As a newly emergent technology, GenAI poses many unfolding ethical challenges. Each of us should take some time to become more informed about the implications of GenAI and how it challenges or aligns with our personal, professional, and disciplinary values. This page provides an overview of some key ethical questions and how they intersect with established York University values, particularly as they relate to teaching and learning.
Ethics for Faculty, Staff, and Students
Some ethical questions for all of us to consider before engaging with any given GenAI tool include:
- How does this technology obtain its training data? What labour practices and standards are used for human workers in the training process?
- How and where does this technology reinforce implicit or explicit biases?
- How accurate and verifiable are the outputs of this technology?
- How transparent is this technology regarding information security and data protection? What happens to content that is shared with it?
- What are the environmental costs associated with this technology? Are there more sustainable options available?
General Limitations and Risks of Generative AI:
This resource, from eCampus Ontario provides an overview of some of GenAI’s risks and limitations.
GenAI and York Values
We recognize that GenAI may at times be in tension with our existing values, while at other times it may serve to further them. York University has longstanding commitments to academic excellence, Decolonizing, Equity, Diversity and Inclusion (DEDI), sustainable development goals (SDGs), experiential education, and educational access and accessibility. GenAI challenges each of these in some ways, while also offering opportunities.
Educational Excellence
- Information accessed or produced via GenAI may contain misinformation (e.g. hallucinations) or disinformation (e.g. propaganda). Learning how to be a discerning consumer of digital content is more crucial than ever.
- GenAI can support personalized learning experiences that boost student engagement and learning.
- Over-reliance on AI may reduce critical thinking skills and learner independence. Determine when and how to best use GenAI as a support rather than a replacement for thinking.
Sustainable Development Goals
- The high energy and water consumption required to train and maintain AI models can have a significant environmental cost. This includes the carbon footprint from energy use and the substantial water needed for cooling data centers, which can offset some of the sustainability benefits AI aims to achieve.
- GenAI can support research and development of sustainability efforts, from large-scale projects like climate modeling to small-scale initiatives such as optimizing local resource use.
Access and Accessibility
- GenAI technologies can exacerbate existing digital divides
- GenAI can assist in identifying and developing opportunities for enhanced educational access through frameworks such as Universal Design for Learning (UDL)
- GenAI can support learning for diverse learners by acting as assistive technology, through improved voice-to-text (or vice versa), image descriptions, summaries of complex information in new ways, and more
Decolonization, Equity, Diversity, and Inclusion
- GenAI can be explicit in extractive approaches to data, including data colonialism. It can also reproduce implicit or explicit bias and limit diverse representation.
- Various Indigenous and other groups are working towards data sovereignty and AI technology informed by Indigenous ways of knowing and being.
- GenAI can support students by illuminating the hidden curriculum of higher education, unpacking and explaining unfamiliar terminology, and providing examples of discipline-specific work.
Experiential Education
- GenAI can support simulation and case studies for hands-on learning.
- Simulated experiences created by GenAI may not fully capture the complexity of real-world situations.
Intellectual Property, Copyright, and Data Security
Along with the ethical considerations mentioned above, GenAI brings significant concerns regarding intellectual property (IP), copyright, and data security. AI-generated content raises complex questions about ownership and compliance with existing copyright laws. Regulators are in the process of addressing these issues from a legal and policy perspective. The vast amounts of data used by AI systems also poses serious privacy risks. All GenAI users should protect sensitive information to prevent data breaches and unauthorized access. Before you share anything with GenAI, think carefully about where that information is going and how it may be used.
You can read more about the data protection afforded to York faculty, staff and students through Microsoft Copilot here.
Discussing GenAI Ethics
For Faculty:
When discussing GenAI ethics with students or TAs, we recommend the following:
- Emphasize the importance of critical thinking and ethical considerations.
- Encourage open dialogue about the implications of AI on privacy, bias, and intellectual property.
- Use real-world examples to illustrate potential ethical dilemmas and foster a collaborative environment where questions and diverse perspectives are welcomed.
For Students:
When approaching instructors about GenAI ethics, we recommend the following:
- Be prepared to learn alongside your professors and TAs. Developments in this field are so new that many researchers and educators are still learning what this technology means for their disciplines and for their classrooms!
- Come prepared with specific questions or concerns. How do your GenAI concerns relate to the course or to the disciplinary context? Would you like to have a one-on-one discussion with the professor? A discussion with the whole class? A guest lecture from an industry expert? Not everything will be possible but be ready to highlight your particular needs.
Want to Learn More?
Some Ethical Considerations for Teaching and Generative AI in Higher Education.
A discussion of ethical considerations pertaining primarily to large language-models such as ChatGPT. Lydia Wilkes touches on the environment, labour, data, privacy, bias and access.
HESA AI Roundtable: Decolonization. Video (1hr2m).
This video, from the Higher Education Strategy Associates AI Roundtable, addresses the question: How can AI either perpetuate or mitigate racism and bias? The conversation explores the historical context of technology as a tool of oppression, the potential of AI to heighten awareness of biases, and the importance of integrating diverse perspectives into AI development to create equitable systems.
Generative AI’s environmental costs are soaring—And mostly secret.
In this article from Nature, Kate Crawford discusses how GenAI’s environmental costs, like energy and water use, are rising and largely hidden. The need for transparency, sustainability, and regulatory action to address these impacts is explored.
Portions of this page have been adapted from: Some Harm Considerations of LLMs | eCampusOntario H5P Studio