As we examine the benefits and possibilities opened up by GenAI in teaching and learning, we also need to consider some current challenges and questions.
Privacy and Data Collection
Generative AI relies on learning from large amounts of data in order to generate new content. This data can include work that has bypassed proper copyright processes and not been permitted by creators, such as novels, artwork, and music, as well as personal information which can be used to identify individuals. As instructors, we need to be mindful that we do not submit original work without permission (e.g., student essay) nor personal information. Another primary consideration is that due to the “black-box” nature of generative AI tools created by third-party companies, educational institutions are not yet able to conduct proper Privacy Impact Assessments on the tools. VCC cannot at this point, assess the impact of the data collected by the companies, for what purposes the data will be used nor who will have access to the data. As such, VCC states that students cannot be required to use the tools and that alternatives must always be provided.
Equitable Access and Accessibility
There are free and differently priced models for generative AI tools. For example, ChatGPT have a subscription-based version that is more current and powerful than the free one. If you are encouraging students to use it, consider how your learning activities can be designed so that students are not disadvantaged if they are not using the paid versions of the tools. Students may also not have easy access to high-speed internet or tech devices, and that could impact their use of the tools. Consider options and alternatives that will allow students to meet learning outcomes without the use of the tools.
Keep in mind that accessibility principles and practices are still applied if and when using generative AI in your courses. The generative AI tools themselves ought to be navigable to all students and be aligned with assistive technology for those who need it. However, with the exception of Bing Chat, most generative AI interfaces are not yet well designed for accessibility with assistive technology. If you are teaching a course to students using assistive technology such as screen readers and keyboard navigation, implementing these tools may be an accessibility risk. Offer options so that students are not required to use the tools.
Perpetuation of Bias and Discrimination
Generative AI, being trained on large amounts of data, can reflect the bias, bigotry and discrimination found on the internet as a whole, which is then re-produced in the content it generates. Most generative AI's have been developed in the western world; therefore, they can also be culturally biased.
It is important to consider the prompts that we are inputting into the GenAI tool to attempt to reduce biased perspectives and perpetuate existing biases. It is also vital to vet the output before sharing with students to assess for accuracy and presence of bias. Likewise, if we choose to allow for GenAI use in our classes, it would be beneficial to encourage our students to use the tools with critical reflection and analysis. (KPU, 2023)
Chapman University provides an introduction to the places where bias can show up in AI as well as identify the types of bias that we can watch out for and apply critical analysis to whenever we are engaging with content generated by AI.
Unreliable Generated Content
Generative AI will provide a response whether it has the information or not. If it does not have any information, it will generate believable inaccurate information to satisfy the prompt or attempt to give you the “best” though not necessarily the most accurate, answer (“AI hallucination”). Generative AI also tends to frame its responses to questions as absolutes, as though every question has a single correct answer. It does not have the discernment built-in to determine whether the content generated is based on fact or not.
For example, when both ChatGPT 3.5 and Perplexity were prompted to “Write me a bio about Jennifer Benson, a nurse in the 1950s in New York City. She was known for her pioneering work with the ethics of care. Make it only 100 words or less.”, the content generated was:
ChatGPT 3.5 Output
Jennifer Benson, a trailblazing nurse in 1950s New York City, left an indelible mark with her pioneering work in the ethics of care. Dedicated to compassionate patient advocacy, she challenged traditional medical norms, emphasizing the importance of empathy and human connection in healthcare. Jennifer's commitment to holistic care redefined nursing practices, earning her widespread respect. Her legacy endures as a beacon for future generations of healthcare professionals, reminding them of the profound impact that genuine compassion can have on both patients and the healthcare system as a whole.
Jennifer Benson: Pioneering Nurse in 1950s New York City
Jennifer Benson was a pioneering nurse in 1950s New York City, known for her groundbreaking work with the ethics of care. She challenged traditional gender roles and advocated for a more compassionate approach to patient care. Benson's work contributed to the development of care ethics, a feminist ethic that emphasized the importance of empathy and emotional connection in caregiving. Her legacy continues to inspire nurses and ethicists to prioritize the human aspect of healthcare. Benson's contributions have left a lasting impact on the nursing profession and the ethical framework of healthcare.
The sources used for the bio about Jennifer Benson, pioneering nurse in 1950s New York City, are as follows:
- Jennifer Benson – Executive Director – Center of the American Experiment - LinkedIn
- Jennifer Benson – American Experiment
- Feminism and Feminist Ethics – Introduction to Philosophy – Rebus Press
- Ethics in Human Biology: A Historical Perspective on Present Challenges – Annual Reviews
- [PDF] Case Studies in Nursing Ethics
The issue above is that Jennifer Benson as a nurse during the 1950s in New York City did not exist, at least not in the way the output was generated. It is a made-up name, and the tools simply combine information and words from different and adjacent sources to create the bio. It is time-consuming for faculty, students, and librarians to check the sources to verify the accuracy of the content. This is especially true for references that is adjacent to or similar in some words to the content but is not directly connected. Note that while Perplexity (and Bing Chat) automatically includes sources, other tools may not.
It is important that we recognize that AI generated content represent a starting point that requires research, reflection, and thoughtful input. For example, the response given may help focus the direction of research but must not be taken as accurate or as research itself. In this, we need to be a subject matter expert and ensure AI-generated content is credible and reliable before distributing to students (unless inaccuracy is part of the assignment) or asking students to follow AI-generated procedures. It would be beneficial to Incorporate information and digital literacy training into the course and inform students that the content generated by AI may be biased and inaccurate. (KPU, 2023)
Digital and AI Literacy
While digital literacy is already part of an integrated framework in BC K-12 education, we are increasingly being faced with extending it to AI literacy in post-secondary education. As AI is finding its way more and more in everyday life, AI tools are also finding their way into workplaces, resulting in an expectation that graduates should have at least a basic understanding of AI to be able to interact effectively with the technology. As with moving towards a responsible and ethical use of GenAI, here are 7 ways to use GenAI in the classroom that helps with building AI literacy. Here is a sample lesson plan that you can adapt to facilitate discussions with your students about AI literacy.