Philosophy of AI / AI Ethics Syllabi, Lessons, Readings
As artificial intelligence (or “artificial intelligence” if you insist) is put to increased use in an increased variety of ways, teaching about it has been increasing, too.
A reader writes in:
With the recent explosion in demand for expertise and/or teaching competence in philosophy/ethics of AI, I wondered if there might be a public thread where folks could share their syllabi. This would be a fantastic resource for people who would like to begin teaching in this area but need examples of how to implement such a course.
Good suggestion.
If you’ve taught a philosophy course that is all about artificial intelligence or covers it in a unit, please link to the course syllabus in a comment (if you don’t have a place to put it online or otherwise need help doing that, send me an email).
If there are some readings you think work well for a course or unit on this subject, please share them.
And if you make use of any lessons in which the students are supposed to be using AI as a way of learning something about it, it would be great to hear about those, too.
Thanks.
I have taught several courses on AI. I’m including two syllabi, in case they might be helpful. I think all (or nearly all) of the readings can be found easily using the names provided (I distribute them via Canvas, rather than providing links on the syllabus). But if not, let me know, I can point you in the right direction.
AI and Epistemology (3000 undergraduate level)
Ethical Challenges of AI (2000 undergraduate level)
The latter course works its way up to generative AI by starting from discussion of simpler ML models. The first one jumps right into gen AI.
I am, to put it mildly, skeptical about using chatbots in the classroom.
I’m having trouble finding the text of the Lisa Miracchi Titus paper, “Does ChatGPT have semantic understanding?” The abstract looks really interesting.
Also, that strikes me as a *lot* of reading for one semester, but I’m glad to have that reading list myself!
Yeah, probably a lot of reading. But I was lucky with respect to my students willingness to read things.
just uploaded it up on philarchive 🙂 https://philpapers.org/archive/TITDCH.pdf
Adriano Mannino & I taught a discussion-based graduate seminar on “The Moral and Political Philosophy of AI” at UC Berkeley this past spring.
The first half was devoted to AI alignment & safety, and the second half addressed deepfakes, the future of work, and the possibility of AI well-being.
You can check out the syllabus (and our class notes/handouts) here.
I have taught a German language course on “Democracy and AI” (and a bit more from a social science perspective) at Erfurt University. The readings and recommendations are mostly in English: Syllabus
Here is a course on AI (and more) that uses films, television shows, and documentaries as its only course materials: https://marcchampagnephilosopher.online/wp-content/uploads/2024/02/Thinking-Philosophically-about-Technology-by-Watching-Films.pdf
Here is the syllabus for a Philosophy of AI course I taught at Dartmouth College a couple years ago.
My main aim in this course was to approach AI from a Philosophy of Science perspective with a focus on explanation and modeling.
For that reason, significant time was spent on figuring out how generative AI systems actually work. If I had the chance to teach it again, I would probably expand the sections on creativity, inductive risk and values, and bias. The section on creativity was probably the biggest hit with the students, and there has been a lot of really interesting work in this area since then.
About a year ago, I taught a philosophy of language class that I just tweaked slightly to focus on LLMs, and motivate seeing what classic readings in philosophy of language might imply about them, and what they might imply for these classic readings. I found that this slight change in focus led me to read much more tension between the classic readings on a behaviorist/anti-behaviorist axis than I had ever noticed when I was teaching before.
I started with one article explaining the basics of how LLMs work: https://arstechnica.com/science/2023/07/a-jargon-free-explanation-of-how-ai-large-language-models-work/
and then followed in successive weeks with:
The biggest surprise to me was unlocking a completely new understanding of everything going on in the Putnam paper outside of the Twin Earth thought experiment – he actually concludes with a theory of meaning that is remarkably compatible with the vectors used by LLMs!
I taught a Phil of AI course many, many years ago, described in a Teaching Philosophy article, online at https://cse.buffalo.edu/~rapaport/Papers/rapaport1986-PhilAI-TchgPhil.pdf
And I have more recently taught a Phil of Computer Science course. Syllabus, with links to another Teaching Philosophy article on it, and to my Wiley textbook, are at https://cse.buffalo.edu/~rapaport/510.html
I also have an as-yet-unpublished paper arguing that AI will “succeed”; current version is at https://cse.buffalo.edu/~rapaport/Papers/aidebate.pdf
Other of my papers on the Phil of AI are online at https://cse.buffalo.edu/~rapaport/papers.html (or contact me for versions that are not there)
I was in one of those UB courses! Circa 2007 or so. Great class, and probably my first exposure to philosophy of AI. 🫡
I taught this survey course at Columbia a couple years ago. I think a lot of these readings would still be useful.
(The books referred to are Mitchell’s “Artificial Intelligence: A Guide for Thinking Humans,” Marcus & Davies’ “Rebooting AI,” Boden’s “AI: Its Nature and Future,” and Schneider’s “Artificial You.” I think the rest of the readings are easy to find online.)
At least some of the material from a Data Ethics course that I taught during Sp18-Sp22 at NC State might be useful. Relevant files are .zipped at this link. I also found it useful to discuss the central counterexample of Ned Block’s “Psychologism and Behaviorism,” the ‘Aunt Bertha’ simulator to discourage naive applications of the Turing Test.
For something short and accessible about ethics of using AI to write apologies or love letters, I’ll put in a plug for my “ChatGPT and Emotional Outsourcing.” You can download the pdf on philpapers or access the online version at Prindle Post.
I haven’t had a chance to teach it yet, but friends have reported good student engagement.
I’ve injected a little AI into an intro-level course I teach on the ethics of persuasion. I use Kelly Weirich’s (excellent!) Prindle Post piece on emotional outsourcing, Emily Bender’s Guardian piece on LLMs abusing our empathy, and the movie Ex Machina. Those are all very accessible, and the students get really into them. (I have mixed feelings about Ex Machina, so I give students the option of reading the screenplay.)
But I also wanted to ask a more specific version the question for this thread. I’m thinking of developing a mid-level course on the ethics of persuasive technology, with special attention to issues of respect and manipulation. If anyone had suggestions for accessible and ethically-rich readings specifically on persuasive tech, I’d be grateful.
I love hearing that you teach my emotional outsourcing piece. Thank you! For persuasive tech, I wonder if a forthcoming paper on AI romance systems coauthored with AG Holdier might be of interest. We argue that, whether or not LLMs generally issue speech acts and regardless of downstream effects, AI romance systems promote misogyny through making the (usually male) user invulnerable to rejection, allowing them to shape their partner’s identity, and treating their desires as a consumer as paramount. These features invite the reader to conceive of intimate relationships in a subordinating way by presupposing a certain ideal or perspective. We don’t directly discuss persuasion, but it might be a nice way to talk about responsibility for what generative technology communicates, how persuasion can be due not only to specific linguistic output but also to the system and its operation on a larger level, etc. I’ll try to return here when we have a preprint up in case it’s of interest!
Preprint is up. AI Romance and Misogyny: A Speech Act Analysis
Thanks so much, Kelly!!
You can find my syllabi here: https://danielweltman.com/teaching.html
The relevant ones are “AI Ethics” and “Ethics and Technology,” along perhaps with “Technology and Human Values.”
Here’s a link to the syllabus and materials for the summer grad seminar that Katie Creel and myself lead: AIDE Summer Syllabus.
It’s designed as an intensive summer program, but people might find the readings on specific topics useful.
Also, we’re running the program again this summer, so please tell grad students you think might be interested. They can find more info at http://www.aidesummer.org.
I have taught a quite technical (but no math or programming required) and intense 3 hour masterclass on transformer language models at Barcelona. The slides are online here: https://dstrohmaier.com/presentations/presentation_masterclass.slides.html
But the interactive part was the highlight.
I would be happy to do a rerun at other universities, so feel free to contact me.
I just taught an intro level Philosophy of AI class at the University of Notre Dame. You can find a list of topics and reading here on my website: https://sites.nd.edu/yuanshan-li/teaching/philosophical-issues-in-ai-fall-2024/
I have taught a class on Philosophy of AI in Amsterdam. We first read Sven Nyholms book on Technology, especially the chapters on robots and then my Elements book on Artificial Minds (with Vincent Müller), which is still under review. Please email me for specifics, if you are interested.
I teach an AI Ethics course as an undergrad senior seminar. The readings draw from recent AI Ethics literature, with some background on computing history and the tech industry. The readings are designed to set up an extended AI Audit project, modeled on Raji et al’s (2020) audit framework, in which students explore different gen AI services through adversarial testing to probe the limitations and guardrails of the services. I describe the audit project for the Critical AI Blog here
AI Ethics Syllabus
AI Audit Project Lesson Plan
Through this project, my students have turned up some interesting examples of bias in machine learning. For instance, most gen AI services have distinct cheese preferences depending on which language is being used.
A course I taught at Dalhousie on the ethics of computing more generally was featured in the APA Blog’s syllabus showcase a few years ago. I teach a descendant of that course at Cornell now, and it has somewhat more on AI at this point. But I still think it’s important to cover ethical issues that arise in computing without anything someone might call “AI” getting involved. Most of the issues raised by various things called “AI” these days (digital intellectual property, privacy, bias, cybersecurity, digital divides, misinformation, responsibility gaps, automation, sustainability, etc.) are persistent issues in computing rather than radically new problems, and a broader perspective grounded in the history of the field is helpful to have.
This is for an interdisciplinary honors GE course about human minds and AI:
Mind in the Machine
Here’s the course description: This course will explore the question of whether an artificial intelligence could have a mind akin to ours. We will also look into AI’s current and future impacts on society, employment, and human identity. To accomplish this, we will examine the nature and current limits of AI, and how the fields of psychology and philosophy have characterized the human mind. In the process we will examine similarities and differences between AI and human cognition in areas such as perception, awareness, reasoning, decision-making, emotion, and self-reflection. By the end of the semester, you will be better equipped to evaluate claims about AI’s capabilities, and to think critically about the nature of intelligence, thinking, consciousness, and what it is to have a “mind”.