Week 1: Introduction to Agentic AI
Agent definitions, ReAct paradigm, autonomous systems overview
Week 1 of 12
Learning Objectives
- Define what constitutes an AI agent
- Implement a basic ReAct agent from scratch
- Understand the agent trajectory structure
Topics Covered
- Agent definitions and taxonomy
- ReAct paradigm (Yao et al., 2023)
- Thought-Action-Observation loop
- Agent vs. standard LLM inference
Resources
Jupyter Notebooks
Required Readings
| Paper | Authors | Year | Link |
|---|---|---|---|
| ReAct: Synergizing Reasoning and Acting in Language Models | Yao et al. | 2023 | arXiv |
| A Survey on Large Language Model based Autonomous Agents | Wang et al. | 2024 | arXiv |
Reading Guide: ReAct Paper
Critical reading of the foundational ReAct paper introducing reasoning-acting paradigm
Primary Paper
ReAct: Synergizing Reasoning and Acting in Language Models
Yao, S., Zhao, J., Yu, D., et al. (2023)
ICLR 2023 arXiv
Yao, S., Zhao, J., Yu, D., et al. (2023)
ICLR 2023 arXiv
Exercise: Agent Classification
Analyze and classify AI systems according to agent taxonomy
Learning Objectives
- Analyze: Classify systems according to agent properties
- Evaluate: Assess strengths and limitations of architectures
- Apply: Apply formal definitions to real-world systems
Tasks
| Task | Points | Description |
|---|---|---|
| System Classification | 40 | Classify 6 systems (ChatGPT, Copilot, Roomba, etc.) |
| Architecture Analysis | 35 | Identify architecture types and memory/tool access |
| Design Proposal | 25 | Design a research assistant agent |
Exercise
Build a basic ReAct agent from scratch that can:
- Answer questions using a simulated search tool
- Perform calculations using a calculator tool
- Show the complete thought-action-observation trajectory
Discussion Questions
- What distinguishes an AI agent from a standard LLM application?
- How does the ReAct paradigm improve over pure Chain-of-Thought?
- What are the limitations of current agent architectures?
Additional Resources
Discussion & Questions
Join the Conversation
Have questions about this week's material? Want to discuss concepts with fellow students?