Generative AI: Some Implications for Higher Education
- Technology & Software
- Course Design
Last modified: May 18, 2026
This page outlines some of the major implications of AI for higher education
Generative AI: Considerations for Academic Use
The following statements were developed by CELT in collaboration with the Sustainability Committee to help students and faculty think critically about the broader implications of generative AI use in academic contexts.
We've provided some student-facing scenarios for each and hope you'll talk with your students and reflect on these impacts together.
While generative AI is often framed as a tool for learning and productivity, overreliance may reduce the active engagement with material necessary for learning and retention. Learning depends on the depth of cognitive encoding—that is, the extent to which information is actively processed, elaborated, and organized. Excessive AI use may therefore contribute to superficial understanding, reduced long-term retention, and weakening of foundational academic skills, such as critical thinking, complex problem-solving, writing, and independent inquiry. In addition, insufficient critical evaluation of AI outputs may lead students to internalize factual inaccuracies, fabricated content, or embedded algorithmic biases present in generated material.
What this looks like in practice for students: You ask an AI to summarize a dense reading before class. The summary feels clear, so skip the original. In discussion, you find you can't explain the argument in your own words or respond to follow-up questions; the understanding didn't stick because you never actually built it. Similarly, if you use AI to draft an essay and lightly edit it, you may turn in something that sounds polished but doesn't reflect your own reasoning and over time, the writing and analytical skills that come from wrestling with ideas yourself don't develop.
Developed by CELT in collaboration with the Sustainability Committee.
The operation of generative AI relies on an immense physical infrastructure that demands substantial resources, including energy, water, and rare-earth minerals for hardware production. This contributes directly to rising greenhouse gas emissions, intensive water consumption for data center cooling, electronic waste, and the environmental strain associated with resource extraction. As AI adoption accelerates, these ecological and socioeconomic costs raise critical sustainability concerns. Students and faculty are encouraged to consider these impacts when determining whether the use of AI is truly necessary for a given task.
What this looks like in practice : A single conversation with a large AI model can consume as much energy as charging a smartphone several times over and that cost multiplies across millions of daily users. Asking an AI to brainstorm ten variations of a sentence you could revise yourself, or regenerating a response because the first one was close enough, represents resource use that quietly adds up. Before turning to AI, it's worth asking: could I do this with a web search, a library database, or my own notes?
Developed by CELT in collaboration with the Sustainability Committee.
Generative AI systems are trained on vast datasets that frequently include copyrighted text, images, code, and other creative works—often without the knowledge or consent of the original creators. When AI generates content that closely resembles or reproduces this source material, it raises serious questions of intellectual property infringement and proper attribution. Students who submit AI-generated work may unknowingly present material that violates copyright law or misrepresents the origins of ideas. Furthermore, the use of AI can obscure the distinction between original thought and reproduced content, undermining the academic values of proper citation, attribution, and scholarly integrity. Students are encouraged to consider who created the material an AI draws upon, and whether their use of that output respects the rights and labor of those original creators.
What this looks like in practice: You ask an AI to help you write a short story and it produces a passage with a distinctive style or a plot element that closely echoes a published author whose work was in its training data without any attribution. Or you use an AI image generator for a class project and submit an image that was assembled from artists' work scraped without permission. Even if you didn't intend to plagiarize, the output may raise real questions about originality and attribution that reflect on your academic integrity.
When students enter prompts into generative AI tools, they may inadvertently share personally identifiable information, confidential course content, or sensitive data with third-party platforms operating under terms of service that permit broad data use, storage, or resale. This creates real risks to individual privacy and institutional data security. Students and faculty should exercise caution about what information they share with AI platforms, review the data policies of any AI tool before use, and avoid inputting any content that is sensitive, confidential, or personally identifying.
What this looks like in practice: You paste a classmate's name and details into an AI prompt while drafting a group project reflection, or you share excerpts from a confidential case study provided by your instructor. Even if the AI's response is helpful, that information has now been submitted to a third-party platform and may be stored, reviewed, or used to train future models. Free consumer AI tools in particular often have broad data retention policies. Try reading through the policy for a tool you use!
Generative AI systems reflect the biases embedded in their training data and design choices, which can result in outputs that perpetuate stereotypes, misrepresent marginalized communities, or present culturally narrow perspectives as neutral or universal. These biases are often invisible within the generated content itself, making critical evaluation especially important. Students and faculty are encouraged to scrutinize AI-generated material for embedded assumptions, seek out diverse perspectives beyond what AI produces, and remain aware that outputs may not be equitable, accurate, or representative across all populations and contexts.
What this looks like in practice: You ask an AI to describe a "typical nurse" or "successful entrepreneur" and the response defaults to gendered or racialized assumptions without flagging them. Or you use AI to help evaluate student writing samples and it rates certain dialects or cultural styles of communication as lower quality. Because AI outputs are generated fluently and confidently, these biases can be easy to miss and easy to pass along uncritically in your own work.