Beyond Generation: Teaching Students to Use AI as Their Toughest Critic

by Kamyar Goudarzi, Ph.D., Assistant Professor of Management

I’ve been working hard this semester on a challenge I think many educators face: moving students past using AI merely for content generation and pushing them toward using it as a sophisticated feedback and refinement tool. If we want our students to thrive in a world saturated with generative AI, they must learn to integrate feedback, both human and algorithmic, to improve their work. I truly believe feedback is the catalyst for improvement, and learning to harness AI for this purpose is essential for intentional, high-value AI usage.

The Real-World Challenge:

To truly embed this practice, I needed an exercise that didn’t just deal with hypotheticals, but exposed students to real-world strategic work that was actually put up for critique. I chose a marketing strategy case that required deep conceptual analysis: the UVA Darden School of Business campaign concepts.

The core of the scenario involved Darden’s Chief Marketing and Communications Officer (CMCO), Julie Daum, who had commissioned Yes& Lipman Hearne (Yes&), a marketing agency, to develop new advertising concepts. Critically, the exercise placed my students in the exact same position as the CMCO: they were reviewing the three potential creative concepts presented by the agency. Students reviewed the same materials, including the creative concepts and a summary of stakeholder feedback, that the CMCO had on her desk. The three campaign options, “Trust Builds Better” (Concept 1), “Trust Makes It Work” (Concept 2), and “Made for Now” (Concept 3), were the focus of their analysis. The strategic element of the case centered on confronting a classic dilemma: inconsistent and conflicting feedback from key stakeholders after the agency’s pitch and needing to select a direction quickly.

Part of my instructions was that students had to critique the three marketing concepts first on their own (the “Human Review”) and then use AI (like ChatGPT or Gemini) to generate feedback. Crucially, they had to isolate the AI’s perspective, treating it as a companion critic, and then synthesize both critiques to form a final recommendation for the agency.

The Necessity of Intentional Prompting:

The biggest learning curve, and perhaps the most important necessity for future professional use, was crafting the AI prompts. The exercise quickly revealed that the quality of the feedback is entirely dependent on the specificity of the prompt, particularly when dealing with strategic complexity.

Students learned that vague prompts like “Which creative is best?” yield generic results, while better solutions required defining a clear role (like a Chief Marketing Officer, an actual student looking for graduate school, or a parent paying for graduate school) and specifying the target audience and strategic goals. For example, one team successfully used a prompt instructing the AI to evaluate the material from the perspectives of the CMO, an actual student looking for graduate school, and a parent who would be financing the degree. This multi-perspective analysis focused on engagement, consistency, and value proposition review.

This forced precision highlighted a key strategic necessity: if you don’t define the target audience and goals clearly in the prompt, the AI cannot deliver useful, targeted critique. This also reinforced the critical importance of choosing a clear target against which to focus all design, because trying to meet the disparate needs of many stakeholders often means not meeting any group’s needs well.

What the Students Took Away:

Watching the student presentations, two major takeaways emerged, solidifying the value of the exercise:

• AI as a Multi-Perspective Strategic Partner: The most crucial lesson was mastering intentional prompting, which transformed the AI from a general tool into a specific, multi-perspective critic. Students learned that effective AI use requires defining a specific role and target, such as asking it to evaluate the materials from the perspectives of the Chief Marketing Officer (CMO), a prospective student, or a parent paying for tuition. This structured approach allowed AI to deliver targeted strategic feedback, such as identifying that the campaign underplayed critical ROI and salary statistics from the CMO perspective, or missed the necessary financial justification demanded by parents. This exercise reinforced the principle that AI feedback is only useful when we guide the tool with clarity about the role, the target, and the strategic direction of the analysis.
• AI as an Aid to Judgment, Not a Replacement: The exercise powerfully underscored that AI is an aid to judgment, not a substitute for human decision-making, especially concerning brand uniqueness and authenticity. Students saw that while AI excelled at identifying gaps in tangible data and suggesting audience-segmented personalization, it often struggled with abstract elements like originality and the authenticity of the “vibe check”. This challenge immediately raised the critical question: “Where is human work or is it all AI?”. Ultimately, students concluded they needed their own human judgment to anchor the final recommendation in Darden’s unique, non-generic proof points, avoiding the risk of adopting a “polished but not bold enough” approach that blended into the competitive landscape.

Ultimately, the exercise proved that by separating their own analysis from the AI’s algorithmic critique, students moved beyond basic content generation and gained a powerful new skill: critical collaboration with AI. They learned that the true upside of AI integration requires thoughtful, intentional human guidance to steer the sophisticated tools toward high-quality, actionable insight and our goal would be to produce great thinkers who know how to use every tool available to sharpen and elevate their output.

Note: Notebook LM was used to draft the blog post and Nano Banana to generate images.