Lecture Notes: AI for Climate Action – Built Environment Optimization and Climate Adaptation

One. Course Details

This is a fast-paced webinar hosted by the Stanford Doerr School of Sustainability as part of its executive and professional education portfolio. The session features two lightning talks by leading experts: Dr. Rishee Jain, Associate Professor of Civil and Environmental Engineering and Director of Stanford’s Urban Informatics Lab, and David Farnham, VP of AI and Engineering at ClimateAi.

Designed for executives, sustainability leaders, policymakers, and technologists, the presentation explores practical applications of AI in addressing climate challenges across two critical domains: the urban built environment and agricultural climate adaptation. It combines cutting-edge research, real-world case studies, and actionable insights to demonstrate how AI can be harnessed to create more sustainable, resilient, and human-centered systems. The session concludes with a preview of Stanford’s upcoming executive programs on AI, climate, and sustainable leadership.

Two. Key Learning Takeaways

• The most powerful AI solutions for climate action combine machine learning with first-principles physics, creating hybrid models that are both accurate and generalizable across different contexts.

• Sustainable design should be reframed from "reducing negative impacts" to maximizing positive impacts on human well-being, recognizing that people are the most valuable resource in any system.

• Hybrid work has fundamentally changed office dynamics, creating new opportunities to optimize both energy efficiency and collaboration through AI-powered space design.

• AI is transforming climate adaptation by translating weather and climate forecasts into actionable business insights, helping organizations manage supply chain risks, optimize agricultural production, and build resilience to extreme weather events.

• Large language models (LLMs) and agentic workflows are unlocking new value in climate tech by improving interpretability, flexibility, communication, and actionability of complex climate data.

• In-situ data scarcity remains a major challenge in both urban and climate modeling, but hybrid AI-physics approaches can help fill gaps and reduce reliance on expensive sensor networks.

• AI will augment rather than replace human decision-making in climate and built environment applications for the foreseeable future, due to issues of hallucination, liability, and the need for human judgment in high-stakes decisions.

Three. Course Gold Quotes

1. "For every concrete block we put together or every kilowatt hour we spend, are we making a positive impact on human life? That’s the framework we should be working under." – Rishee Jain

2. "When we think about energy, we call it a dollar. When we think about space, we say $10. When we think about people, it’s $100. That’s where we need to focus our efforts." – Silicon Valley tech executive (cited by Rishee Jain)

3. "The science is still the science. I don’t want LLMs to be doing science for us. But they can help us scale the application of great science much more seamlessly." – David Farnham

4. "AI is not going to make decisions for us anytime soon. But it can help us explore more scenarios, expand our solution space, and make better-informed decisions faster." – Rishee Jain

5. "Climate adaptation isn’t just about surviving extreme weather – it’s about thriving in a changing climate by using data and AI to anticipate risks and seize opportunities." – David Farnham

6. "The biggest opportunity in AI for climate is not just optimizing existing systems, but reimagining how we design cities, buildings, and food systems from the ground up." – Rishee Jain

7. "Change can be scary, especially when you’re talking about altering decision-making processes that have worked for decades. But the cost of inaction on climate is far greater than the risk of adopting new technologies." – David Farnham

Four. Layered Learning Notes

Module 1: Human-Centered AI for the Built Environment

• Paradigm Shift in Sustainability:

  • Traditional sustainability focuses on minimizing emissions and costs

  • New framework: Maximize positive impact on human well-being for every resource expended

  • Recognizes the interconnectedness of energy use, space design, and human productivity

• Intra-Building Dynamics and Hybrid Work:

  • Hybrid work has transformed office usage patterns, with uneven occupancy across space and time

  • Current building systems often condition and light entire spaces even when only a fraction is occupied

  • Privacy-aware AI inference can map occupancy patterns without compromising individual privacy

  • Initial optimization efforts focused on energy savings (5% reduction demonstrated), but the real value lies in improving collaboration and productivity

• AI-Powered Facilities Management:

  • Generative AI models fine-tuned on building physics can augment building operators’ decision-making

  • These models can process and prioritize occupant requests (temperature adjustments, maintenance issues)

  • Particularly valuable for operators managing multiple facilities simultaneously

  • Designed to augment rather than replace human operators, leveraging AI for scale while retaining human judgment

Module 2: Urban-Scale Energy Modeling with Hybrid AI-Physics Approaches

• Limitations of Current Practice:

  • Most building energy models treat individual buildings as isolated structures in empty fields

  • Ignores critical urban context effects: mutual shading, heat transfer between buildings, and urban heat island effects

  • Leads to inaccurate energy use predictions and suboptimal retrofit strategies

• Hybrid Modeling Framework:

  • Combines fundamental building physics simulations with machine learning algorithms

  • Physics provides the foundational structure and generalizability

  • AI adds speed, scalability, and the ability to learn from real-world data

• Key Findings from Sacramento Case Study:

  • Taking urban context into account allows achieving 80% energy savings targets by retrofitting half as many buildings

  • Window and lighting retrofits show particularly strong synergies in urban environments

  • Unlocks opportunities for large-scale system integration with renewable energy and electric vehicles

• Stakeholder-Specific Applications:

  • Designers/Owners: Evaluate energy impacts of design changes on both their building and adjacent structures

  • Policymakers: Prioritize buildings for retrofit incentives to maximize emissions reductions

  • Planners/Developers: Assess energy impacts of large-scale projects and urban infrastructure investments

Module 3: AI for Climate Adaptation and Agricultural Resilience

• ClimateAi’s Core Mission:

  • Focuses on climate adaptation rather than mitigation

  • Helps organizations build resilience to weather and climate variability and change

  • Serves primarily food and beverage and agricultural customers, who are among the most vulnerable to climate shocks

• Product Portfolio:

  • Monitor: Short-term weather and seasonal forecasts (up to 6 months)

    • Sourcing: Track global crop conditions and anticipate supply shocks

    • Production: Optimize farm operations and mitigate impacts of droughts, heat waves, and floods

    • Sales & Marketing: Align supply chain and marketing strategies with climate-driven demand patterns

  • Adapt: Long-term climate projections (decades into the future)

    • R&D: Develop climate-resilient crop varieties

    • Finance & Risk: Assess water risk and other long-term climate vulnerabilities

    • Sustainability: Support TCFD and CSRD reporting requirements

• Impact Models:

  • Translate raw weather and climate data into specific, actionable metrics for businesses

  • Go beyond generic forecasts to answer questions like: "How will this heat wave affect my wheat yield in Saskatchewan?"

  • Enable proactive rather than reactive decision-making

Module 4: Opportunities and Challenges of AI in Climate Tech

• Transformative Opportunities from LLMs and Agentic Workflows:

  • Interpretability: Automate the translation of climate data into business impacts, eliminating manual spreadsheet calculations

  • Flexibility: Allow customers to create custom dashboards and analyses tailored to their unique workflows

  • Communication: Provide consistent, accurate explanations of complex climate concepts in natural language

  • Actionability: Standardize and automate recommended actions, ensuring no mitigation options are overlooked

• Additional AI Advancements:

  • Accelerated software development: AI copilots have dramatically reduced the time to roll out new features and platforms

  • Improved uncertainty quantification: Fast inference of AI weather models enables more robust probabilistic forecasts

  • Democratization of data access: Natural language interfaces allow non-experts to query and analyze complex climate datasets

• Key Challenges:

  • Data Security: Customers are increasingly concerned about protecting proprietary information when using LLMs

  • Adoption Barriers: Resistance to change and fear of AI-driven decision-making, particularly in traditional industries like agriculture

  • Energy Consumption: The growing power demand of AI systems is itself a climate challenge

  • Data Gaps: Scarcity of high-quality in-situ data, particularly in developing regions, limits model accuracy

Module 5: Future Outlook and Decision-Maker Guidance

• The Future of AI in Climate Action:

  • Continued convergence of AI and physics-based modeling will create more powerful, accurate, and accessible tools

  • Low-cost sensors combined with AI will democratize access to building and environmental data

  • Agentic AI systems will become increasingly sophisticated at automating routine tasks and providing decision support

• Advice for Senior Decision-Makers:

  • Take AI seriously: Invest in understanding both the capabilities and limitations of the technology

  • Avoid extreme positions: Reject both "AI will solve everything" and "AI is useless" mindsets

  • Use AI to expand your options: Leverage AI to explore more scenarios and make better-informed decisions faster

  • Keep humans in the loop: AI should augment, not replace, human judgment, especially for high-stakes decisions

  • Invest in data infrastructure: High-quality, well-organized internal data is essential for unlocking the full value of AI

Wishing you every success as you apply these powerful AI-driven insights to create more sustainable, resilient, and human-centered systems in your work. May you harness the transformative potential of artificial intelligence to not only address the urgent challenges of climate change but also to build a better, more prosperous future for all. Remember that the most impactful solutions come from combining cutting-edge technology with deep human understanding and a commitment to the common good.