AI Suggests Impractical Topics — Feasibility Requires Human Judgment
A doctoral student contacted me last month, frustrated and confused. She’d spent six weeks developing a dissertation
proposal with ChatGPT’s guidance. The topic seemed perfect: studying the lived experiences of adolescent survivors of
sexual abuse to understand their healing processes. She submitted it to her IRB (Institutional Review Board). It was
rejected immediately. Her advisor told her this topic would never be approved—adolescents are minors, sexual abuse
survivors are a vulnerable population, and the psychological risks of asking them to recount trauma are too high for a
dissertation project. Six weeks wasted because AI suggested a topic without understanding basic feasibility constraints
that any experienced researcher would have flagged immediately. Here’s what students need to understand: a dissertation
topic can be original, problem-driven, and intellectually interesting, but if it’s not feasible, it’s not viable. And AI
has no ability to assess feasibility because it doesn’t understand IRB requirements, data availability, access
constraints, or measurement validity issues that determine whether research can actually be conducted. Only human
scholars with research experience can make those judgments.
Let me break down what makes a dissertation topic feasible, because students often don’t understand all the constraints they’re working within.
Your dissertation research must get IRB approval. That means your study must: Minimize risks to participants: Physical, psychological, social, legal, or economic risks must be minimized and outweighed by potential benefits. Protect vulnerable populations: Extra protections required for children, prisoners, pregnant women, mentally disabled individuals, economically disadvantaged people, and victims of trauma or crime. Ensure informed consent: Participants must understand what they’re consenting to and be able to consent voluntarily without coercion. Maintain confidentiality: You must protect participant identity and data security. Avoid deception: Unless absolutely necessary and justified, you can’t deceive participants about the research. Some dissertation topics can meet these requirements with appropriate protocols. Others cannot, no matter how carefully designed. AI doesn’t understand this distinction.
Your dissertation needs data you can actually access within your program’s timeframe: Secondary data: If you’re using existing data, it must be available to you. Some datasets require special permissions, cost money, or restrict access to certain researchers. Primary data collection: If you’re collecting your own data, you must be able to recruit participants, access research sites, and complete collection within your timeline (typically allowing 6-12 months for defense after collection). Temporal requirements: If your research requires longitudinal data or examines effects of recent policies/programs, sufficient time must have passed for those effects to be observable. Some topics sound great but require data that doesn’t exist yet or that you can’t access. AI can’t evaluate these practical constraints.
Even if you can collect data, it must be valid and useful: Social desirability bias: If you’re asking about controversial or sensitive topics, people might not answer honestly, making your data invalid. Self-report limitations: Some things people can’t accurately self-report (unconscious biases, certain behaviors, complex decision processes). Observation challenges: Some phenomena can’t be directly observed or measured reliably. Access to actual behavior: Sometimes you’re studying behavior that’s private, illegal, or occurs in settings you can’t access. If you can’t get valid measurements of your key variables, your study isn’t feasible regardless of how interesting the research questions are.
Let me show you specific categories of impractical topics AI frequently suggests, and why they don’t work.
AI suggests: “A phenomenological study of the lived experiences of sexual assault survivors” Why it’s not feasible: Sexual assault survivors are a vulnerable population. Asking them to recount trauma creates psychological risk. IRBs require extensive protections including mental health screening, availability of counseling services, careful informed consent processes, and often consultation with trauma specialists. Most student dissertations can’t meet these requirements, and many IRBs won’t approve such studies even with protections because the risk-benefit ratio doesn’t favor research conducted by inexperienced researchers. AI suggests: “Interviewing undocumented immigrants about their experiences accessing healthcare” Why it’s not feasible: Undocumented immigrants are vulnerable (economically disadvantaged, at legal risk). Asking about legal status creates risks. Documentation of participation could be subpoenaed. Most students can’t provide sufficient protections to get IRB approval. AI suggests: “Studying children’s responses to discipline in schools” Why it’s not feasible: Research involving minors requires parental consent plus child assent. This creates recruitment challenges and raises questions about whether children can truly voluntarily participate in research about their experiences with authority figures. Additionally, studying discipline raises vulnerability concerns.
AI suggests: “Evaluating the effects of a new state education policy on student achievement” Why it’s not feasible: If the policy was implemented within the past 1-2 years, sufficient data doesn’t exist yet. You need baseline data from before implementation, adequate time for effects to manifest, and post-implementation data—typically 3-5 years minimum. Your dissertation can’t wait five years for data to exist. AI suggests: “Examining how the COVID-19 pandemic affected small business survival rates” Why it might not be feasible: Depending on when you’re conducting research, comprehensive business survival data might not be available yet. Government databases often lag by 1-2 years. Plus, businesses take time to fail—immediate effects versus long-term consequences are different. AI suggests: “The impact of a new drug or medical treatment on patient outcomes” Why it’s not feasible: Medical outcome data takes years to accumulate. If the treatment is very recent, the patient sample is small, and long-term outcomes aren’t yet known. You’d need to wait several years for sufficient data.
AI suggests: “Interviewing people about their racist attitudes toward different ethnic groups” Why it’s not feasible: Even if you could get IRB approval (questionable), people won’t honestly report racist attitudes in interviews. Social desirability bias makes the data invalid. You’ll get politically correct responses that don’t reflect actual attitudes, rendering the research meaningless. AI suggests: “A study of wealthy homeowners’ views on affordable housing development in their neighborhoods” Why it’s not feasible: Wealthy people know the socially acceptable answer is supporting affordable housing. Even if they oppose it, they’re unlikely to admit opposition to a researcher, especially if you’re recording interviews. You’ll get skewed data that doesn’t reflect actual views. AI suggests: “Examining illegal behavior among professionals (tax evasion, insurance fraud, etc.)” Why it’s not feasible: People won’t admit to illegal behavior in research interviews that could be subpoenaed. Your data will be either nonexistent (no participants) or invalid (dishonest responses).
AI suggests: “Observing executive decision-making in Fortune 500 boardrooms” Why it’s not feasible: You won’t get access. Executives don’t let graduate students observe confidential strategic discussions. Even if one company agreed, you couldn’t get a sufficient sample. AI suggests: “Interviewing victims immediately after experiencing trauma” Why it’s not feasible: Multiple feasibility problems: IRB won’t approve (too vulnerable, too risky), you can’t access people in crisis (hospitals, police, crisis centers won’t grant access), and sampling would be impossible (you can’t identify victims in real-time). AI suggests: “Studying classified government programs or confidential corporate data” Why it’s not feasible: Information is classified or proprietary for reasons. You won’t get access, and even if you did, you couldn’t publish the findings.
When you work with experienced dissertation advisors, they evaluate feasibility systematically before you invest time in a topic.
Real professors have submitted IRB applications for their own research and reviewed applications for others. They know: What IRBs approve readily: Low-risk surveys of adults on non-sensitive topics, secondary data analysis of de-identified datasets, interviews on professional practices with voluntary participants. What requires special protocols: Research with minors (requires parental consent), research on sensitive topics (requires debriefing and resources), research with vulnerable populations (requires extra protections). What IRBs typically reject: Research creating significant psychological risk, research with very vulnerable populations without compelling benefits, research where risks clearly outweigh benefits. They can assess your proposed topic and predict whether IRB approval is feasible, requires major modifications, or is unlikely to be granted.
Experienced researchers understand data access constraints: Public datasets: Which ones exist, what variables they include, who can access them, and what restrictions apply. Proprietary data: What companies typically share with researchers, what requires payment or partnership, what’s never accessible. Primary data collection: Realistic timelines for recruitment, typical response rates, access barriers for different populations. They can tell you whether the data you need exists and whether you can realistically access it within your dissertation timeline.
Real scholars understand measurement challenges: Social desirability effects: When people won’t report honestly and when that undermines validity versus when you can work around it. Self-report limitations: What people can accurately report about themselves versus what requires alternative measurement approaches. Behavioral observation: When direct observation is possible versus when it’s not, and what alternative indicators you might use. They can assess whether you can get valid measurements of your key variables or whether measurement challenges will undermine your research.
Experienced advisors know how long things actually take: IRB approval: Expedited review (2-4 weeks) versus full review (2-3 months) versus multiple revisions (6+ months). Recruitment: How long it takes to recruit different types of samples, realistic response rates, seasonal factors. Data collection: How long interviews take, survey completion times, observation period requirements. Analysis: How long different types of analysis take, learning curves for new methods. They can assess whether your proposed study can be completed within the 12-18 months you typically have from proposal approval to defense.
Let me show you specific examples of topics that seemed viable to students but weren’t once feasibility was assessed.
Proposed topic: “A qualitative study of how incarcerated mothers maintain relationships with their children” Student’s thinking: Important social issue, understudied population, clear problem-driven focus. Feasibility problems:
Proposed topic: “Evaluating the impact of a new teacher mentoring program implemented this year on teacher retention” Student’s thinking: Addresses important problem, has organizational access, clear outcomes. Feasibility problems:
Proposed topic: “Interviewing physicians about whether they treat patients differently based on race” Student’s thinking: Important equity issue, physicians are accessible, interviews reveal perspectives. Feasibility problems:
Proposed topic: “Observational study of how emergency department teams communicate during high-stress trauma cases” Student’s thinking: Important for patient safety, can observe real interactions, clear applied value. Feasibility problems:
Students waste months or even years developing infeasible dissertation topics. The pattern is consistent: They develop a topic (often with AI help) that sounds academically interesting. They invest significant time in literature review and proposal writing. They submit to IRB or their committee. They discover major feasibility problems that should have been identified before they invested all that effort. Now they’re starting over, demoralized and behind schedule.
Before you invest weeks or months developing a topic, get it reviewed for feasibility by someone with actual research experience. At Real Professors, we conduct feasibility reviews that assess: IRB likelihood: Will this get approved? What modifications would be needed? Are there insurmountable ethical concerns? Data accessibility: Does needed data exist? Can you access it? What’s the realistic timeline? Measurement validity: Can you get valid measures of key variables? Are there social desirability or self-report concerns? Timeline realism: Can this be completed within typical dissertation timeframes? Resource requirements: What would this cost? What skills or assistance would you need? This review takes about 30 minutes but can save you months of wasted work on topics that won’t pass IRB or committee approval. Let a real professor review your feasibility before you pitch it to your committee.
If you’re still developing your topic, we can work with you from the beginning to ensure feasibility: We help you identify topics that are simultaneously:
Students sometimes resist narrowing topics for feasibility, thinking it compromises their research vision. But feasibility isn’t a limitation—it’s what makes completion possible.
A completed dissertation on a slightly narrowed topic is infinitely better than an abandoned dissertation on an impossible topic. Your committee knows this. That’s why they emphasize feasibility.
Feasible studies actually produce better results than ambitious but impractical ones:
Learning to design feasible research is a critical skill for your post-PhD career. Whether you’re in academia or applied settings, you’ll need to conduct research within resource, time, and ethical constraints. Your dissertation is practice for that reality.
AI can help you brainstorm topics and refine research questions. But it cannot assess whether those topics are actually doable. That assessment requires:
Three Requirements for Dissertation Feasibility
Let me break down what makes a dissertation topic feasible, because students often don’t understand all the constraints they’re working within.
Ethical: IRB Approval Without Insurmountable Obstacles
Your dissertation research must get IRB approval. That means your study must: Minimize risks to participants: Physical, psychological, social, legal, or economic risks must be minimized and outweighed by potential benefits. Protect vulnerable populations: Extra protections required for children, prisoners, pregnant women, mentally disabled individuals, economically disadvantaged people, and victims of trauma or crime. Ensure informed consent: Participants must understand what they’re consenting to and be able to consent voluntarily without coercion. Maintain confidentiality: You must protect participant identity and data security. Avoid deception: Unless absolutely necessary and justified, you can’t deceive participants about the research. Some dissertation topics can meet these requirements with appropriate protocols. Others cannot, no matter how carefully designed. AI doesn’t understand this distinction.
Data-Feasible: Data Exists Now or Can Be Collected Within Your Timeline
Your dissertation needs data you can actually access within your program’s timeframe: Secondary data: If you’re using existing data, it must be available to you. Some datasets require special permissions, cost money, or restrict access to certain researchers. Primary data collection: If you’re collecting your own data, you must be able to recruit participants, access research sites, and complete collection within your timeline (typically allowing 6-12 months for defense after collection). Temporal requirements: If your research requires longitudinal data or examines effects of recent policies/programs, sufficient time must have passed for those effects to be observable. Some topics sound great but require data that doesn’t exist yet or that you can’t access. AI can’t evaluate these practical constraints.
Measurable: You Can Actually Get Valid Data on Your Variables
Even if you can collect data, it must be valid and useful: Social desirability bias: If you’re asking about controversial or sensitive topics, people might not answer honestly, making your data invalid. Self-report limitations: Some things people can’t accurately self-report (unconscious biases, certain behaviors, complex decision processes). Observation challenges: Some phenomena can’t be directly observed or measured reliably. Access to actual behavior: Sometimes you’re studying behavior that’s private, illegal, or occurs in settings you can’t access. If you can’t get valid measurements of your key variables, your study isn’t feasible regardless of how interesting the research questions are.
Topics AI Commonly Suggests That Aren’t Feasible
Let me show you specific categories of impractical topics AI frequently suggests, and why they don’t work.
Sensitive Populations That Trigger IRB Concerns
AI suggests: “A phenomenological study of the lived experiences of sexual assault survivors” Why it’s not feasible: Sexual assault survivors are a vulnerable population. Asking them to recount trauma creates psychological risk. IRBs require extensive protections including mental health screening, availability of counseling services, careful informed consent processes, and often consultation with trauma specialists. Most student dissertations can’t meet these requirements, and many IRBs won’t approve such studies even with protections because the risk-benefit ratio doesn’t favor research conducted by inexperienced researchers. AI suggests: “Interviewing undocumented immigrants about their experiences accessing healthcare” Why it’s not feasible: Undocumented immigrants are vulnerable (economically disadvantaged, at legal risk). Asking about legal status creates risks. Documentation of participation could be subpoenaed. Most students can’t provide sufficient protections to get IRB approval. AI suggests: “Studying children’s responses to discipline in schools” Why it’s not feasible: Research involving minors requires parental consent plus child assent. This creates recruitment challenges and raises questions about whether children can truly voluntarily participate in research about their experiences with authority figures. Additionally, studying discipline raises vulnerability concerns.
Recent Policy or Program Effects With Insufficient Data
AI suggests: “Evaluating the effects of a new state education policy on student achievement” Why it’s not feasible: If the policy was implemented within the past 1-2 years, sufficient data doesn’t exist yet. You need baseline data from before implementation, adequate time for effects to manifest, and post-implementation data—typically 3-5 years minimum. Your dissertation can’t wait five years for data to exist. AI suggests: “Examining how the COVID-19 pandemic affected small business survival rates” Why it might not be feasible: Depending on when you’re conducting research, comprehensive business survival data might not be available yet. Government databases often lag by 1-2 years. Plus, businesses take time to fail—immediate effects versus long-term consequences are different. AI suggests: “The impact of a new drug or medical treatment on patient outcomes” Why it’s not feasible: Medical outcome data takes years to accumulate. If the treatment is very recent, the patient sample is small, and long-term outcomes aren’t yet known. You’d need to wait several years for sufficient data.
Politically Charged Topics With Social Desirability Issues
AI suggests: “Interviewing people about their racist attitudes toward different ethnic groups” Why it’s not feasible: Even if you could get IRB approval (questionable), people won’t honestly report racist attitudes in interviews. Social desirability bias makes the data invalid. You’ll get politically correct responses that don’t reflect actual attitudes, rendering the research meaningless. AI suggests: “A study of wealthy homeowners’ views on affordable housing development in their neighborhoods” Why it’s not feasible: Wealthy people know the socially acceptable answer is supporting affordable housing. Even if they oppose it, they’re unlikely to admit opposition to a researcher, especially if you’re recording interviews. You’ll get skewed data that doesn’t reflect actual views. AI suggests: “Examining illegal behavior among professionals (tax evasion, insurance fraud, etc.)” Why it’s not feasible: People won’t admit to illegal behavior in research interviews that could be subpoenaed. Your data will be either nonexistent (no participants) or invalid (dishonest responses).
Topics Requiring Impossible Access
AI suggests: “Observing executive decision-making in Fortune 500 boardrooms” Why it’s not feasible: You won’t get access. Executives don’t let graduate students observe confidential strategic discussions. Even if one company agreed, you couldn’t get a sufficient sample. AI suggests: “Interviewing victims immediately after experiencing trauma” Why it’s not feasible: Multiple feasibility problems: IRB won’t approve (too vulnerable, too risky), you can’t access people in crisis (hospitals, police, crisis centers won’t grant access), and sampling would be impossible (you can’t identify victims in real-time). AI suggests: “Studying classified government programs or confidential corporate data” Why it’s not feasible: Information is classified or proprietary for reasons. You won’t get access, and even if you did, you couldn’t publish the findings.
How Human Experts Assess Feasibility
When you work with experienced dissertation advisors, they evaluate feasibility systematically before you invest time in a topic.
IRB Experience
Real professors have submitted IRB applications for their own research and reviewed applications for others. They know: What IRBs approve readily: Low-risk surveys of adults on non-sensitive topics, secondary data analysis of de-identified datasets, interviews on professional practices with voluntary participants. What requires special protocols: Research with minors (requires parental consent), research on sensitive topics (requires debriefing and resources), research with vulnerable populations (requires extra protections). What IRBs typically reject: Research creating significant psychological risk, research with very vulnerable populations without compelling benefits, research where risks clearly outweigh benefits. They can assess your proposed topic and predict whether IRB approval is feasible, requires major modifications, or is unlikely to be granted.
Data Access Reality
Experienced researchers understand data access constraints: Public datasets: Which ones exist, what variables they include, who can access them, and what restrictions apply. Proprietary data: What companies typically share with researchers, what requires payment or partnership, what’s never accessible. Primary data collection: Realistic timelines for recruitment, typical response rates, access barriers for different populations. They can tell you whether the data you need exists and whether you can realistically access it within your dissertation timeline.
Measurement Validity Expertise
Real scholars understand measurement challenges: Social desirability effects: When people won’t report honestly and when that undermines validity versus when you can work around it. Self-report limitations: What people can accurately report about themselves versus what requires alternative measurement approaches. Behavioral observation: When direct observation is possible versus when it’s not, and what alternative indicators you might use. They can assess whether you can get valid measurements of your key variables or whether measurement challenges will undermine your research.
Realistic Timeline Assessment
Experienced advisors know how long things actually take: IRB approval: Expedited review (2-4 weeks) versus full review (2-3 months) versus multiple revisions (6+ months). Recruitment: How long it takes to recruit different types of samples, realistic response rates, seasonal factors. Data collection: How long interviews take, survey completion times, observation period requirements. Analysis: How long different types of analysis take, learning curves for new methods. They can assess whether your proposed study can be completed within the 12-18 months you typically have from proposal approval to defense.
Real Examples of Feasibility Problems
Let me show you specific examples of topics that seemed viable to students but weren’t once feasibility was assessed.
Example 1: The Vulnerable Population Problem
Proposed topic: “A qualitative study of how incarcerated mothers maintain relationships with their children” Student’s thinking: Important social issue, understudied population, clear problem-driven focus. Feasibility problems:
- Prisoners are protected class requiring special IRB protocols
- Access to prisons requires corrections department approval (months-long process, often denied)
- Interviewing about children creates additional vulnerability
- Timeline for approvals would exceed dissertation timeframe
- Many IRBs won’t approve prison research for student dissertations
Example 2: The Data Timing Problem
Proposed topic: “Evaluating the impact of a new teacher mentoring program implemented this year on teacher retention” Student’s thinking: Addresses important problem, has organizational access, clear outcomes. Feasibility problems:
- Program just started; no retention data exists yet
- Need at least 2-3 years to measure retention effects
- Dissertation can’t wait that long
- First-year teachers don’t make retention decisions until year 2-3 anyway
Example 3: The Social Desirability Problem
Proposed topic: “Interviewing physicians about whether they treat patients differently based on race” Student’s thinking: Important equity issue, physicians are accessible, interviews reveal perspectives. Feasibility problems:
- Physicians know admitting differential treatment is career-threatening
- Social desirability makes honest responses unlikely
- Data would be invalid even if collected
- IRB might question value of research with likely invalid data
Example 4: The Access Problem
Proposed topic: “Observational study of how emergency department teams communicate during high-stress trauma cases” Student’s thinking: Important for patient safety, can observe real interactions, clear applied value. Feasibility problems:
- Hospital won’t grant access to observe actual trauma cases (patient privacy, liability concerns)
- IRB questions patient consent when they’re unconscious or in crisis
- Staff won’t behave naturally being observed during life-and-death situations
- Sampling would be impossible (when do traumas occur? how many would you need to observe?)
Don’t Waste Time on Infeasible Topics
Students waste months or even years developing infeasible dissertation topics. The pattern is consistent: They develop a topic (often with AI help) that sounds academically interesting. They invest significant time in literature review and proposal writing. They submit to IRB or their committee. They discover major feasibility problems that should have been identified before they invested all that effort. Now they’re starting over, demoralized and behind schedule.
Get Feasibility Review Before Major Investment
Before you invest weeks or months developing a topic, get it reviewed for feasibility by someone with actual research experience. At Real Professors, we conduct feasibility reviews that assess: IRB likelihood: Will this get approved? What modifications would be needed? Are there insurmountable ethical concerns? Data accessibility: Does needed data exist? Can you access it? What’s the realistic timeline? Measurement validity: Can you get valid measures of key variables? Are there social desirability or self-report concerns? Timeline realism: Can this be completed within typical dissertation timeframes? Resource requirements: What would this cost? What skills or assistance would you need? This review takes about 30 minutes but can save you months of wasted work on topics that won’t pass IRB or committee approval. Let a real professor review your feasibility before you pitch it to your committee.
Feasibility-First Topic Development
If you’re still developing your topic, we can work with you from the beginning to ensure feasibility: We help you identify topics that are simultaneously:
- Original (haven’t been done before)
- Problem-driven (address meaningful issues)
- Feasible (can actually be conducted within your constraints)
Why Feasibility Matters More Than You Think
Students sometimes resist narrowing topics for feasibility, thinking it compromises their research vision. But feasibility isn’t a limitation—it’s what makes completion possible.
Finished Beats Perfect
A completed dissertation on a slightly narrowed topic is infinitely better than an abandoned dissertation on an impossible topic. Your committee knows this. That’s why they emphasize feasibility.
Feasible Research Is Better Research
Feasible studies actually produce better results than ambitious but impractical ones:
- Data you can actually collect is more valid than data you struggle to access
- Populations you can recruit are more representative than hard-to-reach populations with 10% response rates
- Questions people will answer honestly produce more useful findings than questions that trigger social desirability bias
Career Preparation
Learning to design feasible research is a critical skill for your post-PhD career. Whether you’re in academia or applied settings, you’ll need to conduct research within resource, time, and ethical constraints. Your dissertation is practice for that reality.
The Bottom Line: Feasibility Requires Experience
AI can help you brainstorm topics and refine research questions. But it cannot assess whether those topics are actually doable. That assessment requires:
- Experience with IRB processes and requirements
- Knowledge of data sources and access realities
- Understanding of measurement validity challenges
- Realistic sense of research timelines and resource needs