Research Article
Hire AI: Teaching College Students AI Literacy by Managing Personal AI Assistants
Synthesis: Hire AI turns generative AI into a personal assistant that students hire, onboard, train and supervise over a semester. The authors designed the frame to reverse the relationship they saw in their own classes, where students deferred to the tool and hid their use for fear of being called cheaters. Two instructors at a large public research university ran the four-stage framework in business writing and in STEM career readiness, using an existing assignment plus documentation requirements rather than a new module. Students reported more confidence and better understanding of the tool's limits. The evidence is anecdotal: two courses, no comparison, no instrument.
Key Findings
- The metaphor does pedagogical work. Calling the tool an eager but untrained assistant gives students a concrete metaphor and shifts the power relation: they are the subject-matter experts, the tool is the junior colleague who needs briefing, correction and boundaries.
- The framework is an add-on, not a rebuild. Four stages run across a term: Groundwork, Discussion, Onboard Assistants, Coach and Collaborate. An existing assignment is minimally changed to state how the assistant may be used and to require a short record of prompts, failures and revisions.
- Teaching the tool teaches the student. The design leans on the protégé effect: students cannot coach an assistant through an assignment without first clarifying the conventions, audience and strategy the assignment requires. Several reported understanding the task better after explaining it to the AI.
- Supervision is where the literacy lives. Naming the assistant, assigning responsibilities and setting limits moved the work from output generation to judgment about output, which is the behavior the authors wanted students to practice.
- The documentation became the weak point. Required written performance reviews for every assignment were experienced as busywork, and students said so repeatedly. Chat transcripts, short audio or video reflections, and shared chat links are offered as lighter substitutes.
- Two disciplines, one frame. Business writing and STEM career readiness produced different assignments and the same structure, which is the authors' main claim for transferability.
Why the relationship, not the tool, is the target
The authors begin from a climate in which students ask whether they are allowed to use AI and whether they will be accused of cheating, while instructors spend their energy on integrity rather than on capability. Their diagnosis is that this arrangement teaches students to hide use and to accept machine output without critique. Reframing the tool as an employee answers both problems at once: use becomes visible and documentable, and the student's job becomes evaluation, briefing and correction. The claim is that this is where AI literacy is practiced, in the supervision rather than in the prompting.
What this means for practice
- Instructors. Start with the assignment you already set. Add one line about how the assistant may be used and one short reflection on what it got wrong, rather than building a separate AI unit.
- Instructors. Make the documentation light. A shared chat link or a two-minute voice note records the same evidence as a worksheet and does not read as busywork.
- Instructional designers. Keep the four stages visible in the syllabus. The framework's value comes from the sequence, and a course that runs only the "onboard" activity gets the metaphor without the supervision practice.
- Researchers. The design is cheap to evaluate: a pre/post measure of confidence and of the quality of students' critique of model output would test the claim the case makes.
Limitations
- No empirical data: two course implementations, no comparison condition, no instrument, and student reaction reported through quotations the instructors selected.
- The authors designed and taught both implementations, so the account is a design case rather than an evaluation.
- Both settings are undergraduate courses at one large public research university, and the transferability claim rests on two disciplines rather than on a sample.
- The framework assumes students have access to a general-purpose chatbot and are willing to share their conversations; neither is guaranteed in a course with privacy constraints.
Citation
Rodak, M. Y., & Braddock, B. V. (2025). Hire AI: Teaching college students AI literacy through the creation, "hiring," and management of personal AI assistants. International Journal of Designs for Learning, 16(2), 135–145. https://doi.org/10.14434/ijdl.v16i2.42010