Week 4

When does a thing stop feeling like an object and start feeling like someone or something we owe attention to?

This week, we move from simulation to relationality and evaluation. Small cues such as motion, timing, memory, need, and response can make a system seem purposeful, vulnerable, or alive. Across artworks, games, toys, and social robots, intelligence is not only performed by a machine. It is also attributed and sustained through an encounter.

We will use the imitation game to test which cues make an audience attribute thought, understanding, personality, or care, and how those interpretations change when the system’s mechanism is revealed.

Class Notebook

Andrea Cheng

Last week, I learned that an interface can change how intelligent a system seems. Comparing my three versions showed me how visible data can clarify the rules, while personal comments and references to past actions can create a feeling of being understood. It made me question how much of that feeling comes from the design of the interaction.

Binjia Li

I learned that automatic color analysis still depends on lighting, sampling choices, and designer-made rules. My question is whether showing these limitations helps users evaluate recommendations more critically？

Jiayi Li

When does showing uncertainty make a system seem more intelligent, and when does it make it seem less capable?

Li Zhang

Acting intelligence is not about full solve the problem super quick. Sometimes hesitation and continuity would seems more intelligence.

Lulah Balde

How to thoroughly build an intelligence system and figure out the design process from scratch for this creative idea

Lulah Balde

consider previous submission answer

Mingfu Yang

The system didn't need to be legible to feel understanding, it needed to be unpredictable in ways that felt alive: memory, hesitation, silence.

Rebecca Davidson

Visible mechanisms can sometimes undermine the system by make the user very aware of the system's limitations.

Seulgi Choei

This made me realize that “How a machine speaks changes how we see it.”

Wendy Wang

This week I learnt that the bee communication game I made is also testing how people construct the idea of intelligence from what a system reveals and hides, connecting to Epistemological Research of Technology.

Xiran Wang

I learnt that rabbits burrowing will disrupt the soil structural integrity and they don't dig randomly. They are planning to build main corridors, nesting chambers, escape bolt holes, and refuge blind spots. Some complex structures are spontaneously built by simple rules with simple goals even without any central planners or commanders. The biggest lesson for me is about swarm intelligence.

Zoe Liu

After adding more random elements, the system began to feel slightly more human. When you repeatedly ask the same question as part of the experiment, it even seems to become genuinely annoyed. If the same question is asked five times or more, the deity will always respond with an angry expression. The first time I saw this happen, it was actually a little unsettling. However, the system still does not actually understand the question. Even so, these small responses can make the user feel that the system is reacting to them. If we amplify this kind of feedback and consider it in the context of people in the past who used these rituals to seek spiritual comfort or guidance, their reactions to such systems become much easier to understand.

Week 4: The Imitation Game

Kismet, an expressive social robot developed at MIT in the late 1990s. Photograph by Daderot, CC0.

Lectures

Readings

Resources

Assignment 4

This assignment is part 4 of the project Epistemological Research of Technology.

Submit Assignment

Last week, you made two versions of the same system: one that exposed its mechanism and one that hid it. Use the feedback from participants to make one final version of your prototype. Combine the strongest parts of the two Week 3 versions and find a meaningful balance between revealing the system and hiding its mechanism. Decide what people need to understand in order to interact with it, and what can remain uncertain enough for them to interpret its behavior.

Based on what you observed, decide what else to add, remove, or change to make the machine appear more like it thinks and understands the person interacting with it. You might work with memory, response timing, continuity, adaptation, surprise, hesitation, vulnerability, language, movement, or another behavioral cue. Do not add complexity only to make the project look more advanced. Each change should help you test what people read as intelligence, intention, personality, feeling, or care.

Project 1 is Epistemological Research of Technology, not simply a reconstruction of a historical system. Your project should move through this structure:

Reconstruction: How does it work?
Intervention: What happens if I change something?
Observation: What changed in the behavior or interpretation?
Claim: What does this reveal?

Making the prototype gives you something to investigate. Your observations and claim are what turn the making into research.

For next week, you will prepare a five-minute presentation that includes a live demo of your final prototype:

1 minute: Introduce your historical research, original epistemological question, and system diagram.
1 minute: Show the two Week 3 versions, explain what you changed between them, and summarize the feedback you received.
2 minutes: Present and live-demo the final combined prototype. Explain what you kept, what you changed, and how the final version responds to your observations.
1 minute: Focus on your main insight or claim, the evidence behind it, the assumptions or limits of your experiment, and one unresolved question.

Do not present only a chronological account of what you built. Take us from the historical research to the diagram, intervention, prototype, observation, and claim. Focus on what the experiment taught you about how intelligence or understanding is constructed and perceived.