Week 2 • Language & Intelligence

Module Questions

How is language connected to thought, memory, and improvisation?
What types of intelligence exist—and how do machines model them?
What can spontaneous language play reveal about how humans and machines "think"?
How do we define intelligence through processing and output, and what’s missing in machine forms?

To Do By Class

Readings:

Follow-Up: Stories of Entanglement

Purpose: Extend your in-class exploration into a free-flowing conversation with ChatGPT to probe human–machine entanglement. Notice how the AI’s role, tone, and your own mode of communication (voice vs. typing) shape the exchange.

Setup

Sign in to the same ChatGPT account you used in class so your chat is saved to your history.
You may type or dictate your starting journal entry (either is fine).
For the conversation itself, you may:
- Use voice only (encouraged) for a more free-flowing exchange, or
- Type if you prefer written reflection.

Conversation Flow (15–20 minutes)

Begin with your story: Start from a personal moment of machine entanglement (you can dictate or type a short paragraph based on your journal).
Assign the AI a role (experiment): Ask it to act as a philosopher, future historian, skeptic, inner voice, or friend. Try at least one role; two is better. Notice how the role shifts the dialogue.
Probe deeply: Explore concerns/fears about AI, how the AI distinguishes human from machine, and how it characterizes the “machine-entangled human.” Push back, ask for examples, and follow threads.
Wrap-up inside the chat: Ask the AI to summarize your conversation and generate three questions for further inquiry.

After the Chat

Ensure it’s saved: Confirm the conversation appears in your ChatGPT history (you must be signed in).
Return to your analog journal: Write a short reflection or summary responding to:
- How did using voice vs. typing affect the flow and depth?
- What moments of entanglement (connection, resistance, misunderstanding) stood out?
- How did the AI’s assigned role change the conversation?
- Reflect on the difference between the student group discussions and the AI discussion.

No deliverables this week. We will build on your experiences in next class’s discussion. Keep your chat in history and your journal notes handy.

Language Games

In this class, we’ll explore playful language games that surface how human communication relies on pattern recognition, memory, spontaneity, and cultural familiarity. These activities serve as warm-ups for understanding how large language models process and reproduce language differently. Each game may later be re-run using ChatGPT to highlight contrasts in creativity, nuance, and repetition.

1. Association Chain

One student says a word, the next responds with the first word that comes to mind, and so on. After 10–15 links, reflect on where the chain started and ended. (you can record the session in a chat to dictate)

Discussion Prompt: What shaped your associations? Habit? Surprise? Cultural scripts?

2. "Fortunately, Unfortunately"

Students take turns continuing a shared story, alternating between “fortunately” and “unfortunately” clauses.

Example: "Fortunately, the robot made me breakfast. Unfortunately, it mistook soap for eggs..."

Discussion Prompt: How do rhythm, contrast, and expectation shape narrative logic?

Slide Talk with Activities

Technical Overview: How LLMs Work

An overview of the technical architecture of large language models:

Tokenization and probabilistic prediction
Transformer architecture and self-attention
Training on massive datasets without grounding
Output as prediction, not understanding
Strengths and weaknesses: fluency, pattern mimicry, hallucinations

AI/Human Learning Comparison

Use ChatGPT to generate two short essays: one on a topic you know well, one you know little about.
Compare clarity, voice, accuracy, and personal resonance.
Ask ChatGPT reflective questions: How was this written? Why these choices?
Begin your AI journal: note what surprised you, what helped, what failed, and what felt human.
Continue journaling outside class with dictation or writing. Save it for reference in future projects.

Create Your Own Custom GPT

Goal: Design a GPT that reflects your interests, passions, or curiosities.

Instructions

Pick a Focus: Choose a topic, role, or purpose for your GPT (e.g., Poetry Coach, Fitness Guru, Dungeon Master, Meme Maker).
Define the Role: Write a short description of what your GPT “is” and how it should act.
Example: “You are a jazz musician who always explains concepts through musical metaphors.”
Set Behaviors: Decide how your GPT responds:
- Formal or casual?
- Short answers or long explanations?
- Playful, critical, or supportive tone?
Give Example Prompts: Write 2–3 sample prompts a user might ask your GPT. Show the kind of interaction you expect.
Add Constraints or Rules (optional): Add boundaries (e.g., “Never use technical jargon” or “Always rhyme in your answers”).
Add contextual documents Upload samples of your writing, documents, spreadsheets and other data.
Test It Out: Try your description in ChatGPT (or another LLM) to see how close the responses are to your vision.
Reflect: Write a few sentences: What worked? What surprised you? How does this GPT reflect your interests?