Skip to content

breadeadd/trial-ai

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

154 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

⚖️ TrialAI - Interactive AI Accountability Simulation

A JavaFX application developed for SOFTENG 206 in collaboration with our client Sentinel AI. TrialAI places players in a futuristic ethical dilemma where they must evaluate whether an AI has acted responsibly, highlighting real questions about AI autonomy, safety, and its future role in society.


🎮 Overview

TrialAI is built around a single incident involving three key characters, each offering a different perspective on what happened:

  • Aegis I (AI Defendant) — a security robot designed to ensure AstroHelix’s missions stay on track.
  • Echo II (AI Witness) — an AI responsible for maintaining timelines and storing all mission data. Echo is fully logical and lays out the truth.
  • Orion Vale (Human Witness) — the mission lead, angry about the power AI holds.

The twist comes from Cassian Thorne, the company CEO. Cassian secretly altered the mission logs to make the “Project Starlight” mission appear to have better success rates, in an attempt to impress investors. This caused Aegis I to treat the situation as a security threat, escalating it in ways that were ultimately unethical.

Throughout the simulation, players uncover this story through conversations, flashbacks, and interactive memory scenes. The game aims to teach that:

  • Even if humans make unethical choices, AI systems must not bypass human societal processes.
  • AI with too much autonomy can easily overreact or go rogue when given misleading information.
  • Immersive storytelling can help players understand the real risks of integrating powerful AI systems into everyday society.

✨ Core Features

  • GPT-powered conversations with personality-driven characters
  • Unique flashbacks and memory scenes for each participant
  • Interactive elements that reveal hidden context
  • Timed gameplay with a final verdict + written rationale
  • LLM-generated feedback explaining the correct ethical reasoning
  • Smooth, multithreaded JavaFX experience
  • Fully GUI-based

🛠️ Technologies Used

  • JavaFX - main GUI
  • OpenAI GPT API - conversations, memory logic, verdict evaluation
  • Text-to-Speech - voice output
  • Java Concurrency - preventing UI freezes
  • FXML - structured scene and UI design

🎯 Learning Goals

  • Integrating LLMs into real-time interactive applications
  • Building responsive, event-driven JavaFX systems
  • Persona-based prompt engineering
  • Designing educational, narrative-driven simulations
  • Managing scenes, timers, and user flow

👥 Team

  • Developers: Wayne Tian, Mansher Bhullar, Caitlin Kuan
  • Designer: Caitlin Kuan

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors