This course introduces generative agents that simulate human behavior in an interactive sandbox environment. Students will learn about the core concepts of ChatGPT powered agents, inter-agent communication, user control, environmental interaction, memory and retrieval, planning, and reacting. The teaching method includes lectures, demonstrations, and discussions. The course is intended for individuals interested in human-AI interaction, generative AI, and large language models.
ChatGPT Powered NPCs in Sandbox RPG Game Smallville - Generative Agents Interactive Simulacra
Software Engineering Courses - SE Courses via YouTube
Overview
Syllabus
Introduction to ChatGPT powered agents having first ever made RPG game Smallville
What is Smallville SandBox RPG Game
The community of the ChatGPT powered Smallville game
Core concept of the run by the GPT engine Smallville game
Inter-agent communication
User control
Environmental interaction
Smallville sandbox environment implementation
Example day in life of ChatGPT powered agents
Why the game is still far away from being a full-fledged game
Each day a new plan is made
Emerging social behaviors
Information diffusion
Relationship memory
The hardest part of such ChatGPT powered agents having RPG game making
Workflow of the Smallville game
Generative agent architecture
Memory and retrieval - the core of the game
Memory stream architecture
Reflection memory
Planning and reaction
Reacting and updating plans
From structured world environments to natural language
Human role playing evaluation
The problems of the ChatGPT powered game
Ethics and societal impact
Demo of the game
Taught by
Software Engineering Courses - SE Courses
