Imaginary Syllabi: AI for Liberal Art Students
[ai
imaginary-syllabi
llm
teaching
mathematica
]
A collection of reading lists for introducing AI concepts and practice to a general population of starting undergraduates…
(with the idea of building up materials that could be used in a variety of courses; perhaps an EP1-style introductory course at Fordham; resources are generally Mathematica-heavy)
Motivations
- “Liberally-educated students need to be more than consumers of AI” (2023)— students should actively engage with AI as creators and critics (and on a technical level), rather than just passive consumers, in order to navigate a future increasingly dominated by AI. The best way to do this is to have some fundamental mathematical/statistical/computational chops, so they can be critical and adaptable.
- Matteo Pasquinelli The Eye of the Master: A Social History of AI — Did you know that Charles Babbage was best known in his lifetime as a labor efficiency planner (which influenced his computing)…and that Karl Marx rips of Babbage and other English factory efficiency thinkers? Or the role of the Austrian School of Economics on inventing neural networks. Oh yes…and more…. (particularly interesting for my other hobby of laboratory automation)
- Adrienne Mayor Gods and Robots: Myths, Machines, and Ancient Dreams of Technology—e.g., the Golden Kourai of Hephaestus as AI agents—as a way to think about automatons through mythology
- LLMs as a practical data point for discussing philosophy of language—“You shall know a word by the company it keeps”, late Wittgenstein’s use theory of language, etc.
Computational Thinking
- Wolfram, An Elementary Introduction to the Wolfram Language
- Christian & Griffiths, Algorithms to Live By: The Computer Science of Human Decision – how to apply ideas of computer algorithms to solve everyday life problems
Machine Learning
- Etienne Bernard’s Introduction to Machine Learning (in the Wolfram language)
AI
- Pre-reading: Bottou & Schölkopf, Borges and AI
- Christian, The Alignment Problem: Machine Learning and Human Values
- Wolfram What Is ChatGPT Doing … and Why Does It Work? … combine with 3blue1brown videos … or perhaps instead a nontechnical 25 minute history of NLP ideas through GPT as an appetizer
- Intellectual history of mathematical representations of words and concepts (from Raymond Llul, Athanasius Kircher, Leibnitz, etc.) –> word/sentence embeddings
- Try some practical things regarding text, such as basic prompt engineering, summarization methods, retrieval augmented generation methods
- Step-by-step guide to building and training a GPT model from scratch on the Shakespeare play corpus
- Philosophical discussion: How much of what we attribute to “reasoning” and “planning” in daily life is just simple pattern matching? And how might we be fooled into think a pattern-matching LLM is “reasoning”?
- Use the language of role-playing as a framework for discussing dialgoue agents, so as to allow us to use folk psychological concepts (beliefs, desires, goals, ambitions, emotions, etc.) without falling into the trap of anthropomorphism. LLMs as non-deterministic simulators capable of role-playing an infinity of characters. See Shanahan et al Nature 2023 — good starting point to motivate deeper exploration of personal identity, Baudrillardian simulcrae, Everettian multiverse models, etc.
Applications
Guest speakers
- Friar Paolo Benanti, TOR: vatican advisor on AI — alas, it appears his writings are all in Italian
- Dennis Yi Tennen @ Columbia, author of Literary Theory for Robots (2024)