What is Prompt Engineering?

Prompt engineering is the art and science of crafting inputs to AI models that reliably elicit high-quality, relevant, and accurate outputs. It is the primary interface between human intent and AI capability — especially critical in educational contexts where precision, appropriateness and pedagogy matter.

Core Prompting Techniques

• Zero-shot prompting: Giving the AI a task with no examples. Works for simple, well-defined tasks. "Explain the water cycle in 3 sentences for a 10-year-old."
• Few-shot prompting: Providing 2–5 examples before the actual task. Improves consistency and format. Essential for structured outputs like rubrics or MCQs.
• Chain-of-Thought (CoT): Asking the AI to reason step by step before answering. Dramatically improves performance on multi-step problems. "Think through this step by step."
• Role prompting: Assigning a persona to the AI. "You are an expert pedagogy coach..." Improves domain-specific quality.
• Instructional scaffolding prompts: Breaking complex outputs into structured, staged sub-prompts.

Prompting for Teaching Use Cases

• Lesson planning: "Create a 50-minute lesson plan on photosynthesis for Grade 9, using the 5E instructional model."
• Differentiation: "Rewrite this paragraph at three reading levels: basic, intermediate, advanced."
• Question generation: "Generate 10 Bloom's Taxonomy questions (2 at each level) on the French Revolution."
• Feedback generation: "Review this student essay and provide formative feedback focusing on argument structure and evidence."
• Simulation: "Act as a Socratic tutor. Ask me guiding questions about Newton's Laws without directly giving the answer."

Prompt Quality Dimensions

• Clarity: Unambiguous instruction with defined scope
• Context: Grade level, subject, student profile, learning objectives
• Constraints: Word count, tone, format, output structure
• Examples: Anchoring outputs with samples reduces variance
• Iteration: Refine through follow-up prompts; treat it as a dialogue

Day 2 Session Highlights

Prof. Anusri P. provides a practical guide for educators on Prompt Engineering for Smart Teaching. The session emphasizes that AI should work as a teaching assistant that streamlines academic workflows rather than replacing teachers.

• Fundamentals: High-quality AI output depends on a clear prompt structure: role, task, context, audience, and goals.
• NotebookLM: Useful for managing PDFs, website links, YouTube links, and generating source-grounded content including audio or podcast-style outputs.
• Google AI Studio: Helps non-coders create interactive web applications and classroom demonstrations using simple prompts.
• Responsible AI: Avoid uploading sensitive student or institutional data into public AI systems; use privacy-focused settings wherever possible.
• Future outlook: Prompt engineering is a modern skill that helps teachers manage time and stay aligned with technology-aware learners.

Pitfalls to Avoid

• Vague prompts yield vague responses — specificity is everything
• Hallucinations increase with open-ended factual queries — always verify outputs
• Overloading a single prompt reduces coherence — decompose complex tasks
• Ignoring the audience parameter leads to age-inappropriate content

Session Overview

Day 3, led by Ms. Sabaita Mohsin, moves beyond basic prompting and introduces AI as a strategic partner for teaching, research discovery, student support, academic integrity, and workflow automation. The session highlights that educators do not need to become coders; they need to identify meaningful use cases, validate outputs, and keep humans in control.

Stages of AI Adoption in Academia

• Level 1 — Basic automation: Drafting emails, summarising notes, creating lesson outlines, and preparing simple assessments.
• Level 2 — Research discovery: Using AI as a research partner to identify papers, compare methodologies, extract limitations, and map knowledge gaps.
• Level 3 — Agentic workflows: Building AI agents that can perform multi-step academic tasks such as collecting student submissions, generating feedback, organising documents, and preparing progress summaries.

Advanced Research Tools Recommended

• Consensus: Helps locate evidence-based research findings and summarize what the literature says.
• Elicit: Extracts methodologies, sample sizes, findings, and research gaps across many papers.
• Research Rabbit: Builds a visual network of related papers and citation pathways for literature mapping.
• Sciite: Helps check whether a claim is supported, contradicted, or extended by later research.
• NotebookLM: Supports Q&A and synthesis from uploaded documents and source materials.

AI Agents and Workflow Automation

The session explains that AI agents can automate recurring academic workflows when the task has a clear goal, input source, evaluation criteria, and review step. Examples include student assessment support, feedback generation, content organization, performance monitoring, and document summarisation.

• AI loops: The system iteratively evaluates and improves its own output against a goal.
• Human-in-the-loop: The teacher or researcher reviews, validates, edits, and approves the final result.
• Goal-based execution: Instead of giving a single static prompt, educators define the objective, constraints, and success criteria.

Co-work, Claude, and Document Piles

Tools such as Co-work and Claude-style document workspaces can help educators manage large document piles by reading, organizing, summarizing, and synthesizing files. This is useful for research folders, student submissions, accreditation documents, project reports, and policy files.

Monitoring Student AI Usage

• Create AI policies: Clearly explain where AI is allowed, restricted, or prohibited.
• Focus on process: Ask students to submit outlines, drafts, reflections, revision histories, and tool-use disclosures.
• Require transparency: Students should mention which AI tools were used and how they were used.
• Verify claims: Teach students to check facts, citations, and possible hallucinations independently.
• Use detection carefully: AI detection tools may help, but final judgment must remain human.

Practical Teaching Applications

• Create custom exam prompts from lecture notes, PDFs, or lesson outcomes.
• Generate differentiated questions for multiple difficulty levels.
• Build rubrics aligned with learning outcomes and assessment standards.
• Use AI as a brainstorming partner for classroom activities and research projects.
• Use iterative prompting to refine outputs until they match the educator's goal.

Session Overview

Day 4, led by Dr. Anu Arora, focuses on how faculty members can use AI tools for assessment, evaluation, lesson planning, rubric creation, research productivity, and academic administration. The session is highly practical and connects AI usage with Outcome-Based Education, NAAC/NBA expectations, and ethical academic practice.

AI for Assessment and Evaluation

• Quizizz: Useful for quickly creating interactive quizzes and classroom assessment activities.
• QuestionWell: Generates multiple question types and helps create diverse assessment items.
• MagicSchool AI: Supports lesson planning, rubric generation, assignment design, and classroom productivity tasks.
• Negative marking: Can be implemented by defining scoring criteria or rubrics, such as assigning negative values for incorrect answers.

Research and Academic Productivity Tools

• ChatGPT: Useful for prompt engineering, curriculum support, assignment design, and structured academic drafting.
• NotebookLM: Helps synthesize uploaded documents and generate source-grounded explanations.
• Perplexity AI: Useful for literature review support with citations and web-grounded answers.
• Research Rabbit: Described as a Spotify-like tool for finding connected research papers through topic and citation networks.

Outcome-Based Education Alignment

AI-generated assessments should not be random. They must connect with course outcomes, programme outcomes, Bloom's Taxonomy, rubric criteria, and institutional quality frameworks such as NAAC and NBA.

• Map questions to learning outcomes before using them.
• Use AI to create rubrics, but validate criteria manually.
• Generate formative and summative assessment items at different Bloom's levels.
• Use AI to prepare feedback that is specific, constructive, and aligned with the rubric.

Action-Based Classroom Task

Participants were encouraged to complete a short challenge using AI tools: create a quiz, generate a rubric, design an assignment, and prepare a literature review outline. This activity shows how faculty can immediately integrate AI into daily academic work.

Ethics, Quality, and Integrity

• Verify AI-generated questions, answers, references, and rubrics before sharing with students.
• Avoid blind copy-paste from AI systems.
• Cite AI assistance where required by institutional policy.
• Protect student data and unpublished institutional content.
• Use AI as an assistant, not as an authority.

Key Takeaway

AI can reduce repetitive academic workload and make assessment more creative, but the educator's role remains central in ensuring relevance, fairness, accuracy, and outcome alignment.

Session Overview

Day 5, led by Dr. Suresh Kannaiyan, focuses on AI for research excellence, future classrooms, literature review acceleration, research paper writing, data visualization, coding support, and responsible research practice. The session closes the FDP by showing how educators can convert AI knowledge into practical academic productivity.

Traditional vs. AI-Assisted Research

The session compares traditional research workflows with AI-supported workflows. Manual research often requires long hours of searching, reading, extracting, summarizing, and organizing papers. AI tools can accelerate these steps by summarizing papers, identifying gaps, comparing methods, and preparing structured notes.

Literature Review with AI

• Use AI tools such as Claude to read and summarize complex research documents.
• Extract key findings, limitations, methodologies, datasets, and future scope from papers.
• Compare multiple papers to build a strong literature review foundation.
• Use AI summaries as a starting point, then verify claims and rewrite in your own scholarly voice.

Research Paper Structure

Dr. Suresh explains a practical roadmap for structuring a research paper. AI can help generate outlines, improve flow, proofread language, and organize sections, but the final contribution must come from the researcher.

• Abstract: Briefly state problem, method, result, and contribution.
• Introduction: Explain background, research gap, and objectives.
• Literature review: Compare previous studies and identify limitations.
• Methodology: Present the proposed approach clearly and reproducibly.
• Experiments and results: Show data, metrics, comparison, graphs, and interpretation.
• Conclusion: Summarize contribution, limitations, and future work.

AI for Coding, Experiments, and Data Visualization

AI can assist with coding research data experiments by generating starter code, synthetic datasets, machine learning workflows, visualizations, and performance comparisons. Tools integrated with Google Colab can help researchers create datasets, run models, and generate graphs.

• Synthetic data generation: Useful when initial data is unavailable for testing a workflow.
• Code generation: AI can create Python code for models such as Random Forest or Linear Regression.
• Experiment comparison: AI can help compare algorithms using metrics such as MAE, RMSE, accuracy, or error plots.
• Diagram generation: AI can support data-flow diagrams, architecture diagrams, and research workflow visuals.
• Refinement: For novel research, the researcher must provide the specific methodology and validate the generated code.

Ethics and Quality Control

• Do not copy-paste AI-generated research text directly into papers.
• Verify all citations, references, equations, claims, and data interpretations.
• Use AI to accelerate drafts, but rewrite with human authorship and academic integrity.
• Protect unpublished data and intellectual property when using commercial AI platforms.
• Document AI assistance where required by journal or institutional policy.

AI Challenge for Educators

The organizers announce an AI Challenge to encourage participants to apply the tools learned during the FDP. Participants are expected to build or demonstrate an innovative AI-supported academic workflow. Winners receive a cash prize and further support through an upcoming workshop.

Future Readiness

The session emphasizes that modern educators must learn to work with AI because academic workloads are increasing. AI can help teachers prepare materials, conduct research, generate experiments, manage data, and create future-ready classrooms while preserving human creativity and responsibility.