Skip to content

DEMO_SCRIPT.md

Latest commit

 

History

History
252 lines (182 loc) · 8.47 KB

DEMO_SCRIPT.md

File metadata and controls

252 lines (182 loc) · 8.47 KB

🎬 IPCC RAG System - Demo Script

A presentation script for showcasing the IPCC RAG system

🎯 Demo Overview

Duration: 15-20 minutes
Audience: Researchers, policymakers, students, technical stakeholders
Goal: Demonstrate the system's capabilities and value

🚀 Demo Setup

Before the Demo

  1. Prepare your environment:

    cd llmrag
streamlit run streamlit_app.py

  2. Load a chapter (do this before the demo starts):

    • Select wg1/chapter02 (Changing State of the Climate System)
    • Click "Load Chapter"
    • Wait for loading to complete

  3. Test the system with a few questions to ensure everything works

  4. Prepare your browser with the Streamlit interface open

📋 Demo Script

Introduction (2 minutes)

"Good [morning/afternoon], everyone. Today I'm excited to show you a powerful new tool for climate research - the IPCC RAG System."

Key Points:

  • This is a local RAG system for IPCC climate reports
  • It runs entirely on your computer (no internet needed after setup)
  • It provides accurate, sourced answers to climate science questions
  • It's designed for researchers, policymakers, and students

"Let me show you how it works..."

System Overview (1 minute)

"The system works in three simple steps:"

  1. Load IPCC chapters - We have multiple chapters from IPCC reports
  2. Ask questions - Natural language questions about climate science
  3. Get answers - Accurate responses with exact source citations

"The magic is that it combines the power of AI with the authority of IPCC reports."

Live Demonstration (10 minutes)

Step 1: Show the Interface

"Here's the web interface. It's clean and simple - just select a chapter and start asking questions."

  • Point out the chapter selector
  • Show that wg1/chapter02 is already loaded
  • Explain this is "Changing State of the Climate System"

Step 2: Basic Question

"Let's start with a fundamental question about climate change."

Question: "What are the main findings about temperature trends?"

What to highlight:

  • The system processes the question
  • It searches through the IPCC content
  • It generates a coherent, scientific answer
  • It shows paragraph IDs for source tracking

"Notice how the answer is scientifically accurate and includes specific paragraph IDs so you can verify the information."

Step 3: Specific Data Question

"Now let's ask for specific data and measurements."

Question: "How much has the Earth warmed since pre-industrial times?"

What to highlight:

  • Specific numerical values
  • Units of measurement
  • Confidence levels
  • Time periods mentioned

"This shows the system can provide precise, quantitative information from the IPCC reports."

Step 4: Complex Relationship Question

"Let's try a more complex question about relationships and mechanisms."

Question: "What causes global warming?"

What to highlight:

  • Clear explanation of mechanisms
  • Scientific terminology
  • Multiple factors considered
  • Proper source citations

"The system can explain complex scientific concepts in a clear, accessible way."

Step 5: Uncertainty Question

"One of the strengths of IPCC reports is their careful treatment of uncertainty. Let's test that."

Question: "How certain are scientists about climate change?"

What to highlight:

  • Discussion of confidence levels
  • Acknowledgment of uncertainties
  • Nuanced language about certainty
  • Scientific rigor

"This demonstrates the system's ability to handle nuanced scientific concepts."

Technical Highlights (2 minutes)

"Let me quickly show you some technical aspects that make this system powerful:"

Source Tracking

"Every answer includes paragraph IDs from the original IPCC reports. This means you can:

  • Verify information by checking the original source
  • Cite properly in your research
  • Understand context by reading the surrounding paragraphs
  • Build trust in the system's accuracy"

Local Operation

"The system runs entirely on your computer:

  • No internet required after initial setup
  • Your data stays private
  • No usage limits or costs
  • Works offline for research in remote locations"

Multiple Interfaces

"We provide multiple ways to use the system:

  • Web interface - for interactive exploration
  • Command line - for scripting and automation
  • Python API - for integration with other tools"

Use Cases and Benefits (2 minutes)

"This system is valuable for several key audiences:"

For Researchers

  • Quick literature reviews - Get answers in seconds instead of hours
  • Fact-checking - Verify specific claims against IPCC reports
  • Teaching - Use in climate science courses
  • Policy support - Access authoritative climate information

For Policymakers

  • Evidence-based decisions - Direct access to IPCC findings
  • Rapid briefings - Quick answers to climate questions
  • Stakeholder communication - Clear, accurate information
  • Report preparation - Proper citations and sources

For Students

  • Learning tool - Interactive way to explore climate science
  • Research skills - Practice asking good questions
  • Source evaluation - Learn to check and cite sources
  • Scientific literacy - Understand climate science concepts

Future Development (1 minute)

"We're actively developing this system with several exciting features planned:"

  • More IPCC chapters - Adding WG2 and WG3 reports
  • Better models - Upgrading to larger language models
  • Multi-language support - Non-English IPCC reports
  • Collaborative features - Share questions and answers
  • Mobile app - iOS and Android versions

Q&A and Discussion (3 minutes)

"I'd love to hear your thoughts and questions. What aspects would be most valuable for your work?"

Potential discussion points:

  • What types of questions would you ask?
  • How could this fit into your research workflow?
  • What additional features would be most useful?
  • How could we improve the system?

🎯 Demo Tips

Preparation

  • Practice the demo several times before presenting
  • Have backup questions ready in case the system is slow
  • Test on the presentation computer if possible
  • Have screenshots ready as backup

During the Demo

  • Speak clearly and at a measured pace
  • Highlight key features as you demonstrate them
  • Engage the audience with questions
  • Be honest about limitations if they come up
  • Show enthusiasm for the technology

Handling Questions

  • Technical questions: Refer to the developer guide
  • Accuracy concerns: Emphasize source tracking and IPCC authority
  • Cost questions: Highlight that it's free and open source
  • Deployment questions: Mention local installation and cloud options

📊 Demo Metrics

Success Indicators

  • Audience engagement - Questions and discussion
  • Understanding - Clear comprehension of capabilities
  • Interest - Requests for more information
  • Feedback - Suggestions for improvements
  • Follow-up - Requests for access or collaboration

Common Questions

  • "How accurate are the answers?" - Based on IPCC reports, with source tracking
  • "Can I use this for my research?" - Yes, it's designed for researchers
  • "How much does it cost?" - Free and open source
  • "Can I add my own documents?" - Currently IPCC-focused, but extensible
  • "What about other languages?" - Planned for future versions

🎉 Closing

"Thank you for your attention. This system represents a new way to access and understand IPCC climate science. I believe it can be a valuable tool for climate research, education, and policy-making."

"If you're interested in trying it out, I can help you get set up. The system is open source, so you can also contribute to its development. Let's work together to make climate science more accessible and actionable."

"Any final questions or thoughts?"

📝 Demo Checklist

Before Demo

  • System is running and tested
  • Chapter is loaded
  • Browser is ready
  • Backup materials prepared
  • Room setup checked

During Demo

  • Introduction completed
  • All questions demonstrated
  • Technical highlights shown
  • Use cases explained
  • Q&A handled
  • Closing delivered

After Demo

  • Follow-up materials provided
  • Contact information shared
  • Feedback collected
  • Next steps discussed

Good luck with your demo! 🌍📚