Rethinking Assignments in an AI World by Moving from Product to Process

by | Oct 26, 2026 | ETS Instructors

For decades, the traditional assignment loop followed a predictable routine: a teacher assigned an essay, problem set, or research paper; students completed it independently at home. However, generative AI tools have altered this approach. Now, a chat interface can draft a coherent five-paragraph essay in ten seconds or solve complex calculus problems with step-by-step explanations. Educators can no longer rely on standard assignments as the primary measure of completed work and student understanding.

Trying to police assignments through AI detectors or blanket bans creates an adversarial classroom culture and fails to address the underlying issue: when the final artifact can be generated effortlessly, the artifact itself loses its value as evidence of learning.

To respond effectively, educators are shifting the focal point of assignments from the static product to the dynamic process.

1. Pair Writing with Oral Defenses

Bringing short oral defenses into K–12 classrooms ensures that students can explain the mechanics behind their completed work.

How it Works in Practice

  • The Two-Minute Defense: Rather than collecting a take-home essay in isolation, pair the written submission with a brief, targeted conversation. Ask questions like: “Why did you choose this source over another?”, “What was the strongest counterargument you considered?”, or “How would your thesis change if X variable were altered?”
  • Video Explainers: If live individual defenses present time constraints, have students submit a 60- to 90-second video recording (using tools like Loom or Flip) where they walk through their outline, highlight their key arguments, and explain how they revised their early drafts.

2. Prioritize Live Problem-Solving and “In-the-Lab” Work

When student work happens entirely outside the classroom, teachers lose visibility into the cognitive heavy lifting. Moving core synthesis work back into the physical classroom restores that visibility.

How it Works in Practice

  • Whiteboard Sessions and Live Modeling: Have students solve complex problems, map out cause-and-effect diagrams, or outline essays collaboratively on whiteboards or shared digital canvases while the teacher circulates to observe reasoning in real time.
  • Process Logs and Revision Histories: If long-form written work is assigned as homework, evaluate the evolution of the document. Require version histories, brain dumps, peer feedback notes, and draft iterations to be submitted alongside the final draft. The homework grade reflects the refinement process just as much as the final text.

3. Emphasize Project-Based Learning (PBL)

Traditional assignments often ask students to summarize existing information, a task at which AI excels. Project-Based Learning shifts learning tasks toward solving open-ended, real-world problems that require local context, community engagement, and hands-on application.

How it Works in Practice

  • Local and Context-Specific Tasks: Assign projects connected to the immediate school or local community (e.g., analyzing local water quality data, interviewing community members, or designing a proposal for a local park). Generative AI lacks access to uncollected real-time local data and human experiences.
  • Multi-Modal Deliverables: Instead of a standard written report, require students to present their thinking through physical prototypes, public presentations, podcasts, or interactive community exhibits where they must interact with an audience.

A Framework for Reworking Assignments

Shifting from product to process requires adjusting how we structure learning objectives and evidence of completed work across core tasks:

  • Essays & Writing Assignments
    • Traditional Product-Focused Model: Assign a 1,000-word essay at home on the causes of the American Revolution.
    • Process-Oriented AI-Resistant Model: Students draft an outline at home (AI permitted for brainstorming), then defend their argument in a 2-minute live Q&A in class.
  • STEM & Problem Sets
    • Traditional Product-Focused Model: Complete 20 calculus practice problems overnight for a grade.
    • Process-Oriented AI-Resistant Model: Students solve 3 core problems in class on whiteboards, verbally explaining their steps and choices to a partner.
  • Research Homework
    • Traditional Product-Focused Model: Turn in a static research paper on local environmental policy.
    • Process-Oriented AI-Resistant Model: Conduct a field study on a local ecosystem, present findings to a live panel, and document research iterations in a process journal.

Critical thinking, spontaneous articulation, iterative design, and authentic problem-solving cannot be outsourced to an algorithm. By designing assignments that makes student thinking visible, educators build classrooms that value the messiness, effort, and growth of real human learning.

Share This