A Comprehensive Guide to AI-Enabled Action Research in the DOLI Program
- Introduction: The Emergence of the Centaur Scholar
- Theoretical Foundations: The Pedagogy of AI-Augmented Action Research
- The AI Ecosystem: Technical Capabilities of Gemini and NotebookLM
- Masterclass in Prompt Architecture: The Anatomy of a Doctoral Prompt
- Phase I: Diagnosis & Scoping (The “Plan” Phase)
- Phase II: The Planning Phase (Literature & Theory)
- Phase III: The Action Phase (Intervention Design)
- Phase IV: The Evaluation Phase (Analysis & Reflection)
- Phase V: Public Scholarship (The Iconic & Symbolic)
- Conclusion: The Future of the Augmented Scholar
- Works cited
1. Introduction: The Emergence of the Centaur Scholar
The landscape of doctoral education is undergoing a seismic shift, driven by the convergence of advanced artificial intelligence and the pragmatic demands of organizational leadership. For students in the Doctor of Organizational Leadership and Innovation (DOLI) program at City Vision University (CVU), this shift is not merely a backdrop but the very terrain upon which their scholarship is built. The program’s design, deeply rooted in the scholar-practitioner model, recognizes that the leaders of the future will not be those who work in isolation, but those who function as “Centaurs”—hybrid intelligences that seamlessly integrate human creativity, ethical judgment, and contextual wisdom with the computational power of AI.1
This guide serves as the definitive manual for navigating this new paradigm. It is specifically calibrated for students utilizing Gemini Pro (leveraging the capabilities of the Gemini 3 class of models) and NotebookLM within the context of Action Research. Unlike traditional academic guides that focus on static information retrieval, this document treats AI as a dynamic “cognitive partner” capable of participating in every stage of the research cycle—from the initial diagnosis of complex organizational problems to the rigorous evaluation of interventions.1
The DOLI program operates on a sophisticated theoretical architecture that blends Action Research, the Lean Startup methodology, and Jerome Bruner’s Spiral Curriculum. Within this framework, AI tools are not shortcuts; they are essential scaffolds that allow students to traverse the “Zone of Proximal Development,” performing research tasks at a scale and depth that would be impossible for an unassisted human scholar to achieve within the constraints of a doctoral timeline.1 By mastering the art of prompt engineering, DOLI students transform from mere consumers of information into “Curator-Synthesizers,” capable of managing vast streams of knowledge and converting them into actionable insights that drive distinctively Christian innovation in the social sector.1
This report will explore the theoretical underpinnings of AI-augmented research, provide an exhaustive library of specific prompts tailored to the Action Research cycle, and detail the rigorous “Trust but Verify” protocols necessary to maintain academic integrity in an age of synthetic media. It is a roadmap for the modern scholar-practitioner who seeks to change the world while studying it.
2. Theoretical Foundations: The Pedagogy of AI-Augmented Action Research
To write effective prompts, one must first understand the pedagogical machinery that drives the DOLI program. The prompts detailed later in this guide are not arbitrary queries; they are strategic instruments designed to facilitate specific cognitive movements within a spiral of learning. The effective use of Gemini Pro and NotebookLM is contingent upon aligning one’s interaction strategies with three core theoretical frameworks: the “Centaur” model of distributed cognition, the Lean Startup approach to research validation, and the cyclical nature of Action Research.
2.1 Distributed Cognition and the Centaur Model
The DOLI program posits that we have entered an era of “Distributed Cognition,” where intelligence is not solely resident in the biological brain but is spread across a network of digital tools and collaborative systems.1 The explicit requirement for students to maintain a Gemini Pro subscription is a recognition that the modern scholar is a “Centaur”—a human augmented by machine intelligence. In this partnership, the division of labor is distinct and critical. The AI, with its vast context window and processing speed, assumes the role of the “Research Assistant that never sleeps,” handling the lower-order cognitive tasks of sorting, summarizing, and pattern recognition across massive datasets.1 This liberation allows the human scholar to focus entirely on higher-order tasks: ethical reasoning, theological integration, contextual application, and the generation of wisdom.
When a student uses Gemini Deep Research to synthesize fifty academic papers on homelessness, they are not bypassing learning; they are engaging in “Scaffolding.” Just as a physical scaffold allows a construction worker to reach heights otherwise inaccessible, the AI allows the doctoral student to grapple with the “topography” of a literature field before diving into the granular details.1 This aligns with Vygotsky’s concept of the “More Knowledgeable Other,” where the AI serves as a provisional expert that helps the student perform beyond their current independent capability until their own “cognitive muscles” strengthen.1
2.2 The Lean Startup Interface: Innovation Accounting in Research
Traditional doctoral research often follows a “Waterfall” model—years of solitary planning and writing followed by a single, high-stakes defense. The DOLI program radically disrupts this by integrating the Lean Startup methodology, popularized by Eric Ries, into the research process.2 This approach emphasizes the “Build-Measure-Learn” feedback loop, which maps perfectly onto the Action Research cycle.
In this context, a research report or a literature review is not a final monument; it is a “Minimum Viable Product” (MVP).3 Students are encouraged to “fail fast” by generating rapid research outputs using AI, measuring their impact through peer feedback in the Community of Inquiry, and pivoting their focus based on data.2 The prompts provided in this guide are designed to facilitate this speed. For instance, creating a “Strategic Plan Analysis” using Gemini is not intended to produce a perfect final strategy but to create a “Research MVP” that can be tested against the reality of the student’s organization.3
This methodology introduces the concept of “Innovation Accounting” to doctoral work. Instead of measuring progress solely by pages written, students measure “Validated Learning”—evidence that their research is solving real problems.2 AI accelerates this cycle by reducing the “cycle time” of research. What used to take months (e.g., coding qualitative transcripts or mapping a knowledge domain) can now be drafted in days, allowing for more cycles of iteration and refinement within the degree program.2
2.3 The Spiral Curriculum: From Enactive to Symbolic
Jerome Bruner’s theory of the Spiral Curriculum provides the developmental logic for how students engage with AI tools throughout the program.1 Bruner proposed that learners progress through three modes of representation: Enactive (action-based), Iconic (image-based), and Symbolic (language-based).
- Enactive Mode (Action): In the early stages (e.g., course ORG700), learning is action-oriented. Students must physically engage with the tools—”Sign up for Gemini,” “Create a Notebook,” “Run a Prompt.” The prompts here are tactical, designed to build muscle memory and technological fluency.1
- Iconic Mode (Image): As students progress (e.g., ORG701), they use AI to summarize complex realities into “Icons”—visual infographics or audio overviews generated by NotebookLM. These iconic representations help students form a mental map or “gestalt” of the field before they can fully articulate it in writing.1
- Symbolic Mode (Language): In the final stages (Dissertation/Project), the student operates in the Symbolic mode. Here, the AI helps synthesize findings into the rigorous, abstract symbol system of academia—APA citations, theoretical frameworks, and formal argumentation.1
Effective prompt engineering requires the student to diagnose which stage of the spiral they are inhabiting. Are they asking the AI to perform a rote task (Enactive), summarize a vast field into a digestible visualization (Iconic), or help refine a complex theoretical argument (Symbolic)?
3. The AI Ecosystem: Technical Capabilities of Gemini and NotebookLM
The DOLI program utilizes a specific suite of Google tools, each serving a distinct function in the research workflow. It is imperative that students understand the technical distinctions between Gemini Deep Research and NotebookLM, as these dictate the strategy for prompt construction.
3.1 Gemini Deep Research: The Autonomous Agent
Gemini Deep Research (accessible via the Gemini Advanced/Pro subscription) is an agentic tool designed to browse the live web and perform autonomous information gathering.4 It is the engine for “broad exploration” and “scoping.” Unlike a standard chatbot that relies on a static training set, Deep Research can formulate multi-step research plans, browse websites, read content, and synthesize findings into long-form reports (often exceeding standard context limits).4
- Core Capability: Autonomous web browsing and multi-step reasoning. It can traverse from a general query to specific data points across hundreds of URLs.
- Primary Application: Initial literature reviews, competitive analysis of peer organizations, finding real-time data (e.g., funding trends, recent news), and generating “Source-Enriched” reports.4
- Prompt Strategy: Prompts for Deep Research must be comprehensive, specifying the scope (e.g., “last 5 years”), the types of sources (e.g., “peer-reviewed,” “government reports”), and the output format (e.g., “Google Doc export”).5
3.2 NotebookLM: The Source-Grounded Socratic Synthesizer
NotebookLM acts as a “Socratic Synthesizer” that is grounded strictly in the sources provided by the user.1 This is the “safe harbor” of the AI ecosystem. It does not “hallucinate” from the open web because its knowledge base is strictly limited to the user’s uploads (the “Ground Truth”).
- Core Capability: “Source-Grounding.” It can ingest up to 50 sources (PDFs, Google Docs, Audio) and reason across them. It also features a “Studio” for generating Audio Overviews (podcasts) and other media.4
- Primary Application: Analyzing specific sets of academic papers found via Deep Research, synthesizing a student’s own field notes, and creating public-facing content like podcasts or FAQs based on verified research.4
- Prompt Strategy: Prompts here are instructional (“Summarize this,” “Compare Author A to Author B”) or creative (“Create a podcast script”). The “Context” is the uploaded document set, so prompts can be more direct.
The following table summarizes the strategic use of each tool within the DOLI workflow:
| Feature | Gemini Deep Research | NotebookLM |
| Primary Function | Autonomous Web Agent | Source-Grounded Assistant |
| Data Source | The Live Internet (Open) | User-Uploaded Files (Closed) |
| Risk Profile | Moderate Hallucination Risk (Requires Verification) | Low Hallucination Risk (Strictly Bound) |
| Best For… | Finding new information, Lit Reviews, Market Scoping | Synthesizing existing info, Study Guides, Podcasts |
| DOLI Stage | Diagnosis, Scoping, Literature Search | Analysis, Synthesis, Public Scholarship |
4. Masterclass in Prompt Architecture: The Anatomy of a Doctoral Prompt
Writing prompts for doctoral research requires a higher level of precision than casual AI use. A vague prompt yields superficial results. The DOLI program advocates for a structured approach to prompt engineering, referred to as the “Anatomy of a Doctoral-Level Research Prompt”.5 Every major research prompt used in the program should contain the following five components to ensure the output meets the rigorous standards of scholar-practitioners.
4.1 Persona: Priming the Cognitive Lens
Assigning a persona primes the AI to access specific knowledge domains, vocabularies, and tonal registers.5 By telling the AI who it is, the student effectively constrains the search space to relevant professional frameworks.
- Academic Persona: “Act as a research expert in organizational leadership and practical theology.” 5
- Professional Persona: “You are the Director of Marketing for City Vision University…” 6
- Consultant Persona: “You are a Christian sociological consultant for the Gospel Rescue Mission movement.” 7
4.2 Task: Defining the Cognitive Operation
The prompt must clearly define the cognitive operation the AI is expected to perform. Students should use active verbs that map to Bloom’s Taxonomy, moving from simple recall to complex creation.5
- Analysis: “Conduct a deep analysis of the existing literature…” 5
- Synthesis: “Blend and integrate three perspectives…” 6
- Critique: “Critique the following argument using the principle of charity…”
- Creation: “Generate a comprehensive literature review…” 5
4.3 Context: Providing the Semantic Bedrock
Context provides the essential background information that allows the AI to understand the nuance of the request. Without context, the AI defaults to generic, “average” responses. For DOLI students, this often involves specifying the “Christian non-profit” or “Rescue Mission” context to ensure theological and operational alignment.5
- Example: “My research focuses on improving volunteer retention within faith-based homeless shelters. The organization is facing high turnover due to ‘compassion fatigue’…” 5
- Example: “Contextual note: This is an article for week 7 in the course. Students will have already read the attached articles on Radical Hospitality.” 6
4.4 Format: Structuring the Output
Explicitly defining the structure of the output is crucial for workflow integration. Whether the student needs a table for a spreadsheet, a narrative for a paper, or a script for a podcast, the AI must be instructed on how to present the data.5
- Data Structure: “Provide a table with columns for organization name, website, and annual revenue.” 7
- Academic Structure: “Structure the report with an executive summary, thematic sections with clear headings, and a concluding section on practical implications.” 5
- Presentation Structure: “Create a 10-slide pitch deck…” 4
4.5 Constraints: Enforcing Academic Rigor
Constraints are the primary mechanism for quality control. They define what the AI should not do and what sources it must prioritize. This is where the student enforces the “Trust but Verify” protocol within the prompt itself.5
- Source Constraint: “Prioritize sources from peer-reviewed academic journals published between 2015 and the present. Avoid opinion pieces.” 5
- Verification Constraint: “Only include information from sources for which a DOI is available.” 5
- Negative Constraint: “Do NOT say the words ‘hashtag’ or ‘outro’ in this audio overview.” 8
5. Phase I: Diagnosis & Scoping (The “Plan” Phase)
The Action Research cycle begins with Diagnosis—understanding the problem in its full complexity. In the DOLI program, this phase is characterized by “Broad Scoping” and “Domain Mapping”.9 The goal is to move from a vague intuition about a problem to a concrete, data-driven understanding of the landscape. Gemini Deep Research is the primary tool for this broad exploration.
5.1 Broad Organizational Scoping and Community of Practice Building
Scholar-practitioners do not work in isolation; they work within a “Community of Practice” (CoP). One of the first tasks in any DOLI project is to identify peer organizations that can serve as benchmarks or partners. The AI can be used to generate large datasets of potential CoP members, which would take weeks to compile manually.
Prompt Template: Building Peer Organization Lists 7
Persona: You are a research analyst for the non-profit sector.
Task: Identify as many nonprofit organizations as possible in the USA that identify themselves as.
Context: We are building a database for a Community of Practice to share best practices.
Format: Provide a table of these with columns for organization name and website.
Constraint: Focus on organizations with a clear digital presence.
Prompt Template: Specialized Niche Scoping 7
Task: Identify large Black (or Hispanic/Latino) churches that have separate nonprofit organizations that they have developed to provide social services to their larger community.
Format: Provide a table of these with columns for church name, church website, nonprofit organization name, and nonprofit website.
These prompts are designed to be “high recall”—they cast a wide net. The constraint “Provide a table” is strategic; it allows the data to be immediately exported to Google Sheets or Airtable, forming the backbone of the student’s “Research CRM”.8
5.2 Problem Identification and Best Practice Research
Once the community is mapped, the researcher must identify “Best Practices” to establish a baseline for intervention. This involves querying the AI to analyze the specific programs of the identified peers.
Prompt Template: General Best Practice Research 7
Persona: You are the of.
Task: Identify the best practices for Christian nonprofit organizations for.
Context: Focus on organizations that have expertise in.
Format: Narrative report with clear headings for each practice.
Prompt Template: Program-Specific Scoping 7
Task: Conduct research on the programs in organizations in the list below.
Context:.
Analysis: Categorize based on how substantial the programs are based on information provided and outcomes listed.
This workflow demonstrates the “Double Diamond” design process: first diverging to find many options (Step 5.1), then converging to analyze the most promising ones (Step 5.2). This allows the student to filter hundreds of potential peers down to a “Top 20” list for deep analysis.7
5.3 Data Enrichment for Strategic Validity
Scoping often requires specific metrics (revenue, staff size) to determine if an organization is a true peer. Gemini can act as a data enrichment tool, filling in the gaps in a student’s spreadsheet.
Prompt Template: Revenue Data Extraction 7
Task: Find the most recent annual revenue for these organizations.
Format: Put into a table with columns for Organization Name, Website, Annual Revenue (number only), Year Revenue Reported.
Constraint: If you cannot find the annual revenue for an organization, then leave the field blank rather than skipping a row. Keep in the original order.
Context:.
Insight: The instruction “Keep in the original order” and “leave the field blank” is a crucial prompt engineering technique. It ensures that the AI’s output aligns perfectly row-by-row with the student’s existing spreadsheet, preventing data mismatch errors during copy-pasting.8
6. Phase II: The Planning Phase (Literature & Theory)
Once the practical problem is diagnosed, the scholar-practitioner must ground their work in academic theory. This involves a shift from broad web scraping to rigorous, constrained searching for peer-reviewed literature. This is where the “Trust but Verify” protocol is most critical.
6.1 The “Trust but Verify” Literature Search
Gemini Deep Research can occasionally “hallucinate” citations—creating plausible-sounding but non-existent papers. To mitigate this, DOLI students utilize the DOI (Digital Object Identifier) as a hard constraint.
Prompt Template: Peer-Reviewed Focus with DOIs 5
Persona: Act as a library research assistant.
Task: I need to identify academic sources for my action research project on. Please find and summarize academic literature on this topic.
Constraint: For every source you use to construct your report, you must provide the full citation and its Digital Object Identifier (DOI). Only include information from sources for which a DOI is available.
Format: Structure the report thematically and provide a full list of the DOI-linked sources at the end.
Rationale: The DOI serves as a proxy for quality control. Since DOIs are almost exclusively assigned to formally published academic work, this constraint effectively filters out blogs, opinion pieces, and gray literature, significantly raising the scholarly rigor of the output.5
6.2 Gap Analysis: Identifying the “White Space”
A requirement for doctoral work is contributing new knowledge. AI can assist in identifying what is missing from the current literature—the “negative space” of the field.
Prompt Template: Research Gap Analysis 5
Persona: Act as a PhD-level academic research analyst.
Context: My dissertation topic is.
Task: Conduct a deep analysis of the existing literature and generate a report that explicitly identifies the key research gaps.
Format: Structure the report as follows:
- What We Know: A concise summary of established benefits.
- What We Don’t Know (Thematic Research Gaps): Detailed unanswered questions relevant to the faith-based context.
- Methodological Gaps: Analyze which research methods are most common and which are under-utilized (e.g., “While survey-based research is common, there is a lack of ethnographic studies…”).
6.3 Technical Workflow: Zotero and BibTeX Integration
Managing citations is a logistical challenge. DOLI students use Zotero as their “external brain.” To bridge the gap between Gemini and Zotero, students must prompt the AI to output data in BibTeX format, a universal citation standard.
Prompt Template: BibTeX Generation for Zotero 7
Task: Research.
Constraint: Where possible try to use sources that have a Digital Object Identifier (DOI).
Format: Include a Works Cited at the end in APA v7 format. After that section, include a Sources in BibTeX Format section including all sources in BibTeX Format.
Execution Workflow 7:
- Generate: Run the prompt in Gemini.
- Copy: Highlight the entire block of BibTeX code (from the first @ symbol to the last }).
- Import: Open the Zotero desktop app. Go to File > Import from Clipboard.
- Verify: Check the imported metadata against the original source. AI often capitalizes titles in “Title Case,” whereas APA requires “Sentence case.” This must be manually corrected.
7. Phase III: The Action Phase (Intervention Design)
In the Action Research cycle, the Act phase involves implementing a change—a new curriculum, a strategic plan, a marketing campaign, or an organizational policy. In the DOLI context, the student does not just write about the intervention; they create it. AI serves as the “Builder” of these Minimum Viable Interventions (MVIs).3
7.1 Developing Strategic Artifacts
Students typically need to create high-level organizational documents to drive change. AI can draft these based on the best practices identified in Phase I.
Prompt Template: Strategic Plan Analysis & Generation 7
Task: Identify strategic plans on the following websites to use as best practice models.
Analysis: Categorize by the most thoroughly developed to least developed.
Format: Provide a table with the organization name and direct link to the strategic plan.
Prompt Template: Fundraising Strategy Diversification 7
Task: Categorize the revenue of the following organizations and identify their revenue diversification strategies to use as best practice models for other nonprofits seeking to develop revenue diversification strategies.
Insight: By analyzing the strategic plans of peers first, the student can then prompt the AI to “Draft a strategic plan for that incorporates the best practices identified in the previous report, specifically the diversification strategy of.”
7.2 Curriculum and Educational Content Generation
For students whose intervention involves education (e.g., training staff or teaching clients), AI can draft lesson plans and reading materials. This prompts the AI to act as an instructional designer.
Prompt Template: Course Content Generation 6
Persona: You are a professor at City Vision University in a course on.
Task: Write a paper for students in the course that focuses on.
Synthesis: Blend and integrate three perspectives: 1) Customer Service, 2) Christian Ministry, 3) Clinical Counseling.
Evidence: Use examples of how these principles were applied and led to transformation from the attached documents.
Format: Provide a section at the end appropriate for discussion questions. Write in a way to avoid being overly technical so it is more accessible to a general audience of frontline workers.
Rationale: This prompt explicitly asks for “blending and integrating” perspectives. This forces the AI to synthesize interdisciplinary knowledge—a hallmark of the DOLI program’s holistic approach.6
7.3 Marketing and Outreach Interventions
For interventions requiring participant recruitment or donor engagement, AI can serve as a marketing strategist, helping to define the “Value Proposition” of the intervention.
Prompt Template: Niche Marketing Feasibility 6
Persona: You are the director of marketing for City Vision University.
Task: Conduct a feasibility assessment of using as a way to target this niche.
Context: Look at whether there are ways that City Vision could advertise on books of particular interest to our market niche.
Output: Determine the minimal annual spend to make this strategy viable.
8. Phase IV: The Evaluation Phase (Analysis & Reflection)
The final phase of the cycle involves analyzing data collected during the intervention and reflecting on its impact. This is where the AI shifts from a “creator” to an “analyst,” helping the student make sense of the “messy” data typical of Action Research.2
8.1 Qualitative Data Coding and Analysis
Action Research generates significant qualitative data—transcripts, interviews, field notes. While AI cannot replace human interpretation, it can perform “cognitive offloading” by coding data into categories.5
Prompt Template: Transcript Analysis and Coding
Task: Analyze the attached interview transcripts.
Analysis: Identify recurring themes related to. Code these themes into categories and provide a count of how many times each theme appears.
Evidence: Identify specific quotes that best exemplify each theme.
Prompt Template: Structured Data Conversion
Step 1: Create the Initial Table. Create a table that lists all courses from the transcript including columns for Term, Course Code, Course Name, Credits, and Grade.
Step 2: Add General Education Categories. Classify each course based on the following rules.
Step 3: Calculate the Total. Sum the Semester Hours.
Warning: Students must be extremely cautious with data privacy. No Personally Identifiable Information (PII) of clients or vulnerable populations should be uploaded to a public AI model. Use anonymized transcripts only.5
8.2 Sociological Reframing and Critical Theory
Doctoral work requires looking at problems through different theoretical lenses. AI can simulate these lenses, allowing the student to “Steelman” opposing arguments or view their work through a critical framework.
Prompt Template: Sociological Reframing 7
Persona: You are a Christian sociological consultant for the Gospel Rescue Mission movement.
Task: Research the various examples of framing in the following list of organizations.
Output: Provide 5 different potential reframing positions amalgamated from the examples that might help the movement achieve critical mass while remaining faithful to its core values.
Prompt Template: Steelmanning and Critique
Task: Provide instructions to students that would help them write to frame their opponents’ argument in the best possible terms to better understand it.
Context: Present various ways of framing that approach including: the principle of charity in rhetoric, steelmanning, the Ideological Turing Test, and Rapoport’s Rules.
Insight: “Steelmanning” (the opposite of strawmanning) is a critical intellectual exercise. Using AI to generate the strongest possible argument against one’s own thesis allows the researcher to strengthen their defense and anticipate criticism.
8.3 Innovation Accounting: Measuring Impact
Action Research requires measuring impact to decide whether to “Pivot or Persevere”.2 AI can help define the valid metrics for this decision.
Prompt Template: Defining Actionable Metrics 1
Task: Based on the Lean Startup methodology and Impact Accounting, identify “Vanity Metrics” to avoid and “Actionable Metrics” to track for a program focused on.
Analysis: Explain how to measure the “Social Validity” of the intervention from the perspective of the beneficiaries.
9. Phase V: Public Scholarship (The Iconic & Symbolic)
The DOLI program emphasizes Public Scholarship through “Open Notebook Science.” Students are expected to publish their ongoing findings to build thought leadership and validate their work in the marketplace of ideas.3 NotebookLM is the primary engine for this, allowing students to convert text-based research into “Iconic” media formats.1
9.1 Creating Audio Overviews (Podcasts)
The “Audio Overview” feature in NotebookLM creates a conversational podcast between two AI hosts. This is an excellent tool for “Working Out Loud”—narrating the research process to a broader audience.3
Workflow and Prompts:
- Select Sources: Ensure only the relevant research documents (PDFs, Google Docs) are selected in the Notebook.
- Generate Audio: Click “Audio Overview” -> “Generate”.
- Customize the Conversation: The student can direct the AI hosts’ behavior.
- Prompt: “Focus specifically on the ethical implications mentioned in the sources. Use a skeptical tone for the second host.” 4
- Prompt (for introductions): “Start by saying ‘In this episode of the Rescue Mission podcast, we are focusing on.’ Do NOT say the words ‘hashtag’ or ‘outro’ in this audio overview.” 8
- Post-Processing: Download the audio (MP3). Use a tool like audio-joiner.com to add a standard intro/outro music track (provided by the university) to professionalize the episode for publication on Substack.8
9.2 Visuals: Infographics and Slide Decks
NotebookLM’s Studio can also generate visual summaries, aiding in the “Iconic” representation of knowledge.
Prompt Template: Infographic Generation 4
Task: Create a timeline infographic showing the key milestones of this research.
Style: Use a professional, minimalist color palette.
Prompt Template: Slide Deck Generation 4
Task: Generate a 10-slide pitch deck for investors.
Content: Include one slide specifically comparing our findings to the current industry standard.
Insight: While NotebookLM generates the content and structure for these visuals, the actual image generation might require using Gemini Pro’s image generation features (e.g., Imagen 3). Students should take screenshots of these outputs or copy the text into PowerPoint/Canva for final design.8
10. Ethical Framework & Verification: The “Trust but Verify” Protocol
The most critical risk in using AI for doctoral research is the erosion of truth through “hallucination.” The DOLI program enforces a strict “Trust but Verify” protocol.5 This is not a suggestion; it is a requirement for academic integrity.
10.1 The Seven-Step Verification Workflow
Every AI-generated claim and citation must pass through this seven-step filter before inclusion in a dissertation 5:
- Isolate the Citation: Copy the author, title, and year from the AI report.
- Primary Verification: Search the title in the university library or Google Scholar. Does it exist? If not, discard immediately.
- Confirm Details: Do the author and journal match the real world? AI often attributes real articles to the wrong journals.
- Access Full Text: Download the actual PDF.
- Verify the Claim: Use “Ctrl+F” to find the specific statistic or argument the AI cited.
- Read in Context: Did the AI strip away nuance? Did it misunderstand the finding?
- Cite the Original: Cite the original paper, not the AI tool (unless discussing the AI process itself).5
10.2 Transparent Citing of AI Usage
When the AI itself is the source (e.g., an AI-generated summary or report), it must be cited transparently to allow faculty and peers to audit the “chain of thought.”
Standard Disclaimer 7:
“This report was generated by Google Gemini Deep Research using the prompt:. It was reviewed by X for accuracy.”
APA Citation for AI Reports 7:
Reference List: Gemini Deep Research. (2025). Report on Vocational Training Best Practices. Google.
In-Text: (Gemini Deep Research, 2025)
11. Conclusion: The Future of the Augmented Scholar
Mastering the prompts and workflows outlined in this guide is not merely a technical requirement for the DOLI program; it is a preparation for the future of organizational leadership. The “Centaur” model—the integration of human wisdom with machine intelligence—is becoming the standard for high-level decision-making.
By following these frameworks, students do more than just complete assignments. They build a personal knowledge infrastructure that scales with them. They move from the overwhelming noise of the information age to a structured, insightful, and actionable wisdom. They learn to diagnose with breadth, plan with rigor, act with precision, and reflect with depth.
As you engage with these tools, remember the core philosophy of the program: Technology is the scaffold; you are the architect. The AI can fetch the bricks and mix the mortar, but the vision, the ethics, and the final construction of the solution belong to you, the scholar-practitioner, acting in service of your organization and your community. The prompts in this guide are your blueprints; use them to build something that lasts.
Works cited
- DOLI Program: Theory and Practice, https://drive.google.com/open?id=1Mg9cZ2SBls0G4inh4zNQ8rlGVVpDtHOqeOJtSCTXN_Y
- Integrating Action Research, Lean Startup and Community of Practice Models, https://drive.google.com/open?id=15JfVibikQq2BSvqTh-JuWtpAQKz_BmAyBsQjDmJ1BU0
- Architecture of the DOLI Program, https://drive.google.com/open?id=1yNdiD3w3pkRVUCmWVxOw3xfw3oSo3sx8ZuXVeaeRU2s
- Instructions for Using NotebookLM for Research and Publication, https://drive.google.com/open?id=16crC1fdyOunyW49rnFIVauYGSAkA-9xEf1h6mKFpLOk
- AI Literature Review Guide , https://drive.google.com/open?id=1MANmRXhafOhJcLBxOSyomM-kckyQVBwYErtTiiElQMA
- AI Prompts, https://drive.google.com/open?id=16Itb_b8Nebfh1F9UV181amzyn_rwwkOFISahXSMqJ7Y
- AI Prompt Suggestions and Guidelines for Research at CVU, https://drive.google.com/open?id=14nmtZyriWS57NaPjUA7KN7mjHT9PDKirbpbMo-Xq5kk
- Newsletter/Podcast & Research Procedures & CVU AI Prompts, https://drive.google.com/open?id=1LuWKo40LBemjcJ5lRwOOrIlWp_K5M1NbeOzOgIruZos
- ORG701: Research Methods for Scholar Practitioners Course Template, https://drive.google.com/open?id=1_4sCulII8hJAdm4quHGKeXrfdADN_HNJX-rq8xa9OtM