Three Perspectives on Using AI in Education

Questions about the pros and cons of using AI continue to pique educators and likely will for some time. This month’s featured articles offer three disciplinary perspectives about using AI. Think of this as a friendly but highly committed debate about the limitations, ethics, and practicalities of tools like ChatGPT for teaching and learning.

Dr. Joel Heng Hartse criticizes the use of generative AI tools like ChatGPT, particularly for writing. In a piece published in The Conversation, he and co-author Taylor Morhpett caution that even if AI-generated prose is grammatically correct or seems factual, it “does not reflect intellectual engagement with its subject matter.” A senior lecturer whose areas of research and teaching include first-year writing pedagogy, second-language writing, and graduate student writing, Dr. Heng Hartse notes that AI can’t think critically, make rhetorical judgements, or evaluate the validity of sources. He wrote an even more skeptical analysis of ChatGPT in The Vancouver Sun (“We’ve Admitted Defeat When We Say That ChatGPT Can ‘Write’ At All,” August 31, 2023).

Writing courses, however, serve as invaluable countermeasures, guiding students to see “the critical variety and power of one of our best technologies: the human act of writing, a system of finite resources but infinite combinations.”

AI tools like ChatGPT pose challenges such as promoting plagiarism, acknowledges Dr. Tenzin Doleck, an assistant professor and Canada Research Chair in Analytics for Learning Design. But does AI offer opportunities to help computing science students with their learning processes? In an article published in Computers in Human Behavior: Artificial Humans, Dr. Doleck and his co-authors describe comparing groups of students who could access AI for hints with those without access. Perhaps surprisingly, they found integrating AI assistance had a minimal impact on how students sought help. In this case, being able to tap into AI “does not necessarily reduce learners’ reliance on support strategies,” such as hints.

Though Dr. Doleck and his team believe AI holds much promise for increasing the productivity and accuracy of data science workflows, they argue that it is “crucial to acknowledge the significance of human expertise and critical thinking because AI systems are limited by their reliance on existing data and algorithms.”

For Dr. Daniel Chang and his co-authors, it is also crucial to regulate and properly integrate AI tools with educational design. A lecturer and researcher who focuses on technology-enhanced learning and writing, Dr. Chang urges educators to ethically and “proactively seek and explore ways to adapt” to AI rather than trying to suppress its use in education.

Ideally, generative AI tools like ChatGPT should support a student’s entire Self-Regulated Learning (SRL) cycle—goal setting, standard reproduction, iterative engagement with GAI systems, and solicitation of evaluative feedback. In reality, this is unlikely to happen. For that reason, Dr. Chang and his team call on designers of AI and online learning management systems to consider and integrate pedagogical principles like goal setting and planning, self-assessment, and personalization in generative AI tools. Collaboration among designers, educators, and other stakeholders can help ensure that AI tools effectively support student learning.

While Dr. Chang and his co-authors urge educators to accept AI as a contemporary educational trend, they insist it’s essential to “establish pedagogical principles for properly utilizing emerging generative AI technologies to promote self-regulation.”