hslu-ki-h2501

3.4 🔍 Discussion and Reflection — Thinking Like a Philosopher

Explore how perception differs between humans and AI systems — not through theory alone, but through short, practical and creative exercises. Each activity helps translate abstract ideas from George Berkeley and Maurice Merleau-Ponty into experiences or examples relevant to computing, robotics, and interaction design.

Activity 1 — Data or Experience?

Goal: Differentiate between processing information and perceiving experience..
Duration: 5–10 minutes.
Format: Pairs or small groups.

Steps

1. Read the following list of examples.
2. Decide whether each is closer to AI perception (data, computation) or human perception (embodied, experiential).
3. Add to the list

Examples:

• Recognizing a cat in an image
• Feeling the cat’s fur
• Detecting sadness in a voice recording
• Feeling sad because of the tone
• Navigating a dark room
• A robot mapping a room with lidar
• …

Debrief

Reflect on

“Where do you draw the line between perception and data processing?”

Then on to:

• Berkeley’s view: perception creates the world.
• Merleau-Ponty’s view: perception happens through the body.

Activity 2 — If the Sensors Go Dark

Goal: Apply Berkeley’s idea that “to be is to be perceived.”.
Duration: 5–8 minutes.
Format: Small groups.

Setup

Consider this scenario:

“Imagine an AI vision system trained on millions of images.
One day, it’s turned off — all sensors disconnected, no data coming in.
Does its world still exist?”

Steps

1. In group write a two-sentence response:

   • “Yes, because…” or “No, because…”

2. Use technical reasoning if possible (data storage, model memory, computation).

   • Example: “Yes, its parameters still encode a version of the world.”
   • Example: “No, without active perception, there is no world at all.”

Debrief

Discuss a few answers. Then connect to Berkeley:

“He believed that when perception stops, the world disappears. For AI — does data storage count as ongoing perception?”

Activity 3 — Embodied or Disembodied?

Goal: Experience Merleau-Ponty’s idea that perception is embodied and situated..
Duration: 7–10 minutes.
Format: Whole class or small groups.

Choose one of the following formats depending on your space and time.

Option A – Blind Navigation

1. One student closes their eyes while a partner guides them 3–4 meters using only voice instructions.
2. Swap roles.

Discuss:

• How did you perceive distance or direction without vision?
• What replaced your usual sensory information?

Option B – Touch Recognition

1. Place small objects (USB drive, apple, coin, marker, etc.) inside an opaque bag.
2. Students identify them by touch.

Discuss:

• What kind of knowledge did your body provide?
• Could an AI replicate that form of understanding through sensors?

Option C – Design Analogy

1. Pose the challenge: “Design a robot that must find and pick up a coffee cup in a real room.”

2. Outline (verbally or on paper):

   • What sensors it needs
   • How it would move
   • How it would know it succeeded

Discuss:

• “Would this robot see the cup, or would it feel its way to it?”

💬 Activity 4 — AI as a Philosopher (LLM Integration)

Goal: Use an AI language model (e.g., ChatGPT, Perplexity, Claude) to explore how machines interpret philosophical ideas.
Duration: 10–15 minutes.
Format: Individual or pairs.

Steps

1. Open an AI chatbot.

2. Give it one of these prompts:

   • “You are George Berkeley. Explain AI perception in your own words.”
   • “You are Maurice Merleau-Ponty. What do you think of robots that can feel touch?”
   • “You are an AI philosopher combining both ideas — define perception in one sentence.”

3. Copy or paraphrase 2–3 interesting responses.

4. Then answer these short questions:

   • What does the AI assume about perception?
   • Does it understand or just recombine ideas?
   • Which answer sounds most “human”?

Debrief

Invite a few to share their favorite “AI philosopher quote.” Then ask:

“Did the LLM truly perceive the philosophers’ ideas, or did it just simulate them?”

→ Does AI perceive, or does it merely process?

🪜 Recommended Flow (30–35 minutes total)

Time	Activity	Focus
5–7 min	Data or Experience?	Warm-up comparison
8 min	If the Sensors Go Dark	Apply Berkeley
10 min	Embodied or Disembodied?	Apply Merleau-Ponty
10 min	AI as a Philosopher (LLM)	Synthesis + reflection

Optional closing question: “If perception is what makes a world real, what kind of world does AI live in?”