AI Ethics w/ Prof. Casey

← All topics  ·  Topic 1 of 10

🤖 What is AI and what can it do?

In 1950, computer scientist Alan Turing proposed "the imitation game," which became known as the Turing Test: a way to test a machine's ability to display intelligence like a human. Since then, what we refer to as "artificial intelligence" has kept evolving, and an exact definition is a challenge even for experts! Here you'll find all manner of "AI" in which a machine performs a task we'd typically think of as requiring human intelligence, from recommendation algorithms to predictive machine learning to large language models to robots.

🎥 Origin of the term "artificial intelligence" (the video above)
July 13, 2025 · YouTube · TikTokThe term "artificial intelligence" first appeared in the proposal for the 1956 summer workshop at Dartmouth that brought together researchers like Marvin Minsky, Claude Shannon, John McCarthy, and Nathaniel Rochester. This workshop has been widely considered the origin point of the AI research field.🎥 Update that further explains narrow vs. general AI (February 23, 2026) · TikTok📖 McCarthy, John; Minsky, Marvin; Rochester, Nathan; Shannon, Claude (1955). A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence.
Solomonoff, Grace (2023). "The Meeting of the Minds That Launched AI." IEEE Spectrum.

🎥 The Turing Test
April 11, 2024 · YouTube · TikTokThe Turing Test framed the question "Can machines think?" as an "imitation game": whether a human could converse with a machine without realizing it was a machine. The original game also involved gender impersonation, which roboticist Ayanna Howard highlighted raises questions about human bias as well.📖 Turing, Alan. Computing Machinery and Intelligence. Mind: 433-460 (1950).
The Turing Test (Stanford Encyclopedia of Philosophy, revised 2021).
Ayanna Howard Discusses Robotics and Unconscious Bias. Grace Hopper Celebration (2018).

🎥 Can Machines Think?
January 19, 2026 · TikTokExploring the question "can machines think?" from Turing's pragmatic test to a more philosophical exploration (re: symbolic AI) of Searle's Chinese Room thought experiment.🎥 Follow-up: the "deep sea octopus" thought experiment · TikTok📖 The Chinese Room Argument (Stanford Encyclopedia of Philosophy).
Bender, Emily M., and Alexander Koller. "Climbing towards NLU: On meaning, form, and understanding in the age of data." ACL, 2020.

🎥 The OG chatbot ELIZA
December 15, 2022 · YouTube · TikTokIn 1966, computer scientist Joseph Weizenbaum introduced ELIZA, one of the first computer programs to simulate human conversation. Using simple pattern matching and substitution rules, ELIZA could mimic a psychotherapist by reflecting users' statements back to them as questions. (Discussion of ELIZA continues in Social media and social AI.)📖 Weizenbaum, J. (1966). ELIZA: a computer program for the study of natural language communication between man and machine. Communications of the ACM 9(1), 36-45.

🎥 Generative AI vs Machine Learning
April 8, 2024 · YouTube · TikTokGenerative AI refers to machine learning models that can create new content (like text, images, audio, or video) based on patterns learned from large datasets. Examples include ChatGPT, which generates human-like text, and Midjourney, which produces images from text prompts. These systems are a subset of machine learning, in which models recognize patterns, make predictions, or classify data.📖 Mark Riedl, A Very Gentle Introduction to Large Language Models without the Hype (2023).

🎥 Bias in Machine Learning
February 26, 2022 · YouTube · TikTokBased on how machine learning works, a simplistic explanation of how bias can happen. (Much more in Bias and fairness.)

🎥 What is a large language model (LLM)?
April 25, 2024 · YouTube · TikTokExplaining it like you're five!

🎥 How does a language model work? (series) · Full YouTube playlist
An attempt to explain how LLMs work without hype and without over-simplification: something between "they're just a next word predictor!" and "now you've taken a machine learning class!"1. The Basics (Feb 23, 2026) · TikTok · YouTube
2. Self-Supervised Learning (Feb 28) · TikTok · YouTube
3. Vector Space and Neural Networks (Mar 2) · TikTok · YouTube
4. Self Attention (Mar 4) · TikTok · YouTube
Interlude: Why You Should Learn About This (Mar 5) · TikTok · YouTube
5. Transformers (Mar 7) · TikTok · YouTube
6. Emergence (Mar 11) · TikTok · YouTube
7. Instruction Tuning (Mar 14) · TikTok · YouTube
8. Reinforcement Learning (Mar 17) · TikTok · YouTube
9. Context Window (Mar 19) · TikTok · YouTube
10. Hallucinations (Mar 21) · TikTok · YouTube
11. Retrieval Augmented Generation (Mar 23) · TikTok · YouTube
12. Tokens (Apr 1) · TikTok · YouTube
Interlude: How Many R's in Strawberry (Apr 2) · TikTok · YouTube
13. Chain of Thought (Jun 6) · TikTok · YouTube
14. Prompts (Jun 12) · TikTok · YouTube📖 Heavily inspired by Mark Riedl's posts: A Very Gentle Introduction to Large Language Models without the Hype (2023), The Intuition Behind How Large Language Models Work, Part 1 and Part 2 (2025).
Also: King - Man + Woman = Queen (MIT Technology Review, 2015) · Vaswani et al., "Attention is all you need" (NeurIPS 2017) · Wei et al., "Emergent abilities of large language models" (2022) · Schaeffer et al., "Are emergent abilities of large language models a mirage?" (NeurIPS 2023).

🎥 What is artificial general intelligence (AGI)?
June 12, 2025 · YouTube · TikTokIn contrast to "narrow" AI, artificial general intelligence typically refers to AI that has human-like abilities across all types of tasks.📖 Morris, Meredith Ringel, et al. Levels of AGI for operationalizing progress on the path to AGI. ICML, 2024.

🎥 Why ChatGPT is not a search engine
June 20, 2025 · YouTube · TikTokLarge language models like ChatGPT don't work by searching for information over their training data. Remember that they are designed to be statistically probable and linguistically fluent, not verifiably accurate!🎥 Follow-up: AI-powered search engines can still hallucinate · YouTube · TikTok📖 Jazwinska, Klaudia and Aisvarya Chandrasekar. AI Search Has a Citation Problem. Columbia Journalism Review, March 2025.

🎥 Why ChatGPT can't tell time
April 4, 2026 · YouTube · TikTokSam Altman was asked in a podcast interview to react to a viral video from TikToker Husk in which ChatGPT in voice mode had "lied" about its ability to time a run. The example demonstrates a technical challenge with (non-)use of tool calling, but Altman's response also suggested a misunderstanding of the more important problem: the "lying."📖 Sam Altman Says It'll Take Another Year Before ChatGPT Can Start a Timer (Gizmodo, April 2026).

🎥 The challenge of image models
April 4, 2026 · YouTube · TikTokWhy does an image model like the one connected to ChatGPT struggle with text (using the example of visual alphabets)?

🎥 Why you can't just ask an LLM to not hallucinate
July 22, 2025 · YouTube · TikTokEven when explicitly prompting ChatGPT to only include reliable, verified searches, and even when it conducts a web search, it might still fabricate information and citations.📖 Huang, Lei, et al. A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions. ACM Transactions on Information Systems 43(2), 2025.
A.I. Is Getting More Powerful, but Its Hallucinations Are Getting Worse (The New York Times, May 2025).

🎥 ChatGPT (Year One) Wrapped
November 30, 2023 · TikTokA year of ChatGPT in the news.📖 Casey Fiesler, ChatGPT Wrapped (An AI's Year in Review) (2023).

🎥 A Taxonomy of Risks of Large Language Models (7-part series)
The 2022 paper "Taxonomy of Risks Posed by Language Models" laid out six categories of actual and anticipated risks. This series maps those (pre-ChatGPT) risks to actual harms that occurred between 2023 and 2025.1. Discrimination, hate speech, and exclusion (Sep 20, 2024) · YouTube · TikTok
2. Information hazards (Oct 2, 2024) · YouTube · TikTok
3. Misinformation (Jul 27, 2025) · YouTube · TikTok
4. Malicious uses (Nov 16, 2025) · TikTok
5. Human interaction harms (Nov 17, 2025) · TikTok
6. Environmental impact (Dec 2, 2025) · YouTube · TikTok
7. Socioeconomic harms (Dec 20, 2025) · YouTube · TikTok📖 Weidinger, Laura, et al. "Taxonomy of risks posed by language models." ACM FAccT, 2022. News examples for each risk are collected in the original syllabus doc.

📚 Further readingNarayanan, Arvind and Sayash Kapoor. AI Snake Oil: What Artificial Intelligence Can Do, What It Can't, and How To Tell the Difference. Princeton University Press, 2024.Riedl, Mark. A Very Gentle Introduction to Large Language Models without the Hype (2023).Hagendorff, Thilo, and Katharina Wezel. "15 challenges for AI: or what AI (currently) can't do." AI & Society 35(2), 2020.Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. "On the dangers of stochastic parrots: Can language models be too big?" ACM FAccT, 2021.Krafft, P. M., et al. "Defining AI in policy versus practice." AAAI/ACM Conference on AI, Ethics, and Society, 2020.Khurana, Anjali, Hariharan Subramonyam, and Parmit K. Chilana. "Why and When LLM-Based Assistants Can Go Wrong." IUI, 2024.Long, Duri, and Brian Magerko. "What is AI literacy? Competencies and design considerations." CHI, 2020.Rehak, Rainer. "AI Narrative Breakdown. A Critical Assessment of Power and Promise." ACM FAccT, 2025.

• Next topic: 📊 Information literacy & mis/disinformation
• All topics