Prompt, Evaluate, Revise: AI-Assisted User Manual Creation in Academic Practice

This study explores the effective use of artificial intelligence (AI) as a tool in technical documentation, focusing specifically on Turkish-language content creation. Conducted with graduate-level students in a translation quality standards course, the project investigates how AI-assisted writing can support the development of user manuals through a structured classroom project. The students were asked to generate specific sections of a user manual in Turkish by interacting with ChatGPT, using prompts aligned with the IEC/IEEE 82079-1 standard and core principles of technical writing. The project unfolded in four stages: prompt development, AI-generated drafting, human assessment of outputs, and human-guided and AI-assisted revision. Each student engaged in iterative prompting and post-editing, evaluating not only the quality of the AI-generated texts but also their own role in guiding and improving the results. As the assignment progressed, the students encountered and addressed various limitations of ChatGPT—such as visual recognition failures, incorrect safety symbol usage, and inconsistencies in warning categories—often learning through trial and error. By analyzing the prompts, outputs, and student-led revisions, the study offers practical insight into the potential and constraints of AI-supported documentation workflows. It highlights the importance of human expertise in both subject knowledge and AI interaction. Ultimately, the paper raises a central question for both education and industry: is AI a supportive means for technical communicators, or is it increasingly shaping the very goals and practices of the field itself?