Instructor: Janie Brennan, McKelvey School of Engineering
Course: Unit Operations Laboratory
TL;DR
Engineering students use a secure generative AI model trained on course materials to critique their own technical writing drafts. The student then manually incorporates the feedback before final submission and authors their own reflection log about the iterative process.
What is being done
In Janie Brennan’s writing-intensive engineering course, students learn to communicate clearly through technical documentation such as a lab report.
To replicate professional engineering quality assurance workflows, she instructs students to upload a draft of their report to a “Gem,” (a customized version of Google Gemini that allows the user to define specific instructions and knowledge files) inside WashU’s secure Google Gemini workspace. The Gem is trained on course rubrics and expectations with specific instructions to provide feedback on structural errors, stylistic flaws, and logical gaps without rewriting the uploaded report. The custom bot evaluates the report and the student can choose to ask follow-up questions for clarity. The student then must manually incorporate the edits they choose into their report.
Following revisions, the students then writes a 500-word reflection log documenting the revision process. The log includes:
- The AI snapshot: A copy of the feedback provided by the bot.
- The improvement: Three distinct gaps caught by the bot and how they resolved each one.
- The critical evaluation: Any feedback that the student disagreed with and how this revision process changed how they view their final document’s readiness.
The assignment positions generative AI as a source of feedback rather than a substitute for student writing or judgment. By requiring students to evaluate, selectively apply, and reflect on the bot’s recommendations, it reinforces revision as an active process and builds students’ ability to assess the quality and relevance of AI-generated feedback.
Adapting across disciplines
This approach can be applied in any course where writing is expected to follow clear disciplinary standards. In public policy, a custom AI tool could review draft policy memos for weak evidence, unclear recommendations, or missing stakeholder considerations. In business courses, students might use a bot trained on investor expectations to test whether a market analysis fully addresses competition, risk, and financial assumptions. In the humanities, the same model could help students examine whether an essay has a focused thesis, coherent organization, and appropriate use of sources.
The most important element is not the automated feedback itself, but the student’s responsibility for interpreting it. Students must decide which suggestions improve the work, which require modification, and which should be rejected.
Generative AI teaching repository
The assignment was developed through the AI Curriculum Corps in partnership with the Center for Teaching and Learning. The CTL Generative AI Teaching Repository serves as a centralized hub for WashU faculty to share, discover, and adapt innovative AI-integrated assignments. To explore peer-reviewed class activities or contribute your own framework, visit the CTL Generative AI Teaching Repository.
Join the AI Curriculum Corps
Looking to responsibly integrate AI into your curriculum? The AICC provides WashU faculty with technical support, peer feedback, and guidance on navigating secure university-supported AI tools. To apply for the next cohort of the AICC, visit ai.washu.edu/curriculum.