The Smithsonian Institution has a long-standing commitment to
sustainability, but sustaining momentum within a large organization
requires consistent effort and fresh perspectives. The Sustainable
Smithsonian team partners with Smithsonian staff to support that work
through research, communications, and strategic thinking. This semester,
one group developed infographics and a proposed newsletter spotlighting
recent global sustainability wins to encourage Smithsonian employees
during a challenging period for federal environmental initiatives, while
another conducted an in-depth investigation into the environmental cost
of artificial intelligence, analyzing whether AI-powered tools like
automated pollen classification offer a net environmental benefit over
traditional, labor-intensive research methods.
Project Overview
Stakeholders
Natascha Finnerty (contact at the Smithsonian Astrophysical
Observatory); Linette Dutari (contact on the PollenGEO project);
SAO Sustainability Group (employee sustainability group in
Cambridge, MA)
Disciplines / Majors
Sustainability research, data analysis, communications
Team Overview
The Sustainable Smithsonian Team (SST) is built on an ongoing
partnership between CUSD and the Smithsonian Institution (SI).
Past work has spanned plastic reduction, zero-waste event
planning, green building certifications, and carbon
sequestration. This semester, the team was divided into two
subteams: Climate Liability (focused on sustainability
communications for Smithsonian employees) and AI x Pollen GEO
(evaluating the environmental tradeoffs of AI use in ecological
research).
Problem Statement
The SAO Sustainability Group has expressed frustration with the
perceived lack of progress in sustainability initiatives at the
Smithsonian, compounded by government restrictions under the
current administration. Additionally, the growing use of AI in
scientific research, including the PollenGEO pollen
classification project, raises underexplored questions about
AI's environmental costs versus its efficiency benefits.
Approach
The Climate Liability subteam pivoted from its prior 'Fly Less
Fly Better' campaign focus to researching and communicating
positive sustainability developments. This included three
research areas: the history of SI sustainability initiatives,
current global sustainability successes, and actionable
sustainable living tips. Findings were synthesized into
infographics, posters, and a draft newsletter for SI employees.
The AI x Pollen GEO subteam researched the environmental
footprint of AI (energy use, water consumption, carbon
emissions, hardware production) and analyzed the tradeoffs
specific to PollenGEO's deep learning approach to automated
pollen classification.
Key Accomplishments This Semester
Climate Liability: produced multiple infographics and posters
covering SI sustainability history (timeline format), global
good news in sustainability, and practical sustainable living
tips. Drafted a newsletter for the SAO Sustainability Group
pending institutional approval. Developed interview questions
for an upcoming virtual meeting with the SAO Sustainability
Group. AI x Pollen GEO: completed a comprehensive literature
review on AI's environmental impacts, including energy, water,
and hardware footprints, and identified strategies for improving
AI sustainability through algorithmic efficiency, data center
design, and standardized reporting.
Next Steps
Climate Liability: launch newsletter if approved; hold virtual
meeting with SAO Sustainability Group; continue building the
relationship with the SI around positive sustainability
communication. AI x Pollen GEO: present research findings to the
PollenGEO team; refine research questions based on their
feedback; advance toward a comparative analysis of AI versus
manual research methods for pollen classification.
Risks & How They Were Addressed
The government shutdown limited communication with Smithsonian
contacts for the AI x Pollen GEO subteam, preventing direct
collaboration with the PollenGEO team. The subteam addressed
this by conducting independent research to build foundational
knowledge in preparation for a meeting with the PollenGEO
director in mid-December. The Climate Liability newsletter
requires institutional approval before distribution, so the team
prioritized infographics, which have received positive feedback
previously, as the primary deliverable.
