On Sept. 15, all Smith students received access to the institution’s license for Google Gemini, a generative artificial intelligence platform.
Alex Keller, Dean of the College, emailed students about this implementation a week and a half earlier. The announcement stated the subscription would “level the playing field,” explaining that the institutional license to Gemini gives students who cannot afford private subscriptions the same resources as students who can.
“Without institutional access, students who can afford private subscriptions gain access to protected tools, while others are left using free consumer models that harvest their data for AI training,” Dean Keller stated.
The statement emphasized the importance of the data and privacy protections that this model provides versus other public consumer models.
The Google Gemini Subscription comes from the Smith Google Workspace for Education license, which supplies core applications such as Email, Calendar and Google Drive. In June of 2025, Gemini became another core application available at no extra cost. Smith granted this tool to faculty in the spring of 2026 and now to students this fall.
According to Smith’s Chief Information Officer, Heather Woods, both Mount Holyoke College and Amherst College have also made the Google Gemini license available to students and faculty.
“We didn’t have to go out and think about buying something new. The question was, we have the tool available. Should we let people use it?” stated Woods.
Woods also said there is a “good sense” that the Consortium of Liberal Arts Colleges (CLAC), a cohort of 75 American liberal arts colleges, has given their students access for much longer than Smith.
At its core, Gemini is an AI chatbot that can interpret and create text, video, audio, images and code. Smith’s Gemini is protected, meaning that any data input into the chat will not be used to improve or train the bot or spread to outside sources, unlike other public AI.
Essentially, Smith wants to ensure that data is as safe as possible.
Smith’s AI use as an institution is not new. Last year, Smith partnered with University of Massachusetts, Amherst, to access their platform UMass Campus GenAI, in which professors can request a range of models on behalf of themselves or students, including ChatGPT, Claude and Gemini. However, that changed this year.
“[Smith] made a decision to pull back from [the UMass] partnership because the administrative cost for it was higher than we wanted to pay,” said Woods. “So we have made the decision instead to set up something that looks very similar here at Smith.”
This model is called LibreChat, or Smith GenAI.
In the spring of 2026, the Student Government Association (SGA) surveyed Smith students and faculty about their AI use, concerns and thoughts about it in the classroom. They found a wide variety of opinions.
“The faculty interviewed for this report are not asking the institution to refuse AI. They are insisting instead that Smith must deliberate carefully, with faculty and students as genuine participants, about what adoption means, what it costs and what it serves,” the report wrote.
In early spring, the SGA committee surveyed 71 students and interviewed an unspecified number of professors from across all divisions. The committee acknowledged that the student data set is not large enough to generalize the entire community, but can give insight into individual perspectives.
Of the students surveyed, 43.3% said that they have never used AI, while 9.0% reported using it daily. The most common motivation for its use was to save time and to prepare for class.
“Survey responses indicate that students generally favor a moderated approach to AI in academic settings. The most common response was that AI use should be permitted with guidelines,” stated the report, as it also found concerns about the ethics, environmental impacts and academic integrity.
Kate Queeny, a Smith chemistry professor of 27 years, asserted her curiosity as to what students think about institutional AI.
“So much of what you do when you’re helping a person understand something is reading the person. I’m always poking at my students when they say, ‘I’m really confused.’ I say, ‘you’re not confused– you have a question,’ said Queeney. “There’s a lot that you learn in the process of answering a question that seems like one AI could answer.”
Many students are surprised about the Gemini license. Popular Instagram Smith satire account “The Orb” posted about the announcement with the text: “Forget what we just said: Smith Announced Gemini Licensure Days After Affirming Commitment to Sustainability.” It is overlaid on a picture of a Smith “Carbon Neutral” Geothermal Energy Project Poster.
This post reflects a common criticism of AI chatbots like Google Gemini: their environmental impact. In August 2025, the Chief Scientist and Chief Technologist of Google published an article stating that the average text prompt into Gemini uses “0.24 Wh of energy.”
For reference, an average computer charger uses anywhere between 40 and 50 Wh. Google also added in their report that their average single text prompt uses “5 drops of water” and emits 0.03 grams of carbon dioxide equivalent. Google states that they are “heavily” investing in methods to reduce power consumption and water use needed per prompt.
Despite the addition of the licensure, the AI policy still remains individual to each department, class and professor.
Smith announced that they are hiring student interns to teach their peers how to use AI for the 2026 school year and new students are required to complete the Academic Integrity Module, which discusses Smith’s AI policies.
On their website, Smith states that their five main AI academic expectations are “1. Know your course policy; 2. Practice full transparency; 3. Be accurate and responsible; 4. Guard personal data; and 5. Embrace ‘desirable difficulty.’”
Alumna Amanda Ballantyne ‘02, who assisted the National Artificial Intelligence Advisory Committee and the White House on AI policies, is coming to speak on Oct. 20. Additionally, there is a teaching and learning seminar running from September to December, Scaffolding Deep Learning in the Age of AI, exploring the “pedagogical problems” of AI.
“Members of the Smith community are particularly skeptical and apprehensive about the emergence of AI in academia,” the SGA report stated, “Rather than computational skills and data safety, they’ve expressed a desire to ask moral, epistemological and social questions.”
Be First to Comment