In a world of instant content, the value of learning lives in how students think, apply, and transform understanding over time.
The recent attention around tools like the now-offline Einstein, an AI agent capable of completing coursework on a student’s behalf, has put academic integrity at the forefront of faculty conversations. But what matters less than the tool itself and more what it represents: a new category of AI that takes can remove the student from the learning process entirely.
And this isn’t the first time we’ve faced this type of tension.
For years, educators have known that many assessment practices reward task completion over genuine understanding. Agentic AI tools have accelerated the conversation by going beyond generating a draft or summarizing a reading. They can log in to online platforms, navigate course content, interpret assignment prompts, and submit work, all without the student ever engaging with the material. It has made visible a challenge that was already there and made it even harder to ignore.
Trying to “solve” this by chasing each new AI tool misses the larger point. Detection strategies and technical safeguards will continue to lag behind rapidly evolving technologies. This is not a race we can win through control alone. It is a moment to redesign.
That does not mean resisting AI. In fact, the opposite is true. AI holds enormous potential to enhance learning and is already proving to be enormously valuable in higher education, and its long-term impact will be transformative across nearly every aspect of the institution, supporting personalization, accessibility, and new forms of engagement. The goal is not to eliminate AI from the learning experience, but to ensure it is used in ways that support meaningful learning. That requires intentionality.
As institutions explore how to move forward, one thing is becoming clear: we must rethink how learning is assessed.
When content can be generated instantly, the value of learning shifts. It is no longer defined by what a student produces alone, but by how they think, apply, and transform understanding over time.
That shift calls for changes in assessment design…practical, grounded changes that can be implemented now. Here are some ways to begin that process:
1. Anchor Assessments to the Specific
The most effective way to reduce outsourcing is to design assignments an agent can’t easily fake.
Tie assignments directly to class discussions or lectures, current events, or a live scenario introduced in your course. An AI agent can write a strong essay about the ethics of fast fashion; it cannot write one that references the debate your class had last Thursday.
Project-based and scenario-based assignments further strengthen this approach. When students are asked to apply knowledge in context, they must demonstrate judgment, interpretation, and decision-making elements that are more difficult to automate.
2. Use Project-Based and Scenario-Based Work
Assignments that ask students to apply knowledge to real-world situations require judgment and context that is harder to manufacture. Build in specificity wherever possible.
3. Make Thinking Visible Over Time
Shift the emphasis from the final product to the process of arriving at it.
- Ask for all the scaffolding. Require students to submit planning notes, outlines, draft iterations, and annotated sources alongside final work. The documentation trail shows whether a student actually worked through the problem or assignment.
- Incorporate journals. Ongoing reflective journals throughout a course create a record of intellectual development that accumulates over time.
4. Vary Your Question Types
If your assessments rely heavily on true/false and multiple choice questions, they’re particularly vulnerable to automated completion. Consider mixing in:
- Fill-in-the-blank and matching questions
- Multi-part questions that require explanation across steps
- Timed assessments with question bank randomization
These aren’t foolproof, but they raise the difficulty for automated tools and reintroduce an element of real-time engagement.
5. Be There
This one doesn’t require any platform features or policy changes. Research consistently suggests that instructor presence is one of the strongest deterrents to academic dishonesty of any kind and AI is no different.
When instructors are actively engaged, e.g., giving substantive feedback, moderating discussions, following up individually, patterns of student work become visible. Anonymity decreases, and the relationship between student and instructor creates a level of accountability that technology alone cannot provide.
Regular and substantive interaction (RSI); active discussion moderation; timely and specific feedback: these aren’t just good pedagogy. In this environment, they also serve as an everyday defense for academic integrity.
Faculty who lean into authentic design, who stay present and engaged, and treat this moment as a pedagogical opportunity are strengthening education by designing learning experiences that are more meaningful, more resilient, and more human.
Lisa A. Clark, EdD, is the Associate Vice President of Academic Transformation at Blackboard LMS. Named one of the Top 50 Voices to Follow in Higher Education for 2026, she is recognized for her thought leadership in AI, pedagogy, and academic transformation. Dr. Clark serves on the Board of Directors of the United States Distance Learning Association and speaks internationally on the future of teaching, learning, and digital innovation.