HotEthics 2024: A Reflection

I attended HotEthics at San Diego yesterday. It was somewhat strange that I ended up attending: Scott encouraged me to submit a rant about my views on whether optimizations improve social utility, and the workshop required in-person presentation, prompting a trip to San Diego¹. This blog is a reflection on what got published and said at the workshop.

What do Computer Scientists mean by "ethics"?

The workshop purportedly exists to discuss ethical concerns in the design of systems. But there seemed to be no conversation about what ethical concerns mattered. The talks varied between conversations about:

• Reducing environmental impact, which I think is an important concern for humanity, but I do not know where it fits into the ethics, which I have always understood to be questions around how people should behave.
• How should systems designed to be implanted in the human body (e.g., brain-computer interface devices) be designed, that is, should they be designed to be flexible or not, how general should they be, etc. This is getting closer to ethics, but the answers seem to be somewhat addressed by medicine, and I could not figure out why those guidelines did not suffice (beyond perhaps a NIH like desire common to computer science).
• Designing equitable systems, which again leads to questions such as: is equity an ethical concern; and how should individuals (systems designers, etc.) ensure equity without limiting user choice (which itself is a different form of inequity).
• Discussions about teaching students that repair is possible, and encouraging repair. This was the one topic that did seem to suggest an ethical framework and a way to adopt it, but alas there was only one talk.

The audience discussion at the end seemed to be similarly mixed, suggesting computer scientists take the type of professional ethics courses required of engineers (thou shalth not take bribes to sign off on poorly designed bridges), which are of course useful but do we really need an entire workshop to learn that people should not be breaking anti-bribery laws?

The Role of Education

A lot of the discussion focused on the role of education in getting students to think about ethics when designing systems, One suggestion was engineering ethics courses (seemingly required in Canada), and the previous section talks about my concerns on that front, so we won't discuss that proposal. The other suggestions were to: (a) borrowing from the ML community; (b) borrow from courses at various schools; and (c) creating a new course. I am in favor of all three of these alternatives, but am unsure what they would cover. I wonder if this entire discussion is not a sign that computer science students need to be encouraged to take courses in other areas, a return to the original liberal arts way.

So the workshop as a whole?

So now that the workshop is done, what next? I am not sure: submitting a rant to a workshop is not a ton of work, but I am not sure whether it contributes to the community. If the goal is to really design systems ethically, we need to start by defining what that even means.

Cynically, I worry our definition of ethics will be dictated by what funding agencies and universities find important, which in turn will mean our ethical considerations will be dictated by the fears that appear in the newspaper opinion pages. If that is where we go, then HotEthics is likely going to be just another workshop on early work. As a community we seem to have an insatiable appetite for new venues.