“The Making of….Statement on AI and Assessment”
The “Statement on AI and Assessment,” drafted by the MLA Task Force on Research and Teaching and approved by the MLA Executive Council, responds to the rapid integration of AI technology and “agents” into the apparatuses of our work: from the AI summary icons in our Canvas discussion forums to products being launched by publishing and ed tech companies that promise to pre-sort and group student assignment submissions. The substitution of human evaluation for AI “grading” is on the horizon (if not already here). In this blog post, we write about motivations for taking a direct position on the use of AI for assessment.
Educators are facing increasing encouragement, pressure, and even mandates to use technology to provide students with feedback on their written work. Despite outrage about Einstein, an AI agent that was pitched to students as a way to complete tasks in their learning management systems, Instructure (which owns Canvas, the LMS used by many institutions nationally), announced the launch of its own agent. Referencing unspecified “low educational value” tasks, the company promised to “orchestrate” these “time-consuming” activities with the goal to “amplify human potential” (Watkins). Educators find themselves grappling with the implications of these automatically-integrated features into systems that they are often required to use, whether they want to or not.
We hope the statement can help instructors by articulating the core value of the organization: that education is not a transaction between teachers and students of words for points or content for grades. Education is relational, and the ultimate goal is the building of skills, knowledge, and dispositions that authentically assess student progress and motivate students to grow. We hope that instructors can use the statement both as a touchstone for their own thinking and as a message to circulate within their own local contexts.
We recognize this statement barely scratches the surface. Those of us who teach writing, literature, and digital humanities find nothing ground-breaking about a caution against using AI in place of our own time and expertise to provide feedback to our students. No one wants to see the realization of the dead classroom, where students’ bots write for their professors’ bots, and learning is not only beside the point but a distraction to the efficiency of credential production. This statement was designed to be a tool of advocacy for colleagues in the humanities to use as a bulwark against pressure from administration and those outside our fields to “leverage the efficiency” of AI in uncritical ways.
But of course “assessment” implies much more than AI-assisted grading. In the coming weeks, the MLA will be providing more detailed guidance around assessment that is more targeted to our teacher-scholar members. For the past 8 months, the MLA AI and Pedagogy Working Group has been putting together frameworks and resources to help faculty who are facing the difficult reality that many time-tested and formerly effective assignments and grading practices just don’t work anymore.
The forthcoming working paper, “Designing Assessment for Motivation and Accountability: An MLA Framework for Responding to GenAI” lays out a two-pronged approach for AI-aware assessment design that promotes both transparency and engagement, as well as clear guardrails and accountability measures.
Generative AI has been a disruption to humanities instruction, but it hasn’t changed the fundamental principles we all hold at the center of our practice. The argument has been made many times in the last few years that college students need the skills humanities teach now more than ever. It’s in our name. We are teaching the one thing AI can never be—human-centered.