Reminders

Renew Our Schools - Register Today!

Space is still available for both the fall and spring Renew Our Schools competitions, a six-week challenge to save energy at your school!

Green Team Mini-Grant applications now open!

Students can apply for a $300-$700 mini-grant to do a project to save energy at their school or improve energy education. 

Energy Educator of the Year nominations are open!

KEEP is seeking nominations for the Wisconsin Energy Educator of the Year Award.

Watt's Your Appliance now available in Spanish!

KEEP's popular Watt's Your Appliance? activity is now available in Spanish as ¿Investigando y comparando las potencias de electrodomésticos comunes?.

Register for NAAEE 2025!

Register now for the online North American Association for Environmental Education (NAAEE) 2025 Conference (November 3–6) and Research Symposium (October 30).

What's the scoop on energy and artificial intelligence (AI)?

As students return for a new school year, AI – and discussions about how or whether it belongs in the classroom – is likely to continue at the top of people’s minds. So how can AI connect to energy education?

One of the most basic questions is: how much energy does AI consume? Up until recently, the general rule of thumb was that asking AI a question took 10 times as much power as a more traditional search while generating an image took as much power as charging a smartphone. And energy is not the only resource AI needs; data centers that power AI require significant amounts of water to cool them – estimated to be one cup of water per prompt.

As a result, some companies are investing effort into improving the efficiency of AI, and it is paying off. Just this month, Google stated that to answer a median text prompt, their AI, Gemini, uses only five drops of water and .24 watt-hours of electricity, the equivalent of using an 8-watt LED for less than two minutes. That is a sharp decrease in the power required for the same prompt just one year ago. However, Google has yet to release data about the electricity and water required for Gemini to generate images or video, which is estimated to significantly exceed the energy input of generating text.

How much energy did it take to produce this image? Was it enough to charge a smartphone? The image is the result of a prompt asking Adobe Firefly to create an engaging graphic for an article about artificial intelligence’s energy use. Adobe says, “Generative AI can use a lot of energy,” but does not provide any specifics.

Besides the question of the electricity used by individual prompts, there’s also the larger question of how the AI boom will affect the nation’s electric grid as a whole. Tech companies are racing to build massive new data centers to train and run AI models, all of which need to be powered by massive quantities of electricity. The Department of Energy reports, “data centers consumed about 4.4% of total U.S. electricity in 2023 and are expected to consume approximately 6.7 to 12% of total U.S. electricity by 2028.” To meet these needs, tech companies and utility companies are working on adding new sources of electricity to the grid. 

For the last few years, most new power has come from renewable energy sources such as solar, wind and batteries, which have become fast and inexpensive to install. However, with the challenges solar and wind projects face under new legislation and the desire for data centers to run around the clock, other options like restarting shuttered nuclear power plants and building methane gas power plants are being considered. This leads to another question: will the increased energy demand required by data centers and reduced investment in solar and wind increase individuals’ cost of electricity?  

However, there are a number of ways AI could support a smarter grid and therefore make our energy systems more sustainable. One possible use for AI would be predicting and managing solar and wind power on the grid. Our electric grid is delicately balanced, with the supply of electricity providing exactly the same amount of power as is demanded from the grid, at every single moment of every day. That becomes more challenging as more of the grid is served by renewables. AI could be trained to predict solar and wind production, and electric demand, and optimize getting users the cleanest, cheapest electricity at any given moment. 

Areas where AI could support a smarter grid

Donti, Priya L.;Kolter, J. Zico

Annual review of environment and resources, 17 Oct 2021, Vol. 46, Issue 1, pages 719 - 747 

Demand management is another area where AI may shine. In our current system, the electric grid has to be prepared to supply as much electricity as any user wants, whenever they want it. This can be a major challenge when everyone in a city wants a lot of power all at once – for example mid-afternoon on a hot summer day when everyone’s air conditioner is running simultaneously. That might require turning on more power plants, which can often be the most expensive – and most polluting – ones. Long term, it might require building entirely new power plants coupled with expensive upgrades of power lines to distribute that electricity. With AI, during moments of peak demand the electric utility can send signals to program participants (whether individual’s smart home appliances or even entire data centers) and tell them to reduce their electricity consumption for a few hours, which could eliminate the need for additional power plants.  

AI uses significant amounts of energy, and is likely to require massively more over the coming years. Yet what AI is used for will likely be as significant as the electricity used to power it. AI can be used to help manage electricity use and integrate renewable energy onto the grid. It can also be used to boost production of fossil fuels. Ultimately, the impact AI has on energy will be dependent on how people choose to use it.

In the meantime, this is a topic with a wide scope for classroom discussion to brainstorm, research, and evaluate how AI could and should impact a 21st century energy system.  

Upcoming Professional Development Opportunities

Data Nuggets: Bringing Authentic Soil Carbon Research to the Classroom

Authentic datasets have the unique ability to capture the attention of students and engage them in the work of scientists. Data Nuggets are free classroom resources, co-designed by scientists and teachers, designed to build data literacy through analysis and visualization of authentic data. Join The Wisconsin Energy Institute for a FREE webinar on October 23rd from 6 to 7pm to learn more! 

ATTENTION! Educators in WPS service territory

The below opportunities are intended for educators in the Wisconsin Public Service territory. Check this list to verify your eligibility to participate based on the utility service provider of your school building.

Solar on Schools fall '25 grant apps open 9/1 - 10/17

The FALL 2025 Solar on Schools grant application will be open September 1 – October 17, 2025.

The Solar on Schools grant provides Wisconsin public K-12 schools and colleges access to an in-kind module grant, valued up to $20,000. Applicants are eligible to apply for one grant per district installing solar. Grant applicants must demonstrate project development and the ability to install their PV system within 12 months of being awarded the grant.

Did someone forward you this email?

The KEEP program is funded with generous support by Alliant Energy, Madison Gas and Electric, We Energies, Wisconsin Public Service, WPPI Energy, and Xcel Energy. 

Published September 2025