Week 4: How Do We Talk to Machines?

This week we move on from the more technically intensive readings to think about the question: how do we talk to machines?

This question is, on the one hand, the topic of a thousand social media posts promising to teach you everything you need to know about the new discipline of prompt engineering. On the other hand, it also opens up a more profound space for exploring what it means to talk at all: with machines, with each other, and with ourselves.

The Practical Resources are likely to be useful for learning techniques and strategies for doing things with AI. Some, like the Park and Choo article, will be useful as overviews of prompting strategies, and it is also worth noting the links at the end of that article. Others, like the Karpathy tweet and Anthropic’s Cowork article, cover emerging approaches to using AI for research or information synthesis.

The Theories about Practice go in a different direction. These aim to stretch our thinking about “talk” as a performance (Austin, 1962), as a defiant act of truth (Foucault, 2020), and as user interaction in the context of AI (Weizenbaum, 1966; Mollick & Mollick, 2023; Magee et al., 2024; Magee, 2026).

I want first to reflect a little on J. L. Austin’s seminal text, How to Do Things with Words. This book, a series of lectures, instigated attention to the pragmatics of language and inspired, among others, Searle’s Chinese Room thought experiment. Again, I do not suggest reading this end to end, but rather focusing on getting some conceptual understanding of Austin’s distinctions between:

Perhaps the question for us is: How do we get AI to do things with words?

Then, with Ethan and Lilach Mollick’s paper, we will examine how different prompts can be used to play roles on the pedagogical “stage.” That idea is reworked in two long papers by myself and other colleagues. For those interested in treating AI as artificial “actors,” I do not suggest reading these through in full, but rather focusing on the principal idea: that LLMs have internalized vast amounts of information about how people perform with language, and that this internalization can simulate quite complex character enactments.

Both papers are, in some ways, throwbacks to the early days of AI, when psychoanalysis and dramaturgy were more prominent theoretical rubrics. As if on cue, we then look at a classic AI paper, Weizenbaum’s 1966 presentation of ELIZA: a system that simulated a certain therapeutic approach, to surprising effect given the technical limits of the time.

Finally, for those with theoretical inclinations, I have included Michel Foucault’s work on Discourse and Truth and Parrhesia. This is perhaps more of a placeholder for discussions in future weeks, but Foucault’s analysis of parrhesia in classical Greek and Roman discourse holds special relevance for how we view AI’s social role: as truth-teller, as performer, as stochastic parrot, as assistant, and so on.

Several of these texts also move across traditions and vocabularies that not all of you may be familiar with, and I would encourage you to supplement your reading with queries to AI itself. For example:

Practical Resources

Theories about Practice

Assignment(s)

In your respective Google Sheet tab, respond to one or more of Week 4’s questions.