What are the specific challenges of generative AI in the classroom? How can we use AI to help students learn better? How can we adjust assignments to prevent AI misuse? What are some best examples and/or practices of beneficial AI use among BMCC faculty? The answers to these questions first emerged from an exploratory discussion that we convened among our colleagues within the Department of Academic Literacy and Linguistics (ALL) and continued in a March 24th CETLS event with BMCC faculty members across academic departments.

We found that participants in both virtual discussions were eager to share their concerns, strategies, and potential models for addressing AI’s impact on academic integrity and pedagogy. To that end, we found that three key discussion points guided those robust conversations: 1. dealing with the challenges of AI-created or enhanced student work; 2. figuring out the appropriate approach to AI use in teaching; and 3. considering the broader pedagogical implications of AI use.

Challenges of AI in Student Work

Faculty members across disciplines expressed concerns about students using generative AI tools to complete assignments in their courses. Furthermore, a number of discussion participants pointed out that the challenges are more pronounced in asynchronous courses, where monitoring is especially difficult. This overarching concern about AI detection–or lack thereof–has led to strict policies that range from outright prohibitions to swift penalties.

Approaches to AI in Teaching

To that end, we discovered that ALL and other BMCC faculty members seem to utilize three primary models for integrating and/or resisting AI use in their courses: prohibitive, deterrent, and integration.

In the Prohibitive Model, or so-called “red light” approach (FULL STOP), instructors strictly forbid AI use in coursework and implement stringent rules, such as requiring handwritten papers. They also frequently use detection software to identify suspected AI-generated content; yet, they still expressed concern that prohibition may be a losing battle, as such tools continue to evolve and have become more difficult to detect over time.

In the Deterrent Model, or so-called “yellow light” approach (SLOW DOWN), instructors offered varied strategies for allowing some level of AI use within their courses. Those strategies include designing assignments that make AI-generated responses less useful; crafting prompts that require personal experiences or firsthand accounts; encouraging discussions with students about AI use and ethical concerns; including a rubric section assessing ‘personal investment in the answer’ to measure originality; and requiring students to conduct interviews as part of their assignments to ensure authenticity.

In the Integration Model, or so-called “green light” approach (GO AHEAD), instructors more fully facilitate AI use as a pedagogical framework within their courses. Again, they shared an array of avenues toward this purpose that include teaching students how AI works and its role in academic and professional settings; encouraging students to use AI for research while maintaining academic integrity; assigning comparative exercises where students write their responses to prompts and generate AI responses along with analyzing the differences; and discussing AI’s reliance on averages and predictability by highlighting contexts where these are valuable such as summarizing research, structuring essays, and presentations.

Lastly, faculty members across departments seem to agree that focusing solely on one model might not be desirable; instead, they believe that a multi-pronged approach may be the best way forward, including methods from all three models, depending on the context and teaching-learning scenario.

Broader Pedagogical Considerations

Both ALL and other BMCC faculty members debated what type of instruction remains essential in an AI-driven era. Finding balance in different disciplines and courses may require tailored approaches to AI integration or resistance. Indeed, encouraging students to critically engage with AI rather than simply prohibiting its use may foster deeper learning and adaptability.

Conclusion

The ALL Department and CETLS discussions underscored the need for a flexible, context-sensitive approach to AI in teaching. While strict prohibitions may work for some courses, deterrence and integration strategies can offer more sustainable solutions. Moving forward, we encourage all BMCC faculty members to experiment with the different models and share best practices to maintain academic integrity while embracing technological advancements.

We invite any BMCC faculty member to share their favorite AI tools, strategies, and/or teaching methods with us in the comments below.