Hacker News Reader

FWIW, I've never seen any value whatsoever in these ethics classes (note: I'm not saying it's not an important topic, it's extremely important).

The hypotheticals and situations presented and discussed in many of them are very similar to yearly mandatory classes we have to take at FAANG (and many other workplaces I assume) and it boils down to having a checkbox that says "I'm not stupid".

Real world ethics is much more subtle and difficult to teach imo. Like how do you do trade-offs around cost/societal issues/potential financial consequences and do it under pressure with your job on the line.

In Canada, we have the iron ring given to engineers to remind them of their duties to society, I think that serves as a good reminder even if it's symbolic.

Sounds like you just haven’t taken a good ethics class. I was required to take one in grad school (comp bio/biost at) taught by a law professor. Every week we were assigned a case study on a real-world ethics problem, such as the Duke/Potti scandal [1]. We wrote up answers to 5-6 discussion questions ahead of class, then spent the class time in small group discussion rather than lecture. I think it was one of the most valuable courses I took in grad school.

[1] https://en.wikipedia.org/wiki/Anil_Potti

I don't understand, what's the controversy with Potti? The dude was a piece of shit - what else is there to discuss? I'm not being condescending I genuinely don't see any nuance like you're stating. When is it ever okay to fabricate data?

> When is it ever okay to fabricate data?

When you want to deliberately sabotage something. What may have its place, but obviously not in such case (obviously not applicable in cancer research like in this case).

Which should be obvious to anyone who has made it to 18 years of age.

> a real-world ethics problem, such as the Duke/Potti scandal [1].

Why is this an ethics problem? That's just "normal" scientific fraud, isnt it? What is there to discuss?

I think you're right the problem is the course. For example I also had a great ethics class, but it was bioethics, not technology.

Part of the problem is you think that software engineers should take a course in software engineering ethics. The problem here is like the GP says, it boils down into a class of are you dumb enough to pick the obviously unethical answer and fails to teach any ethics.

I wonder if requiring an ethics class that focuses on an orthogonal topic, for example requiring software engineers to take a course in business ethics, might be more effective since they can focus more on the actual system process of ethics and less on how it effects their own personal career, and thus be more willing to honestly evaluate ethically grey scenarios.

I had a required ethics course for my CS degree. I can't recall anything from it. I also took a one day ethics course at a FAANG that was excellent. Far better than a "I'm not stupid" checkbox. I think both of these are true:

- Ethics should be a required course for CS.

- CS classes in general do a poor job of tracking well with the industry. Fine for some things (eg, theory), but awful for others (eg, any Software Engineering course which touches on waterfall.)

For those who have missed out, here's an ethics course in the size of a paragraph:

Most engineering ethics focus on life and death, such as making sure your bridge doesn't collapse; Most software is not a matter of life and death, so ethics may seem irrelevant. However, the incredible power of technology means your design choices will have an impact on the lives of many. Seeing red lines where you need to tell your employer no is important, but everyday ethical thinking will find problems other approaches to design will be blind to. This in turn will make better products, and increase your value to your employer. So throw an "ethics" section in your design doc, and ask the question "Does this technology as designed hurt any groups of people with specific characteristics?" You might find that your camera app needs better calibration for dark skin tones, that your UI elements are too hard to use for women with large fingernails or people with minor motor disabilities, or perhaps that one of the JS libraries your considering is very large, and will make a much worse experience for people with rural and slow internet. So simply stop, think, and approach your design from the perspective of different cultures, races, religions, ages, genders, and life circumstances.

I like your approach here. As strange as it sounds the path to ethical action in tech requires cultivating the kind of imagination required to think of all the ways tech can be used for ill (sometimes called 'design noir'). I'd hope a curriculum would have a hefty dose of this.

Gone are the days we can just build something and act naive about what ways tech can be abused or think we can solve those issues as they come up.

Most of the time, it's not even in companies interest to look too closely at the negative implications of their products, because then they'd have to act.

Hence, its better to anticipate problems at the design phase than deal with them after the thing is built and corporate inertia and sunk costs demand it stay the same or only get band-aid fixes to those problems.

The one ethics course I took as an EE undergrad taught me exactly one lesson - that the whistleblower always has their life ruined, and even ones who (in the US) have received significant payments typically regret it.

I think it's the reverse of the lesson they were trying to teach.

Not directly related to tech, but I really recommend the movie The Insider for a look at how whistleblowing goes. It’s about the tobacco industry but can likely be generalized to any situation.

https://en.m.wikipedia.org/wiki/Jeffrey_Wigand

Hacker One's business model is creating a pen for all the whistleblowers (security researchers doing bug bounties) so they are trapped inside it and can't escape unless they want to lose their ability to get paid in the future.

It's a classic Monopsony.

I agree that most ethics courses aren't well done and scream 'mandatory training' rather than 'interesting subject'. The required two week one I took at Stanford GSB was pretty weak, despite business ethics being extremely important. I think they've made it better since I took it; it was a sideshow when it should have been much more central to the business curriculum.

Pedagogically, going purely book- and reading-based like most courses is a bad idea, as that can't capture the paradoxes and nuances involved in ethical decision making. I think the field would benefit from more decision-oriented approaches including a review of past ethical dilemmas perhaps as case studies, more embodied/situated formats like simulations, and practical advice on how to identify and/or avoid getting into ethically compromised situations in the first place. But you do need to couple those with the foundational frameworks that you'd use in those situations, and ideally contextualization with much broader ideas like the social contract in society.

Ethics guidelines from professional engineering societies can help with some amount of ongoing ethical awareness, but I do think it has to start in the formal education system.

Also, the engineers are often well aware when something is wrong or dubious.

But there's a difference between knowing that and having the power to do something about it.

This list is also a bit incomplete, or more likely just refers to classes taught in the current school year. I co-teach a different ethics-focused class in the Stanford CS department (CS 181, which in this incarnation started in 2017). I think it's a pretty good class too, although as you suggest, there are limits on how much a quarter-long class can realistically accomplish. (We try to push the envelope with our project -- some of our students have protested outside Apple HQ or Robinhood, given presentations to the Palo Alto School Board, interviewed farmworkers and U.S. congresspeople about automation, handed out leaflets about net neutrality to tourists, etc.) The reading list and goals, etc., are up at https://cs181.stanford.edu.

In addition to the new "embedded EthiCS" program you mentioned, the CS program also offers a bunch of more focused classes on, e.g., trust and safety engineering (CS 152); law, order, and algorithms (CS 209); race and gender in Silicon Valley (CS 80Q), fair, accountable, and transparent deep learning (CS 335); computational social choice (CS 366); digital technology and law (CS 481); the Modern Internet (upcoming). And, of course, a much bigger range of classes about engineering, ethics, and public policy offered by colleagues across the university.

The truth, though, is that neither CS 181 or 182 is required per se. All Stanford undergrads have to take at least one course in ethics, and all undergrads in the School of Engineering have to take at least one course in Technology in Society, and both CS 181 and 182 fulfill both of those requirements. But many of our students do fulfill them by taking other classes.

Thanks for the practitioner perspective Keith, this makes me feel a lot better about the approach Stanford is taking (and makes me want to return to the farm!)

Honestly the ethical questions being brought up are some of the most interesting things happening in tech right now. Society & the human API is like the next abstraction layer for tech, and the dust has hardly settled. It's really created a lot of opportunities to ask ourselves what kind of society we want to live in. Your CS181 syllabus looks great!

Having the computer science department, specifically, offer more ethics classes is controversial and has gone different ways at different universities. It may seem like an obvious win if you care about society, but many people interested in tech ethics (esp. those who don't consider themselves computer scientists) think it pushes in the wrong direction as far as systems and institutions go.

At one place I've taught, there was quite a bit of pushback from ethicists against the CS department expanding ethics education in its departmental courses. They viewed this as part of a "CS eats everything" trend that would hollow out humanities and social science departments, as well as diminish the role of a broad liberal education codified in the core non-major-specific curriculum. Instead, they preferred CS to stick to more technical classes, but simultaneously were pushing to include at least one ethics course as part of the gen-ed requirement that all students have to take, with ideally a 2nd course more tailored to type of major. In any case, courses taught by social scientists or philosophers, not by CS profs with a CS course number.

Some of this is just university turf warring, but I think some reflects real differences in opinion on how to structure things. Particularly in the U.S. version of university education, where majors are not as dominant a proportion of total course requirements as in some countries. (Depending on the university, courses in your major are something like 35-60% of the total.)

Interesting, thanks for the context. I get that its tough to find the right dividing lines on these in a university context, though I admit I'm partial to more integrated approaches. Like, I can't help but imagine teaching ethics within a tech context would be far more stimulating for CS students than a general ethics course.

I guess the good part about a general ethics course would be exposure to students in other majors that think a bit differently, which would hopefully open up CS students perspectives.

On the other hand, a CS specific course would be able to do something like assign students a project to build something, and then follow it up with questions of whether that should be built at all, because the thing in question was ethically problematic.

From my personal experience, the introductory ethics classes taught under philosophy often feel like "philosophy of ethics", with tenuous connections to anything taught elsewhere. Classes taught under science and technology studies are more useful, because the field is empirical. Even if none of the case studies are from your field, the ways STS approaches them are widely applicable.

CS 181W/182W is pretty close to a required class for a Stanford CS major. There are "Technology in Society" and "Writing in the Major" requirements, and taking one of those classes is the simplest way to satisfy both.

https://ughb.stanford.edu/courses/approved-courses/technolog...

https://undergrad.stanford.edu/programs/pwr/courses/writing-...

Ah good to know, happy its at least a soft req!

Stanford student here. You’re right that there aren’t a ton of dedicated ethics classes within the CS department (though there are many tech-oriented ethics courses around the university).

That being said, there has been a major push recently to _embed_ ethics inside the intro sequence.[1] All of the major intro classes (106A, 106B, 107, and 109, to name just a few) have dedicated ethics components.

[1] https://hai.stanford.edu/news/building-ethical-computational...

Thanks that's really good to hear, embedding and repeating the theme throughout different courses really is the best way to build ethical awareness and mindset.

When i was at uni (definitely nothing as high powered as standford) we had one ethics class. It was the most useless class i have ever taken.

The prof seemed very uninformed about the computer side of the material (it was taught by someone from philosophy dept). Examples were from like the 80s (cyberstalking people logged into the mainframe via finger and w) and just generally had nothing relavent to any actual ethical conundrum.

But i did learn being a cartoon villian is wrong.

Ditto moral philosophy, civics, epistemology, logic. The earlier, the better.

Started binging History of Philosophy Without Any Gaps during the apocalypse. So grumpy that I missed all this stuff. If nothing else, I would have learned that most of our debates in society are truly ancient, maybe save myself some angst.

https://historyofphilosophy.net/all-episodes

When I was at university (mathematics; not so long ago) there was basically no ethics course in the curriculum[1] at all. Common career paths were:

- academia

- intelligence agencies

- teaching

- various “data science” or “AI” type things (could include eg pricing insurance or selling/targeting ads amongst other things)

- computer programming

- finance (investment banks, quant shops, etc)

[1] A few lecturers started trying to do a few lectures in ethics topics while I was a student but I think they weren’t really a proper part of the curriculum

I've been asked to do things I was ethically conflicted about. What I learned is that the best way to be ethical is to be rich, because telling your boss you refuse to do things is easier said than done when you have a family to support.

I have found when it comes to money/career; 99.99% of people look the other way, or rationalize their cowardness, or greed.

In college, 1 out of a 1000--10000 (I'm pulling these numbers out of the air) students ever publicilly questioned a teacher who wrong, or unprepared. Yea--I get it. They can ruin your life, esecially the bloated senile fools with Tenure.

And yes--it's easy to be ethical if your are wealthy.

I walked out of a Chiropractic School because Subluxations are maybe 1 in a miiilion, and any success in Chiro is based Placebo.

So many in health care are absolute ethical frauds that fall back on The Art of Medicine in order in order to make more money.

My elderly mother has pain in both shoulders. He doctor's office said the doctor can only exam one shoulder per visit. Just rediculious bill padding. I'm not worried because the pain in in both shoulders, and related to movement.

I won't get started on my Psychiatrist who demands I see her monthly in order to get a prescription refilled. And yes--the rediculious meetings are me giving her $159 cash for a 5 minute session with someone I despise personally/professionally. Why--because I'm addicted to the two highly addictive drugs, and I've been on the drugs for decades. I figure why put my body through a detox at this point. I have to play the charade.

This problem has been the basis of my own studies.

I’m studying computer science in an interdisciplinary build-your-own major style program, focusing on AI ethics. I’m not in any position to make these kinds of claims but I do think that, as you’ve mentioned, these issues will only become more serious in the near future.

At first I was a little worried that not having the traditional CS degree on my resume would hinder me from career opportunities but learning about problems within this space has been incredibly inspiring in its own right.

I'm a lawyer and now developer, I'm not sure how many classes would be enough? Ethics in the technology sense is not the same as legal ethics and may be closer to philosophy?

Stanford Law and Biz schools both offer ethics courses.Butdoesnt look like any are required.

Ethics courses are presumably optional at the law school and the business school.

Ethics required for a degree, sure.

Ethics required for CS in particular, no.

There is nothing special about a CS degree which requires an ethics course, and this is even if one conflates CS with software development.

Embedded ethics.

Does Harvard CS have an ethics course?

If it does, did Mark take it?

If he didn't, that explains it. If he did, the class is useless.