Week 3

When does a simple set of rules begin to look like behavior?

This week, we will move from **representation to simulation**. A model never reproduces the world completely. It selects certain relationships, states, differences, and behaviors and treats them as sufficient. By studying random walks, automata, cellular systems, flocking, collective behavior, and other simulations, we will ask what a model preserves, what it leaves out, and why we nevertheless recognize its behavior as meaningful, lifelike, or intelligent.

Class Notebook

Andrea Cheng

I learned that the way a system looks and responds can make simple rules seem intelligent or personal. I’m curious about how much our trust comes from understanding how a system works, and how much comes from how it presents itself.

Binjia Li

This week, I learned that the same computational rule can be presented as either a transparent calculation or a personal interpretation. Building both versions made me distinguish what the system actually computes from what its language suggests it understands. I have not yet tested how participants respond. My question is: does revealing the mechanism increase trust, reduce perceived intelligence, or do both and how might seeing one version first influence the other?

Jiayi Li

This week, I realized that whether a system feels “intelligent” does not necessarily depend on how complex its underlying logic is. This made me wonder: If users are less aware of how much they contributed, are they more likely to give the system credit for the final answer?

Li Zhang

/

Lulah Balde

A question I still have is how to make my system as intelligent as possible but also as abstractly creative and pleasing to look at + interact with.

Mingfu Yang

If everything a companion animal currently provides — comfort, response, even something like touch or memory, even a two-way sense of being understood — could eventually be reproduced by something built specifically to provide it, would an electronic pet become the better choice?

Rebecca Davidson

The system's apparent ability to translate human activity into data and then back into life-like feedback is actually what makes the system feel most intelligent, even if it operates on just a simple algorithm.

Samridhi Jain

I learnt how to code using objects this week. I also learnt that intelligence can be scarier when you don't know how it operates or works. After multiple versions of intelligence being scary, I'm confident that it is something to fear. I have further questions on how intelligence can be described as scary.

Seulgi Choei

When a machine talks like a person, do we think it understands what it sees?

Wendy Wang

This week, I learned that hiding the rules can make a simple system feel more intelligent. What surprised me most was that the second version of my bee game felt more mysterious, even though it used almost the same rules. It made me realize that intelligence can depend on how we experience and interpret something.

Xiran Wang

This week, I realised that swarm intelligence can be a form of intelligence without a central planner or a commander. It's more about emergence and decentralisation. Simple rules can eventually form complex behaviour. What an individual has done can be simple, but multiple individuals can finish a huge and complicated project.

Zoe Liu

In this system it is quite easy to get the same result several times in a row. People using Version 1 might think that the system just happened to roll the same random number, since there are only four possible combinations. But people using Version 2 might see it as fate. If they keep getting “yes”, they might feel that they are somehow favored by fate. If they keep getting “no”, they might even become discouraged because of it. But in reality, the system does not understand what they are asking at all. This also brings me back to a question I had before. If someone becomes skilled enough to consistently throw the Jiaobei and get the answer they want, does that answer still count as an answer? Or does it become a performance made for others, where the result has already been decided in advance?

Week 3: Ghost in the Machine

[Ghost in the Shell](https://en.wikipedia.org/wiki/Ghost_in_the_Shell) is a 1995 Japanese animated science fiction film based on the manga of the same name by Masamune Shirow. Created by Mamoru Oshii, the film is set in a futuristic world where cybernetic enhancements and artificial intelligence are commonplace.

Ghost in the Shell is a 1995 Japanese animated science fiction film based on the manga of the same name by Masamune Shirow. Created by Mamoru Oshii, the film is set in a futuristic world where cybernetic enhancements and artificial intelligence are commonplace.

Lectures

Workshop

Resources

Assignment 3

This assignment is part 3 of the project Epistemological Research of Technology.

Submit Assignment

Return to the small working model you built last week. This week, make two versions of the same underlying system to study how representation and interface affect the way its behavior is interpreted.

In the first version, make the mechanism legible. Let someone see the categories, rules, decisions, inputs, or transformations that produce the output. For instance, you might show the state of the system at each step, display the rules that govern its behavior, or provide a way to trace how one input leads to another output. The goal is to make the underlying logic of the system clear and understandable.

In the second version, keep the underlying logic substantially the same but hide the mechanism. Make the system appear to behave intelligently, lifelike, or meaningful without revealing how it works. You might use a more abstract interface, a different set of symbols, or a more complex surface that obscures the underlying rules. Could you make the system appear to have a mind of its own, even though it is still following the same rules as the first version? Could you make it “think”, or “feel”, or “decide” in a way that seems more human, even though it is still a simple set of rules?

The goal is not simply to make the second version more polished. Use the pair to ask what changes when a designed surface performs qualities the underlying mechanism may not possess. Document both versions and compare them.

Bring both versions, your documentation, observations, and question to class next week. We will use them to prepare for the imitation game and consider how an audience might evaluate a system from its behavior alone.