Maryyann Crichton

Wetware

Co-Director / Animation Lead / Art Director

Project Goals

What is your definition of consciousness when applied to AI technology? Our goal is to open up discussions around AI ethics.

Game Website

Unreal Engine 4 / Maya / Substance Painter / Substance Designer

Download Game

Summary

Wetware is my MFA thesis. It is a first-person game about the interplay between AI ethics and consciousness. You are a lab scientist tasked with probing information from a captured enemy drone. Your country is at war and you're the only hope. As you work to uncover enemy plans, you learn there's more secrets in the lab than just those from the enemy drone.

research & concept dev worldbuilding experience flow texturing playtesting

Research & Concept Development

I started the project by researching existing games and opinions around AI ethnics. We explored texts by philosophers like Daniel Dennett and his writings on free will. We also interviewed notable scientists like David Hanson at Hanson Robotics about his thoughts on AI ethics and general intelligence. This initial research helped us build our game goals.

Interaction and Theme

Our goal was to create an experience that explores the moral implications behind interacting with AI agents through different perspectives. To tackle this topic, we created experience goals and interaction goals to drive our decisions.

research notes ; determining experience goals

Experience goals :

1. Evoke moral conflict through gameplay

2. Create characters with opposing opinions

3. Open discussions about AI ethics

Interaction Goals :

1. Emotional weight through repetition.

2. Your decisions affect your coworker's trust.

3. Small interactions have larger implications.

We were inspired by Papers Please, which emphasizes the simple interaction of stamping "approved" or "denied" to determine if an immagrant is accepted into the country. In Wetware, players repeatedly probe the drone everyday. It combines the weight of moral and civic reponsibility with the emotional numbness from repitition as you directly determine the fate of lives.

Players speak with items in the lab, rather than inspecting them. In a world filled with "intelligent" agents, it is easier to accept the enemy drone's level of intelligence.

interaction inspiration: Papers Please

Worldbuilding

The Laboratory and Characters

The first version of the lab contained rectangular rooms connected by hallways, but players often got lost. I upgraded our design into a simpler circular design with a single hallway. The final design of the lab is in the shape of a cell with the player working with the robot in the center nucleus.

3D mockup of the lab

Sketches of the interiors of the lab

Each character has 3 important traits — opinion on AI ethics, characters stakes, and character trust.

We define a character's opinion on AI ethics from the start of the game. Their trust is determined by dialogue responses from players and willingness to help characters resolve issues at stake. Aiding these characters allow players to build up empathy.

I also created character behavior trees which determine the progression of a character's actions. The map below shows the changes in the robot's reaction depending if players decides to probe or not probe the robot. This development creates a more realistic and believable world.

Day 1 - 2 interaction behavior tree for robot

Day 3 - 4 interaction behavior tree for robot

Experience Flow

Story Flow

I created several versions of story flow to show the different stages of the Wetware game.

Throughout the process, I continued to reference our "experience goals". How do we introduce conflicting moral decisions on AI ethics? Does the player have enough context to feel like they have agency? How do we encourage the player to empathize with this world?

Story flow iterations with pain points & improvements

Key Issues

• Players lacked context about the story and became unmotivated.

• Players become indifferent to theme after repeatly making difficult decisions.

• Players felt emotionally disconnected from the robot.

Solutions

• I added an introduction animation about the state of the war. We found players were much more motivated to complete their tasks after this change.

• I added the non-interactive dreamscape with interviews from notable researchers on AI ethics. This gave players a moment to self-reflect.

• I included additional reaction animations on the robot, directly connected to a player's actions. This produced a real emotional reaction from players.

Day 1 Tutorial Experience Flow

Day 1 Tutorial teaches players all of their tasks and interaction mechanics.

After making initial sketches, I transferred the sketches into an experience flow wireframe that marks the important mechanic introduction points coupled with key events in the day. This was then mocked up in 3D inside of Unreal Engine and Maya.

experience flow of the day 1 tutorial

IPAD Interaction Sketches & Wireframe

The IPAD is the player's inventory, store, and escape. Players access this IPAD often, so it's crucial to make easily navigable. I created quick prototypes using sticky notes and built them into wireframes.

IPAD layout and interaction sticky pad sketches

Design Goals:

1. Large icons with Simple UI.

2. No data can be erased.

3. Viewing the IPAD during work hours will NOT stop the clock.

The last point calls back to our 3rd interaction goal — small interactions have larger implications. Spending time on the IPAD instead of working directly influences the trust of your co-workers.

Texturing

One major visual reference was the aethestics of stop motion puppetry. Similar to how puppets require a puppet master, AI robots require a developer. This theme influences everything from visual design to animation to texturing.

Texture Guidelines

For each character, I created texture guides that dictated the number of types of material assets in the game.

Using the texture guides, I textured all of the assets in the game. This included procedural, static, and interactive materials in substance painter and Unreal to achieve our final look.

Final Look

Coupled with simple soft lighting, we achieved a tactile visual style in the final game.

Playtesting

A final step for our game was making rounds through several national conferences, including PixelPop, Indicade, and Dreamhack. At each conference, we conducted a Playtest Questionnaire to collect feedback on player experience. This feedback informed the last iterations of changes.