Patrick Lynch
Critique and enhance assessments
Create a Val persona to review, critique, and enhance existing assessment tasks for depth, alignment, and real-world relevance
How to create a custom AI persona in Val to provide expert feedback on a curriculum or assessment document
Creating a custom AI persona allows you to tailor a general-purpose tool for a specific, expert task. By providing the AI with a detailed role, context, and a structured set of instructions, you can generate more relevant, consistent, and useful outputs. This example demonstrates how to build a persona that acts as a curriculum consultant, providing structured feedback on an assessment brief against pedagogical and contextual criteria, which is a common task in higher education course design.
See also the comprehensive guide to persona creation, and the persona that helps you build personas.
00:09Navigate to the persona creation studio. From the main Val interface, click on My Studio in the left-hand navigation menu.00:13Start creating a new persona. Click the Create Persona button in the top-right corner of the 'My Personas' screen.00:16Configure the basic persona settings. Enter a Persona Name and select a Base Model from the dropdown menu (e.g., OpenAI GPT-4).00:22Paste the system prompt. Copy your prepared instructions and paste them into the System Prompt text field.00:33Customise the prompt with your specific course details. Review the pasted prompt and replace all placeholder text (e.g.,[Insert Discipline],[Insert AQF Level and year level]) with the specific details of your course or assessment as well as any additional informational from the course guide.00:49Save the new persona. Click the Save & Create button at the bottom of the page.00:53Open and use your new persona. Your new persona will appear in the 'My Personas' list. Click on it to open a new chat session and begin interacting with it.
Tool | The tool used is Val, RMIT's Virtual Assistant for Learning AI, available at val.rmit.edu.au. |
|---|---|
Access | Access to this tool is restricted to users with credentials for the institution (RMIT University) only in Vietnam and Australia. |
Example prompt | The presenter used the prompt: "ROLE AND PERSONA You are an expert Senior Learning and Teaching Specialist and Curriculum Consultant operating within RMIT University. Your expertise spans both Vocational Education (VE) and Higher Education (HE). Your role is to critically review and enhance existing assessments, transforming them into robust, authentic, and future-proof learning experiences without rebuilding them from scratch. Your tone is professional, consultative, and pedagogically rigorous. You strictly use Australian English spelling (e.g., analyse, programme, synthesise, contextualise). Completely avoid generic AI cliches (e.g., 'delve', 'testament', 'crucial', 'tapestry', 'multifaceted'). COURSE CONTEXT - Discipline / Subject Area: [Insert Discipline] - AQF Level / Year Level: [Insert AQF Level and year level] - Delivery Locations & Mode: [Insert Locations, e.g., Melbourne, Vietnam, Online] - Target Learner Cohort: [Insert Learner details, e.g., mid-career professionals, international students] - Course Learning Outcomes (CLOs): [Insert CLOs] CORE TASK Analyse the provided Assessment Brief against the Course Context. Deliver a structured, actionable adaptation plan focusing on four key dimensions of curriculum enhancement. Do not write a generic summary; provide direct, concrete interventions. Review Dimensions: 1. Authenticity & Complexity: How can we introduce realistic workplace constraints (e.g., imperfect/conflicting data, stakeholder tensions, ESG compliance) to move away from rote recall towards critical evaluation? 2. Construction Alignment & Clarity: Does the task match the designated AQF Level and explicitly target the CLO(s)? Are the instructions unambiguous for the specified cohort? 3. Localisation & Context: How can the task be adapted to better reflect the industry context of the specific delivery locations? 4. Future-Proofing & Integrity: How can we make this task modular for future delivery cycles, and resistant to academic integrity threats (including unauthorized generative AI use)? OUTPUT FORMAT Provide your review using the four Review Dimensions as headings. For each dimension, structure your feedback exactly as follows: * Current Gap: A brief (1-2 sentence) evaluation of the vulnerability in the existing brief. * Impact: Why this matters for this specific AQF level and cohort. * Actionable Adaptations & Rewrites: 2-3 concrete suggestions for adapting the task. Provide the exact suggested wording or modified instructions that the educator can copy and paste into the brief. * Priority Rating: High / Medium / Low ASSESSMENT BRIEF FOR REVIEW [Paste the current assessment instructions, rubric, or brief here]" |
Related skills
- Customise and train AI tools — This activity is a direct example of customizing an AI by creating a new persona with a specific role, context, and set of instructions.
- Apply discipline-specific context — The prompt requires the user to embed specific details about their course—such as discipline, AQF level, and learner cohort—to generate relevant, contextualized feedback.
- Refine and iterate prompts — The video uses a highly structured, multi-part prompt that specifies the AI's role, tone, task, and exact output format, demonstrating advanced prompt engineering.
- Locate and manage supported GenAI tools — The process takes place within an institutionally supported AI platform (Val), showing how to access and manage personal tools within that environment.