CS182: The Bug in Stanford’s CS Department

CS182 is Stanford’s flagship CS & ethics course, advertised to solve the problem of big tech by patching engineering students’ minds. But it is really a bug in their education. Praised by the media, the course supposedly introduces a novel pedagogical method that teaches engineers to think ethically. I took the inaugural offering of the course in 2019, and it has since been codified as a means to fulfill writing and ethics requirements for computer science students.

While the professors teaching CS182 have noble intentions, the course ultimately fails at its objective of getting students to engage deeply with tech and ethics issues. CS182 needs to change. The current offering treats technical subjects with little rigor and caricatures engineers as optimization junkies who cannot think ethically without instruction from enlightened ethicists. The course is riddled with armchair abstractions and laced with progressive ideology. It would do better to focus on implementing actual systems to resolve existing ethical issues in tech rather than engaging in cheap philosophy.

The course paints engineering students as a crowd of myopic amoralists in need of philosophical enlightenment. Its lecture slides begin by accusing engineers of being stuck in a dangerous “optimization mindset.” Computer Science students are told their curriculum is “built around an entire value system that knows only utility functions, symbolic manipulations, and objective maximization.” It is assumed that we are living in the cave of techno-optimism, where all who write code naively think that technological progress is an end in itself or that optimization is a first-order value.

Students supposedly see the light by listening to a 5-minute primer on John Rawls’ Veil of Ignorance, reading a couple of pages about the Panopticon, imagining themselves in Robert Nozick’s Experience Machine, and reading a chapter from Mill’s On Liberty. But there is no serious philosophy being done — no context, no tradition, no set of common ethical principles are articulated. Few questions, responses, or rebuttals are addressed. This kind of education is akin to trivia or quiz bowl. What is most fair? Adopt the Rawlsian Veil of Ignorance! A problem with utilitarianism? Enter the Experience Machine! Privacy? Life is a Panopticon!

Despite the shallow talk bereft of substance, the instructors speak as Silicon Valley’s philosopher kings endowed with special ethical wisdom. They would do well to recall those words of Socrates in the Apology, that those who think themselves wise may not be wise at all. To sum it up in another student’s words, “Stanford CS students are not a wayward congregation completely ignorant to the issues, stop treating us like it.”

The class encourages students to develop sloppy solutions to the few real technical issues it seeks to address (for example, algorithmic bias). One assignment has students evaluate a machine learning algorithm that predicts criminal recidivism — with the intention of showing students the risk of racial and gender bias in algorithms. Students train a model with data containing race-based input features and find disparities in prediction between African Americans and Caucasians. The assignment was seen by many students as trivial and non-rigorous. In training a model to predict criminal risk, one student realized he could just disable training on any data about previous offenses to satisfy the “fairness” requirement. In simpler terms, he decided not to look at perhaps the most important data (defendants’ previous convictions), in determining whether they would recidivate, so he could be “fair.” Professors are encouraging kindergarten-like tinkering far from any principled engineering.

CS182 claims to cherish diverse opinions and thoughts, but it is steeped in progressive ideology. In the words of another student, the class contains a “big-tech bad, capitalism bad, union good” bottom-line. Consider one of the first assigned readings, “Optimize What?” by Jimmy Wu, which criticizes computer science and mathematical subdisciplines as reinforcing capitalist structures. The article literally calls for a Marxist people’s revolution of Silicon Valley. Workers would take control of the technology sector and implement a “communist realism in science” where algorithms are “developed for economic planning” and “decentralized protocols are utilized for coordinating supply chains between communes.” If that is not ideology, I do not know what is.

Next, consider the course’s coverage of privacy. Lecture slides point to the 1619 project, as instructive of the U.S.’s deep privacy violations. Technology facilitating workplace supervision, like taking drug tests and checking emails, is actually motivated by the impulse behind “technology that pervaded plantations.” But are United States prisons and American structural racism truly the prime examples of invasive surveillance? There is a certain literal surveillance state which tracks and judges its 1.3bln residents with a social credit system. But is there any mention given or any examination of this reality? I found no mention.

Also, look at the course’s coverage of diversity and inclusion in technology. Plenty of attention is given to the lack of women or minorities in technology. It's not just Stanford, but Silicon Valley and Wall Street seem to decry the injustice. Yet, no mention is given to how Silicon Valley’s managerial class overwhelmingly votes progressive or how conservatives and people of faith often experience alienation in many of these workplaces. The professors’ claims of diversity ring hollow.

CS182 undermines quality computer science education by essentializing philosophy into tropes, encouraging sloppy technical solutions, and silently pushing a certain strand of progressive thought. Rather than trying to enlighten engineers with special ethical knowledge, CS182 should focus on teaching engineers actual practices that serve the nation. Things like patching vulnerabilities and building secure databases. As for teaching students to think deeply about tech and be more “ethical” — a shared political culture, commitment to the development of virtue, and a responsibility to the nation over the self must be restored. But emotivism and moral skepticism now permeate the campus spirit. Little allegiance to virtue remains. Instructors are desperately trying to get students to grasp onto moral responsibility, but they have no foundation to build on. It is a futile exercise.