Final Ethics Project
We've been slowly working through some ethical issues throughout the course, and now it's time for you to do a deep dive into an ethics topic on your own. In this project, you will write a position paper on a topic of your choosing. Anything from the suggested topics or from class topics (in more detail than we've done in class) is fair game, but you can also certainly pursue something else that's not listed, as long as you clear it with me first. Your work will culminate in a paper that's roughly 2000 to 3000 words (~4 pages single spaced), satisfying the following parameters.
- Present the problem clearly to a general audience, and explain both how AI and ethics factor into the issues at play.
- Present both the risks and possible mitigation strategies for these risks.
- You should include roughly 10 cited references from reputable sources (NOTE: In the examples below, I've provided links to some more popular magazines to whet your appetite, but I went to see some serious academic references as well). You may want to look at the references in Race After Technology or on Google Scholar, for instance. If you find something is behind a paywall, check Ursinus's interlibrary loan.
- At least two of your references should show contrasting viewpoints.
- At least one reference must be a technical academic paper, in the style of the sorts of things we've talked about in this class. Note that it's totally fine if you want to debunk this paper or point out its dangers! (e.g. the Kosinski 2018 paper on using convolutional neural networks to detect sexual orientation)
- Take a stance! What are your thoughts, and how would you back them up with your references?
You may partner up with one other person, though you are not required to do so.
The final paper will be due on Friday 12/15, and feedback to other students will be due by Sunday 12/17.
Grading
Below is a grading breakdown for the paper, which will make up 15% of your overall class grade
Category | Percentage of Grade |
Quality of References, according to specs above | 40% |
Thesis statement, overall clarity, and formatting (I don't care much about the format of references, but the paper should follow in a logical manner from intro, to thesis, to body, to conclusion) | 20% |
Quality of argument, with regards to ethical issues | 20% |
Correct and precise description of the AI systems at play at an appropriate level of technical detail for a general audience, but demonstrating insider knowledge learned from this course (examples could include demonstrating sufficient knowledge of supervised vs unsupervised vs reinforcement learning, or demonstrating a working knowledge of deep learning in the context of your application) | 20% |
Peer Feedback
To increase a breadth of knowledge and to keep the dialogue going, each student will provide reflections and constructive feedback on three other student papers from the class. If everyone in this class does an individual paper, this will also guarantee that it gets read by at least 3 people other than me. If everyone pairs into a group, then it will get ready by 6 people other than me. Feedback will be graded for completion, so long as it is thoughtful and constructive.
Suggested Topics
- AI and the Environment
- AI and Augmented Reality
- Self-driving cars
- Forbes: AI Ethicists Clash Over Real-World Aptness Of The Controversial Trolley Problem, But For Self-Driving Cars It Is The Real Deal
- Business Insider: "Elon Musk's worst nightmare"
-
AI and online advertising
- The Correspondent: The New Dot Com Bubbles Is Here, and It's Called Online Advertising
- AI Surveillance, data centers / government privacy
-
The ethics of recommendation systems
- CNBC: "A woman shared her tragic story of how social media kept targeting her with baby ads after she had a stillbirth"
- Netflix: The Social Dilemma
-
AI and healthcare
- IEEE Spectrum 2021: Andrew Ng X-Rays The AI Hype (about deep learning-powered radiology)
- Robot pets for dementia patients
- Fabbri 2019: Health Care Vendor Data Management in the Era of Big Data and Machine Learning
- Ars Technica: "UnitedHealth uses AI model with 90% error rate to deny care, lawsuit alleges"
-
Face recognition technology
- Nature News Feature 2020: The ethical questions that haunt facial-recognition research
- Buzzfeed: "Surveillance Nation A BuzzFeed News investigation has found that employees at law enforcement agencies across the US ran thousands of Clearview AI facial recognition searches — often without the knowledge of the public or even their own departments."
- Wired: The best algorithms struggle to recognize Black faces equally
- NYTimes: Wrongfully accused by an algorithm
-
Predictive policing
- The Verge: Heat Listed (how Chicago's predictive policing became a self-fulfilling prophecy)
- Race After Technology Ch. 2
-
The ethics and implications of AI-powered creativity
- Colt 2009: Computational Creativity: Coming of Age
- Futurism: A New AI Can Write Music as Well as a Human Composer: The future of art hangs in the balance.
- City.am: 20 times that AI has tried to make it in the music business
- Epstein 2020: Who Gets Credit for AI-Generated Art?
- Where Does the Buck Stop? Ethical and Political Issues with AI in Music Creation. Transactions of the International Society for Music Information Retrieval.
- Mystery AI Hype Theater 3k Ep. 4 - Is AI Art Actually "Art"? (ft. Johnathan Flowers, Jennifer Lena, & Negar Rostamzedeh)
- Wired 2022: "Picture Limitless Creativity at Your Fingertips (a more positive take)
- AI and music copyright at scale
-
AI and LGBTQ+ issues
- The Guardian 2017: "New AI can guess whether you're gay or straight from a photograph"
- Original 2018 Kosinski article on "Deep Neural Networks Are More Accurate Than Humans at Detecting Sexual Orientation From Facial Images"
- Rebuttal to the science of the Kosinski article by Margarte Mithcell, Blaise Agüera y Arcas, and Alex Todorov
- AI and emotion recognition
-
AI And Beauty
- Race After Technology Ch. 1
- http://beauty.ai/
-
AI in the military
- 2021 Forbes Article on Project Maven
-
Deep fakes, revenge porn, and female domination in AI
- Technology Review 2/12/2021: "Deepfake porn is ruining women’s lives. Now the law may finally ban it"
- Salon: "From Siri to sexbots: Female AI reinforces a toxic desire for passive, agreeable and easily dominated women"
- RollingStone: "An AI Nightmare Has Arrived for Twitter — And the FBI"
- Kara Kelleher from Boston University School of Law: "Revenge Porn and Deep Fake Technology: The Latest Iteration of Online Abuse"
- "Jailbreaking" and the difficulty of keeping large public generative models "on the rails"
- Isaac Asimov's law of robotics, and the definition of robotic "harm"
-
The ethics AI-powered therapy / other implications of generative AI on mental health
- Eliza: The Rogerian Therapist
- Vice: "Eating Disorder Helpline Fires Staff, Transitions to Chatbot After Unionization"
- Psychiatrist.com: "NEDA Suspends AI Chatbot for Giving Harmful Eating Disorder Advice"
- Mystery AI Hype Theater 3000, Episode 13 - Beware the Robo-Therapist
- Liberation: "Comme une drogue dans laquelle il se réfugiait» : ce que l’on sait du suicide d’un Belge ayant discuté avec une intelligence artificielle" ("Like a drug in which he took refuge," What we Know About The Suicide of A Belgian Man Based on What He Told An AI ChatBot)
-
The ethics of chat bots
- Mett Claira: The First Female Afro-Latina Bilingual AI Bot
- The Verge: Twitter taught Microsoft’s AI chatbot to be a racist asshole in less than a day
- QZ: Microsoft's Politially Correct Chatbot Is Even Worse Than Its Racist One
-
The ethics of AI news summaries / news generation
- Nature: tl;dr: tl;dr: this AI sums up research papers in a sentence
- CNET: Facebook's news summary tool
- The Vanderbilt Hustler: "Peabody EDI Office responds to MSU shooting with email written using ChatGPT"
-
AI in the legal profession
- Donahue 2018: A Primer on Using Artificial Intelligence in the Legal Profession
- Reuters: "New York lawyers sanctioned for using fake ChatGPT cases in legal brief"
- Algorithms in the gig economy
-
AI and the philosophy of religion
- Anderson 2020: "A New Hermeneutics of Suspicion? The Challenge of deepfakes to Theological Epistemology"
-
AI and lending decisions
- MIT Technology Review: Bias isn’t the only problem with credit scores—and no, AI can't help
-
AI and hiring decisions
- Futurism:You’re Hired! This Site Generates Random Neural Network Résumés
- The Verge: Automated hiring software is rejecting viable candidates
- The Guardian: How algorithms rule our working lives
- AI and armed conflict prediction
-
The Internet of Things and ubiquitous computing
- Wired: Be Healthy or Else: How Corporations Became Obsessed with Fitness Tracking
- AI and animal welfare
-
AI and the public health
- Nature: Millions of black people affected by racial bias in health-care algorithms
- "Mystery AI Hype Theater 3000, Episode 16 - Med-PaLM or Facepalm? A Second Opinion On LLMs In Healthcare"
- Trust and safety
- Openness vs closedness in AI Models
- Pitfalls in activism
- Other ideas: online dating algorithms, social credit, policing hate speech, smart device assistants (e.g. Amazon Alexa)