WhoDunit - AI game development on steroids
AI Summary
Summary: Who Done It Game Development
- Game Overview:
- Title: Who Done It
- Concept: Advanced prompt engineering with AI
- Development: Pair programmed with GPT
- Content: All generated by GPT (storylines, pictures, characters)
- Characters: Roleplayed by GPT
- Database: AI-powered administration system
- Gameplay:
- Players pick a mystery (e.g., the stolen clock)
- Interrogate suspects and solve the mystery
- Game loop: Talk to characters, ask questions, search crime scenes
- Mystery Generation:
- Mysteries are AI-generated
- Players can choose a theme or let GPT get creative
- Advanced prompting techniques used to generate content
- Roleplaying and Interrogation:
- GPT roleplays characters during interrogation
- Players can interact with characters and the environment
- Admin Interface:
- Admin tasks performed through chat with AI
- AI handles database queries and admin functions
- Cost:
- Using GPT-4 is expensive ($230 for the demo)
- GPT-3.5 is cheaper but less effective
- Development Insights:
- Productivity significantly increased with AI assistance
- AI helped with design, architecture, coding, debugging, UI design
- New types of products can be built with AI technology
- Game Mechanics:
- Accuse characters and convince a police officer
- Game master ensures the story remains coherent
- Players can interact with the game master for hints or to change the game
- Technology Stack:
- Front-end: Next.js, React (new to the developer)
- AI as a “genius coach” for coding assistance
- Conclusion:
- AI-driven development offers a huge productivity boost
- Enables creation of innovative products previously not possible