce students’ ethical deliberation.

Simmons notes, “We want to send the message that ethical reasoning is part of what you do as a computer scientist.”

The program has tripled in size since its beginning in Spring 2017. In addition, the course materials are available online, so that other institutions can access them.

AI and Ethics Coursework: Duke University’s Pratt School of Engineering

The Lane Family Ethics in Technology Program was recently created within Duke University’s Pratt School of Engineering with a $300,000 gift. This gift will support new coursework that interweaves ethics with engineering, as well as an annual symposium focused on technology and ethics.

An interdisciplinary oversight committee is composed of faculty members from Engineering, Law, Medicine, Divinity, Sociology, and Computer Science. This committee will evaluate new coursework ideas.

Nita Farahany, director of Duke’s Science and Society Program, is one of the committee members. She says, “The increasing reality is that students will encounter significant ethical issues in the future of their engineering practices. We have not traditionally and increasingly must explicitly prepare students to thoughtfully and effectively navigate these challenging issues by integrating ethics into their core engineering curriculum.”

Suzanne Shanahan, the Nannerl O. Keohane Director of the Kenan Institute for Ethics, is another committee member. She adds, “The Lane Family Ethics in Technology Program represents a wonderful opportunity for Duke to become a national leader in integrating normative analysis and ethical decision-making into its engineering and computer science curriculum.”

AI and Ethics Coursework: Rigorous Thinking Required

Jeffrey Behrends, a fellow-in-residence at Harvard’s Edmond J. Safra Center for Ethics, has co-taught a design and ethics course with Barbara Grosz. He notes: “Once students who are unfamiliar with philosophy are introduced to it, they realize that it’s not some arcane enterprise…. A lot of students who are attracted to computer science are also attracted to some of the methodologies of philosophy, because we emphasize rigorous thinking.”

Given the popularity of computer science right now – 40 percent of MIT undergraduates now either major in computer science or earn a joint degree that includes computer science – there’s definitely no time to lose for universities to step up and focus on the integration of ethical thinking and engineering.

In our next post, we’ll examine efforts both large and small by universities and nonprofits to do just that.

What do you think? Let us know in the comments.