Group Assignment 4

Note: M4 - Final Assignment Deadline: Friday 7th Mar at 12:00 PM

GPT Models and AI Agents for Real-World Problems

Introduction

Welcome to the final assignment for the Agentic System developed by CrewAI. This project aims to deepen your understanding of modern AI techniques—particularly the integration of AI agents with GPT models—and their applications across diverse fields such as social science, business, education, healthcare, and more. In this assignment, you will leverage prompt engineering techniques to boost the performance of your model and develop an interactive Streamlit application to showcase your results.

Task Description

Develop an innovative project that employs GPT models and AI agents to address a real-world problem, analyze a dataset, or explore a social science question. The assignment guidelines are intentionally broad to foster creativity. Below are some general project ideas to inspire your work; you are welcome to propose your own.

Project Ideas

Description:
Build an AI agent that autonomously collects and analyzes public datasets (e.g., social media feeds, public health statistics, community survey results) to identify emerging trends.

Key Requirements:

  • Prompt Engineering: Design custom prompts for GPT models to extract insights, summarize key topics, and highlight patterns.
  • User Queries: Implement features that allow the agent to respond to user questions and provide clear summaries.
  • (Optional) Knowledge Graph: Visualize the connections between social trends, demographics, and other factors.

2. Educational or Training Chatbot Agent

Description:
Create a conversational AI that assists learners in understanding complex topics—be it in science, language studies, or career development.

Key Requirements:

  • Interactive Explanations: Use GPT models to deliver step-by-step explanations and real-life examples.
  • Prompt Engineering: Develop prompts that ensure consistent, accurate, and engaging responses from the chatbot.
  • Streamlit App: Incorporate interactive elements to make the learning experience dynamic and engaging.
  • (Optional) Knowledge Graph: Optionally, use a knowledge graph to demonstrate how different concepts interrelate.

3. Sustainability and Environmental Analysis Agent

Description:
Develop an AI-driven tool that analyzes environmental data, climate trends, or sustainability metrics to support decision-making.

Key Requirements:

  • Data Collection & Processing: Gather and preprocess open-source environmental datasets.
  • Prompt Engineering: Use structured prompts for the GPT model to produce summaries, risk assessments, or recommendations.
  • Streamlit Dashboard: Create visualizations of key metrics, scenario models, or action plans.
  • (Optional) Knowledge Graph: Optionally, illustrate the relationships between climate factors, emissions, policies, and societal impacts.

Data

  • You may utilize open-source datasets or generate your own data.
  • Ensure that your data aligns with your project’s objectives and effectively showcases the capabilities of your AI solution.

Emphasis on Prompt Engineering

Given the importance of prompt engineering in modern LLM development, focus on:

  1. Prompt Templates: Crafting precise prompts that guide the model effectively.
  2. Few-Shot Examples: Including minimal examples to shape output quality.
  3. Role/System Messages: Leveraging system messages (if using OpenAI or similar APIs) to define overarching behavior.

Knowledge Graph (Optional)

  • Optional Usage: Integrate knowledge graphs if they enhance your project’s demonstration of relationships and structured information.
  • Visualization: If implemented, use the knowledge graph to visualize interconnections within your chosen domain.

Streamlit Application

Develop a Streamlit app to:

  1. Demonstrate User Interaction: Showcase how users interact with your AI agent.
  2. Display Outputs: Present your agent’s analyses, visualizations, and insights.

Submission Guidelines

  1. GitHub Repository: Create a dedicated GitHub repository for this assignment. Include all code, datasets, and a descriptive README.md.
  2. Team Formation: Work individually or in groups of up to three members.
  3. DigitalExam Submission: Additionally, submit via DigitalExam, compiling all previous assignments into one file for your overall module exam portfolio. You are encouraged to refine or improve your past submissions for this purpose.

Good Luck!

We look forward to seeing how you creatively integrate GPT models, AI agents, prompt engineering techniques, and optionally knowledge graphs to tackle a real-world or social science problem. Enjoy your journey into advanced AI applications!