AI summary by TwinMind
Charlotte Degot: A more accurate way to calculate emissions | TED
TED
9m 26s
Current State of Carbon Emissions Measurement
- Emissions continue increasing despite decades of reduction commitments
- Key problem: inaccurate measurement of climate impact
- Analogy: like saving money without price tags or losing weight without measuring portions/calories
- Greenhouse gas emissions inherently difficult to measure
- No color, invisible
- Cannot place sensors everywhere (buildings, fields)
- Reliance on estimations and conversion factors
- Margin of error: 30-60% in most measurements
- Targets and action plans based on inaccurate data
Corporate Emissions Reporting Statistics
- CDP (nonprofit global disclosure system) data analysis
- Over 2/3 of companies not accurately measuring emissions
- Only 7% of companies ultimately reducing impact
- Core principle: "cannot reduce what you cannot measure"
Need for Comprehensive Carbon Accounting
- Corporations must measure across all activities and sources
- Apply same rigor as financial accounting to carbon measurements
- Financial accounting took 100+ years to develop modern automated systems
- Climate urgency: don't have 100 years for carbon accounting development
- Essential for meaningful targets and successful action plans
AI as Solution for Carbon Measurement
- AI can process data automatically from diverse, unstructured sources
- Invoices, consumer behavior data
- Capabilities: modeling to estimate missing information, simulation, optimization
- Accelerates carbon accounting development timeline
Case Study: Wine and Spirits Company
- Scale: billions in sales, hundreds of brands, global consumers
- Emission sources requiring measurement:
- Direct emissions from facilities
- Purchased electricity
- Raw materials, leased assets, IT emissions
- Business travel, transportation, waste
- Product end of life
- Challenge: most data inaccessible (outside direct scope, suppliers can't calculate emissions)
Glass Bottles Example - Traditional Method
- Current calculation: 1,000 tons glass × 950 kg CO2/ton = 950,000 kg CO2
- Highly inaccurate: ignores numerous emission-impacting factors
- Difficult to set targets and action plans
AI-Enhanced Method
- Data scientists process detailed factors:
- Glass type, color, recycling share
- Supplier country of origin, transportation mode
- Brand-specific data
- Simulation of design and supply chain integration
- Key impact factors identified:
- Glass color: 1.5x more emissions (clear vs green bottles)
- Country of origin: 2x emission difference based on energy mix
- Design: 1.5x more emissions for same weight (design variations)
- Results: granular level correlation and calculation
- Emissions correction: 30-50% adjustment from traditional method
- Enables actionable outcomes:
- Meaningful target setting
- Concrete initiative identification
- Performance tracking capability
Case Study: Cement Industry
- Scale: if cement were country, would rank 3rd largest emitter
- After China and US
- Before European Union and India
- Primary emission source: clinker production (key cement ingredient)
- Process requirements: maintain 1,400°C temperature
- High fuel consumption
- Carbon-containing materials
- Solution focus: cleaner, higher quality clinker production
- Higher quality = less clinker needed = fewer emissions
AI Implementation in Cement Production
- Complex science with multiple interdependent factors:
- Process parameters (machine rotation speed, fill rate)
- Fuel type, raw materials, chemical composition
- Current approach: manual parameter maintenance by operational teams
- AI enhancement:
- Better measurement through diverse sources
- Direct measurements, material/mass balance
- Simulation of potential decisions
- Optimal recommendation delivery to operators
- Results: substantial emissions reduction within months
Universal Applications and Conclusion
- Infinite applications possible across industries
- No company/industry cannot derive significant climate impact from AI
- AI not sole solution but critical enabler
- AI capabilities enable emissions reduction through:
- Accurate measurement
- Simulation
- Optimization
- Benefits: fast, cheap, easy implementation
- Opportunity cannot be missed
