thoughtful institutional policy. It underscores the need for researchers with expertise and critical thinking skills, who see AI as a useful tool but not a source of definitive truth. It means communicating clear guidelines to all those involved in visitor-facing content creation, outlining what aligns with the organization’ s mission and voice, and what needs to be carefully reviewed. The goal is not to block or hinder AI adoption— that would result in missed benefits and opportunities— but to approach its use critically, with eyes wide open.
5
The Bigger Picture
Microsoft announced plans last fall to re-open Pennsylvania’ s Three Mile Island nuclear power plant in 2027, almost 50 years after the notorious partial meltdown that caused it to be taken off-line. Meta signed an agreement in early 2025 to keep a nuclear reactor in Illinois running for another twenty years. Amazon and Google have struck similar deals. For all the talk of“ virtual” technology, AI uses up almost unimaginable amounts of very real energy and natural resources.
From the critical minerals used in computers, devices and their batteries, to the energy required to run data centers and train large language models, AI has a voracious appetite and an equally large impact on the environment. Running a single LLM training model produces almost 660,000 pounds of carbon dioxide emissions, the equivalent of five gas-powered cars run over their total lifetime( Crawford, 2021). One ChatGPT query uses as much electricity as a lightbulb needs to produce light for 20 minutes( Mazhar, 2024). Creating a single image with a generative AI app consumes the same amount of power it takes to fully charge your cell phone( Heikkilä, 2023). Where does this fit with your museum’ s commitment to sustainability? While it’ s unlikely and unrealistic that AI can be avoided completely, it may provide an opportunity to raise both staff and public awareness through thoughtful internal policies and visitor engagement opportunities.
Next steps
AI is already having a profound impact on how we, as museum professionals, do our jobs and engage with the public. The key to a successful implementation and integration is to become informed, ask the right questions and work collaboratively with staff to establish ground rules about its use. A good starting point might be to have a guided conversation with one of the existing templates, such as MuseumWeek’ s Internal AI Policy Guide, or exploring how other institutions and working groups have tackled AI’ s challenges and opportunities. We owe it to our mission and our visitors to carefully weigh AI’ s promise and innovation against potential perils. References available on page 60
Sabrina Greupner is a consultant for museum digital, an instructor at the University of Toronto in the Master of Museum Studies program, and a former Director of Interpretation and Design at the Ontario Science Centre in Toronto, Canada. She can be reached at sabrina. greupner @ gmail. com
42 MATAHEF ISSUE 01