When to Use AI Tools — and When to Rely on Your Mentor

A student asked me last week: “I know I shouldn’t write my dissertation with ChatGPT, but where’s the line? When is AI helpful versus harmful? I’m confused about what’s okay.” This is exactly the right question. Because here’s the reality: AI tools aren’t going away, and pretending they don’t exist isn’t practical. But treating AI as a substitute for human mentorship will destroy your dissertation and possibly your academic career. The line is clearer than you might think: AI can assist with mechanical tasks. Human mentors must guide intellectual work. Let me show you exactly where that line is, why it matters, and how to navigate it successfully. Because used appropriately, AI can save you time on tedious tasks—but only human mentorship develops the scholarly thinking that makes dissertations succeed.

The Fundamental Distinction: Mechanical vs. Intellectual

Before discussing specific uses, understand the core principle separating appropriate from inappropriate AI use.

Mechanical Tasks: Where AI Can Help

What qualifies as mechanical:

• Tasks that don’t require judgment, reasoning, or original thinking
• Tasks where correctness can be objectively verified
• Tasks that save time without replacing intellectual work
• Tasks where AI assists YOUR process rather than doing your thinking

Examples:

• Formatting references according to APA rules
• Catching grammatical errors in text YOU wrote
• Suggesting search term synonyms for database queries
• Converting statistical output to formatted tables
• Checking whether documents meet formatting requirements

These tasks are tedious but don’t involve the intellectual work that defines doctoral-level scholarship.

Intellectual Tasks: Where Mentors Are Essential

What qualifies as intellectual:

• Tasks requiring reasoning, judgment, or scholarly expertise
• Tasks where multiple approaches exist and selection requires wisdom
• Tasks where your thinking is the actual work product
• Tasks where understanding matters more than output

Examples:

• Identifying meaningful research gaps
• Selecting appropriate theoretical frameworks
• Justifying methodological choices
• Interpreting data and findings
• Constructing scholarly arguments
• Defending research decisions

These tasks ARE the dissertation—they represent your scholarly contribution and cannot be outsourced. According to research from Yale’s Graduate School of Arts and Sciences, the ability to reason through intellectual decisions independently—while incorporating mentor guidance—is the primary competency doctoral defenses assess. Students who cannot articulate reasoning behind their choices fail defenses, regardless of document quality.

AI Use Cases: The Green, Yellow, and Red Zones

Let me be specific about what’s appropriate, questionable, and unacceptable.

Green Zone: Clearly Appropriate AI Use

These uses are ethically sound and help efficiency without compromising intellectual integrity: 1. Grammar and spell-checking Appropriate use: Running your draft through Grammarly or asking ChatGPT “Is this sentence grammatically correct: [sentence YOU wrote]” Why it’s okay: You wrote the content. AI is just checking technical correctness—like a copyeditor would. Important: AI is checking YOUR writing, not generating content for you. 2. Citation formatting Appropriate use: “Format this citation in APA 7th edition: [citation information for source YOU found and read]” Why it’s okay: You found the source, read it, and decided to cite it. AI is just handling formatting mechanics. Important: Never ask AI to “find sources about [topic]”—that’s you not doing your literature search. 3. Search term brainstorming Appropriate use: “I’m researching teacher retention. What additional search terms should I consider?” Why it’s okay: AI suggests terms you might not think of. You still do the searching, reading, and evaluating. Important: You must understand terms before using them and verify results are relevant. 4. Synonym suggestions Appropriate use: “What’s another word for ‘mitigate’ that might be clearer?” Why it’s okay: You know what you mean. AI is helping find the right word—like a thesaurus. Important: Only use words you actually understand. Don’t use sophisticated terms you can’t define. 5. Table formatting Appropriate use: “Convert this data into APA-formatted table: [data YOU analyzed]” Why it’s okay: You did the analysis. AI is formatting output for presentation. Important: You must understand the data and analysis. AI is formatting, not interpreting.

Yellow Zone: Proceed With Caution

These uses might be acceptable but require careful judgment: 1. Sentence clarity improvement Use case: “Make this sentence clearer: [sentence YOU wrote expressing YOUR idea]” Why it’s questionable: Where’s the line between clarity help and content generation? Proceed if:

• The original idea and content are genuinely yours
• You could explain the concept without AI’s help
• You’re improving expression of YOUR thinking, not having AI think for you
• You could defend the reasoning if questioned

Don’t proceed if:

• You don’t understand what the sentence means
• AI is generating new ideas or arguments
• You’re using AI to make vague ideas sound sophisticated

2. Outline structure suggestions Use case: “I have these main points: [YOUR points from YOUR analysis]. Suggest logical organization.” Why it’s questionable: Organization is somewhat intellectual, not purely mechanical. Proceed if:

• The content and ideas are entirely yours
• You understand why AI’s suggested structure makes sense
• You could have organized it yourself with more time/effort
• You’re getting organizational suggestions for YOUR ideas, not having AI generate the outline

Don’t proceed if:

• AI is determining what content should be included
• You don’t understand why the organization works
• AI’s structure changes your arguments substantially

3. Literature search strategy Use case: “I’ve searched [these databases] with [these terms] and found [these patterns]. What search strategy might reveal additional relevant research?” Why it’s questionable: Search strategy involves judgment about relevance and coverage. Proceed if:

• You’ve already done substantial searching yourself
• You understand your field well enough to evaluate AI suggestions
• You’re getting supplementary search ideas, not outsourcing searching
• You critically evaluate all sources found through AI suggestions

Don’t proceed if:

• You haven’t done your own searching first
• You’re relying on AI to determine what’s relevant
• You cite sources without reading them because AI suggested them

Red Zone: Never Appropriate

These uses constitute academic dishonesty and will result in failure or dismissal: 1. Content generation NEVER APPROPRIATE: “Write my literature review about [topic]” or “Generate my problem statement” or “Write my methodology chapter” Why it’s dishonest: These sections require YOUR reasoning, YOUR synthesis, YOUR justification. Having AI generate them is plagiarism—presenting AI’s work as yours. Consequences: Proposal rejection, failure, possibly program dismissal for academic integrity violations 2. Research question formulation NEVER APPROPRIATE: “Based on this literature, what research questions should I ask?” Why it’s dishonest: Formulating meaningful research questions demonstrates doctoral-level thinking. If AI does this, you’re misrepresenting your capabilities. Consequences: You won’t be able to defend your questions or explain why they matter 3. Data analysis or interpretation NEVER APPROPRIATE: “Here’s my data. Identify themes/run analysis/interpret findings.” Why it’s dishonest: Analysis and interpretation are the core intellectual work of empirical research. Outsourcing them means you didn’t do the research. Consequences: You can’t defend your analysis or explain your reasoning. Possible research misconduct charges. 4. Theoretical framework selection or application NEVER APPROPRIATE: “Which theory should I use?” or “Apply this theory to my research” Why it’s dishonest: Theoretical reasoning is central to doctoral work. You must understand and justify theoretical choices. Consequences: Defense failure when you can’t explain theoretical connections 5. Citation without reading sources NEVER APPROPRIATE: “Summarize this article [PDF]” then citing the article based on AI’s summary without reading it yourself Why it’s dishonest: You’re claiming to have engaged with work you haven’t read. This is fabrication. Consequences: Committee catches you not knowing sources you cited. Academic integrity violation.

When You Need Your Mentor: The Essential Guidance Moments

Now let me show you specifically when human mentorship is irreplaceable—where AI cannot substitute even if you’re tempted to try.

Moment 1: Identifying Research Gaps

What’s needed: Analyzing patterns across literature to identify what’s genuinely missing and worth studying Why AI fails: AI describes what exists but cannot assess what’s missing or whether gaps matter What mentors provide:

• Field knowledge to recognize understudied areas
• Judgment about which gaps are meaningful versus trivial
• Strategic thinking about gaps you can feasibly fill
• Understanding of what your committee will view as original

The conversation: You: “I’m interested in teacher retention. How do I find a gap?” Mentor: “Tell me what you’ve found so far in your searching.” You: Share findings from systematic searching Mentor: “Notice how most studies examine salary but few examine working conditions in high-poverty schools specifically? That’s a meaningful gap because those contexts face different constraints. Let’s explore whether that’s really understudied…” This guided discovery—helping you learn to identify gaps while verifying your judgment—cannot happen with AI.

Moment 2: Selecting Methodologies

What’s needed: Matching research methods to your specific questions, context, and constraints Why AI fails: AI describes methods generically but doesn’t understand epistemology, design logic, or feasibility assessment What mentors provide:

• Understanding of what different methods can and cannot reveal
• Judgment about which methods fit your questions
• Assessment of what’s feasible given your constraints
• Preparation for defending methodological choices

The conversation: You: “Should I use phenomenology or grounded theory?” Mentor: “What are you trying to understand? Are you exploring the essence of an experience or generating theoretical explanations of processes?” You: “I want to understand how teachers experience administrative support—what makes actions feel supportive versus controlling.” Mentor: “That focus on meaning and essence suggests phenomenology. Grounded theory would be appropriate if you were developing theoretical models of how support relationships develop over time. Let’s think through the implications of phenomenological design for your data collection…” This reasoning about method-question fit requires expertise AI lacks.

Moment 3: Interpreting Findings

What’s needed: Making sense of what data mean theoretically and practically Why AI fails: AI can describe findings but cannot reason about theoretical implications or practical applications What mentors provide:

• Theoretical knowledge to connect findings to existing frameworks
• Field expertise to assess whether findings are surprising or expected
• Judgment about what findings mean for practice or policy
• Guidance on avoiding overinterpretation or underinterpretation

The conversation: You: “Three teachers mentioned lack of resources. What does that mean?” Mentor: “How did they describe resource lack? Physical materials? Time? Support staff? Each has different theoretical and practical implications. Also, three mentions from how many participants? We need to assess whether this is a dominant theme or minor issue. Walk me through the specific contexts where resources came up…” This analytical guidance—helping you interpret data rigorously—requires human expertise.

Moment 4: Constructing Arguments

What’s needed: Building scholarly arguments that synthesize evidence and advance claims Why AI fails: AI describes but doesn’t argue. It lists points without constructing logical chains of reasoning. What mentors provide:

• Modeling of how to build arguments from evidence
• Feedback on logical gaps or unsupported claims
• Guidance on acknowledging alternative interpretations
• Teaching how to position yourself in scholarly debates

The conversation: You: “I’m trying to argue that administrative support matters for retention.” Mentor: “That’s too general. What specifically about support matters? For whom? Under what conditions? Your data suggests support matters more for early-career teachers than experienced ones. That’s your argument—support needs vary by career stage. Now build that argument: start with theory predicting why support needs differ, show your evidence of differential effects, explain what this means for support program design targeting different career stages…” This instruction in argumentation—the heart of scholarly writing—requires human teaching.

Moment 5: Defending Your Work

What’s needed: Preparing to explain and justify every research decision under questioning Why AI fails: AI can’t anticipate your committee’s questions or prepare you to think on your feet What mentors provide:

• Mock defense practice with questions committees will actually ask
• Feedback on your reasoning and explanations
• Guidance on handling challenges to your decisions
• Preparation for discussing limitations without defensiveness

The conversation: Mentor: “Your committee will ask why you chose 15 participants. Explain your reasoning.” You: “Um, that’s what the guidelines say?” Mentor: “That’s not reasoning. Why is 15 appropriate for YOUR study specifically? Think about information power—your aim is narrow, your sample is specific, your interview guide is structured. These factors mean fewer participants needed. Now explain that logic…” This defense preparation—teaching you to articulate reasoning under pressure—requires human coaching.

How to Use Both AI and Mentors Strategically

The most effective approach: use AI for mechanical efficiency while relying on mentors for intellectual development.

The Strategic Workflow

Step 1: Get intellectual guidance from your mentor Work with your mentor to:

• Identify meaningful research gaps
• Select appropriate theories and methods
• Develop research questions
• Design data collection procedures
• Plan analysis approaches

Step 2: Do the intellectual work yourself Execute the work yourself:

• Conduct literature searches and read sources
• Collect and analyze data
• Interpret findings
• Draft content expressing YOUR thinking

Step 3: Use AI for mechanical assistance Let AI help with tedious tasks:

• Check grammar in content YOU wrote
• Format citations for sources YOU read
• Create tables from data YOU analyzed
• Generate alternative search terms for additional searches

Step 4: Get mentor feedback on intellectual work Share drafts with your mentor for:

• Assessment of reasoning quality
• Feedback on argumentation
• Identification of gaps or weaknesses
• Guidance on improvements

Step 5: Iterate with mentor guidance Revise based on mentor feedback:

• Strengthen arguments
• Add missing justifications
• Clarify reasoning
• Improve synthesis

What This Workflow Achieves

Efficiency: AI saves time on mechanical tasks Quality: Mentor guidance ensures intellectual rigor Learning: You develop scholarly capabilities through mentor teaching Defense readiness: You understand your work because you did it with human guidance Integrity: Your work represents YOUR thinking, informed by mentorship

Why Mentors Cannot Be Replaced by AI

Even if AI becomes more sophisticated, human mentorship provides irreplaceable value.

Mentors Know You

Good mentors understand:

• Your strengths and areas needing development
• Your learning style and what types of guidance you need
• Your specific program’s requirements and culture
• Your career goals and how your dissertation supports them

This personalized guidance helps you develop as a scholar, not just complete a document.

Mentors Know Your Field

Expert mentors bring:

• Decades of field expertise AI cannot replicate
• Understanding of what your specific committee values
• Knowledge of current debates and emerging directions
• Connections and professional insights benefiting your career

This contextual knowledge shapes advice AI cannot provide.

Mentors Teach Scholarly Thinking

Mentors don’t just help you finish—they teach you to think like a scholar:

• How to evaluate evidence critically
• How to construct rigorous arguments
• How to position yourself in scholarly conversations
• How to handle intellectual challenges

These thinking skills serve your entire career, long after AI tools evolve.

Mentors Hold You Accountable

Mentors provide:

• Motivation and encouragement through difficult phases
• Accountability for progress and deadlines
• Perspective when you’re overwhelmed or discouraged
• Celebration of milestones and achievements

This human relationship sustains you through the dissertation process.

Get the Right Kind of Mentorship

Don’t try to write your dissertation with ChatGPT when you need human expertise. Work with mentors who develop your scholarly capabilities.

What Real Professors Provide

When you work with us, you get mentorship AI cannot replicate: Strategic guidance: We help you identify viable research directions based on decades of field experience Methodological expertise: We teach you research design logic, not just describe methods Analytical training: We guide your data interpretation, ensuring rigor and depth Argument development: We teach you to construct scholarly arguments that convince committees Defense preparation: We prepare you to defend every decision under questioning Career development: We position your work to support your post-PhD career goals Get dissertation mentorship from scholars who develop your thinking, not just help you produce documents.

Our Balanced Approach

We embrace appropriate technology while maintaining intellectual integrity: We teach you to use AI ethically: Showing you exactly where AI can help and where it undermines your work We guide your intellectual development: Ensuring you understand your dissertation deeply enough to defend it confidently We prepare you for success: Not just graduation, but for careers requiring scholarly capabilities We maintain standards: Upholding the integrity of doctoral work while helping you work efficiently

Complete Dissertation Support

Get comprehensive dissertation help that combines efficient use of available tools with irreplaceable human mentorship.

The Bottom Line: Tools Assist, Mentors Teach

AI tools can help with mechanical tasks—formatting, grammar, search term brainstorming. But they cannot provide the intellectual guidance and scholarly teaching that doctoral work requires. Use AI for:

• Grammar checking
• Citation formatting
• Search term suggestions
• Table formatting
• Synonyms and word choice (for concepts you understand)

Rely on mentors for:

• Identifying meaningful research gaps
• Selecting appropriate methodologies
• Interpreting findings
• Constructing scholarly arguments
• Defending research decisions
• Developing as a scholar

Don’t try to write your dissertation with ChatGPT and miss the mentorship that develops capabilities serving your entire career. Use AI strategically for efficiency while getting human guidance for intellectual work. The line is clear: AI assists with tasks. Mentors develop thinking. Know the difference, and you’ll use both effectively.