Can AI predict how energy will be used or made before it happens?

The Project

This project continues the Smart Meter journey by showing students how AI can help forecast energy use or generation. Grades 4–6 stay focused on consumption, using the same energy-use questions from The Smart Meter to predict savings and lower demand. Grades 7–8 shift into generation, using solar as the forecast source and modeling how sunlight changes output over time. All students investigate the energy cost of AI, train a predictor on clean data, test it with a Digital Twin, and share a final launch with a real audience. The project connects directly to Utility Integrated Resource Planning (IRP) and Public Utility Commission (PUC) hearings, where data is used to make decisions about future energy infrastructure.

1. 1
   Phase I: From Metering to PredictionPhase 1 of 3

   1. Innovation Launch: From Metering to Predicting

      Students reconnect their Smart Meter work to a new mission: grades 4–6 continue forecasting energy use and savings, while grades 7–8 shift to solar generation and predict how output changes over time. They discover that a digital twin is a virtual copy of a real energy system, and they produce a team launch plan that sets roles, focus, and the public audience for the project.

   2. The Energy Cost of AI

      Students measure the power cost of running AI and compare it with the work their sensing system already does. They discover that prediction is never free: every model has an energy footprint, and they produce an integration checklist that prepares their hardware, code, and data for the training phase.

   3. Training the Oracle

      Students organize clean training examples from their energy data and write the logic that helps a pocket computer (microcontroller) recognize patterns. They discover that a model only learns from the examples it sees, and they produce a first-pass predictor ready to test against new data.
2. 2
   Phase II: Building and Testing the ForecastPhase 2 of 3

   1. Forecasting the Future with Data

      Students run their predictor on new scenarios and compare the results against what they expected. They discover the difference between training and inference, and they produce a forecast report that shows where the model is reliable and where it needs revision.

   2. Digital Twin Systems

      Students map their small-scale prediction to a larger building, campus, or grid system. They discover how a digital twin helps people test "what if" decisions before making real-world changes, and they produce a systems map that connects local data to community-scale energy planning.

   3. The Bias Check

      Students stress-test their model with edge cases, missing data, and changing conditions. They discover that bias and noise can make a forecast look stronger than it really is, and they produce a validation summary that explains what the model can and cannot claim.
3. 3
   Phase III: Public Launch and Career PathwaysPhase 3 of 3

   1. The Creative Brief

      Students turn their technical findings into a public-facing case for action. They discover that good AI work matters only when people can understand and use it, and they produce a concise policy brief or presentation outline that names the key insight, audience, and call to action.

   2. The Visionary Launch

      Students deliver their final video, podcast segment, or summit-style presentation for a real or simulated community audience. They discover how predictive AI can support better energy decisions at scale, and they produce a polished launch artifact that communicates both the forecast and its consequences.

   3. Career Roadmap and Reflection

      Students reflect on the full journey from measuring energy to forecasting the future and map the careers touched by the project. They discover how the same skills transfer to utility planning, data science, policy, and engineering, and they produce a personal or team roadmap that names next steps and future roles.