A common assumption in discussions of artificial intelligence is that efficiency is inherently beneficial. If an AI system can perform a task faster, more accurately, or with less effort than a human being, many assume that delegating the task is obviously rational. This assumption overlooks a critical ethical problem. For students, junior professionals, and other developmental users, some forms of AI delegation may undermine the very capacities that education and professional experience are intended to build. The issue is not whether AI can perform a task. The issue is whether the user remains capable of understanding, evaluating, and taking responsibility for the work produced.
The growing popularity of the slogan “let AI do the doing, not the thinking” reflects this misunderstanding. Doing is not merely execution. In many educational and professional contexts, procedural labor functions as the scaffolding through which judgment, expertise, self-monitoring, and practical wisdom are developed. When those developmental activities are prematurely delegated to AI systems, users may experience improved short-term performance while simultaneously weakening the cognitive capacities required for independent reasoning.
Why Existing Approaches Fail
Much of the current discussion surrounding AI ethics focuses on familiar concerns such as bias, privacy, misinformation, or regulatory compliance. While these concerns are important, they often overlook a more personal ethical question: Can individuals misuse AI in ways that are harmful to themselves?
Current discourse frequently reduces ethics to compliance, policy, or organizational governance. This framing neglects the developmental consequences of AI dependence. It also tends to confuse successful task completion with genuine understanding. A student who submits an AI-generated essay may receive a passing grade. A junior employee who relies heavily on AI-generated analysis may produce an acceptable report. However, neither outcome demonstrates that the individual has developed the underlying competencies required for independent judgment.
Empirical research increasingly suggests that excessive reliance on generative AI can reduce critical engagement with tasks. Studies have found associations between AI reliance, cognitive offloading, diminished critical thinking, and weakened metacognitive accuracy. The concern is not that AI necessarily makes individuals less intelligent. Rather, improper AI use may shift users from the role of author, reasoner, and evaluator into the role of requester, editor, and approver. This distinction matters ethically because responsibility for judgment remains with the human user.
The Holcombe Case-Based Moral Reasoning Framework
To analyze this problem, Mark T. Holcombe’s Holcombe Case-Based Moral Reasoning Framework (HCBMR) provides a structured approach.
The Holcombe Case-Based Moral Reasoning Framework (HCBMR) is a structured approach to ethics education that develops moral judgment through systematic analysis of real-world cases rather than abstract theory alone. It trains individuals to identify morally relevant facts, competing values, and decision tradeoffs before applying ethical principles.
The central strength of HCBMR is that it begins with a real-world dilemma rather than abstract theorizing. Improper AI use presents precisely the type of practical ethical challenge that the framework was designed to evaluate.
Applying HCBMR to Improper AI Use
1. Case Study
Consider a student who uses a large language model to complete assignments she cannot independently perform. Alternatively, consider a junior employee who submits AI-generated reports that he lacks the expertise to verify.
Both cases involve successful output production. The ethical question is whether the users remain capable of exercising independent judgment regarding the work they claim as their own.
2. Relevant Facts
Several facts are morally relevant:
- AI systems can generate highly fluent
- Users may experience a subjective feeling of understanding without possessing actual
- Educational and professional development depend upon repeated engagement with procedural tasks.
- Responsibility for submitted work remains with the human These facts distinguish developmental AI use from routine automation.
3. Value Conflict Identification
- The central conflict is between competing
- On one side are efficiency, convenience, productivity, and immediate
- On the other side are autonomy, competence development, intellectual honesty, and long-term professional growth.
- The ethical challenge emerges because the short-term benefits of AI delegation may directly undermine the long-term development of rational agency.
4. Reasoned Evaluation Through Kantian Ethics
Kantian ethics provides one normative structure for evaluating this conflict.
Kant’s moral philosophy treats rationality and autonomy as intrinsically valuable characteristics of persons. Individuals possess dignity because they are capable of self-governance through reason. Consequently, actions that weaken the exercise of rational agency raise ethical concerns even when they appear beneficial in the short term.
From this perspective, AI becomes problematic when it ceases to function as a tool that extends rational agency and instead begins to replace rational agency.
A student who uses AI as a tutor, coach, critic, or simulator may strengthen her capacity for judgment. In contrast, a student who uses AI to bypass the cognitive struggle necessary for learning may weaken the very capacities education is intended to cultivate.
Similarly, a professional who uses AI to accelerate work that s/he fully understands preserves responsibility and autonomy. A professional who cannot verify AI-generated recommendations but nevertheless presents them as his/her own risks misrepresenting both his/her competence and his/her agency.
5. Justification With Restrictions
- HCBMR requires acknowledgment of tradeoffs rather than simplistic
- The argument is not that AI should be Nor is the argument that efficiency lacks value.
- The relevant distinction concerns whether AI supplements judgment or substitutes for
- When AI extends human reasoning, it can enhance
- When AI replaces human reasoning in contexts where competence is still being developed, it may erode autonomy.
The New Dunning-Kruger Effect
One of the most significant risks identified in recent research is a phenomenon that might be described as an AI-mediated form of the Dunning-Kruger effect.
The pattern is straightforward:
- The user receives fluent AI
- The output creates a feeling of
- The user lacks the competence required to verify the
- The user nevertheless relies upon
- Confidence increases faster than
The resulting condition is not simply ignorance. It is synthetic competence, the appearance of mastery without the corresponding capacity for independent evaluation.
This distinction matters because professional responsibility requires more than successful outputs. It requires understanding.
Implications for AI Ethics Education and Organizational Governance
The implications extend beyond individual users.
For ethics education, the lesson is that AI literacy must include metacognitive literacy. Students must learn not only how to use AI systems but also how to recognize the limits of their own understanding.
For organizations, the lesson is that AI deployment should not be evaluated solely through productivity metrics. Leaders should consider whether AI systems are strengthening or weakening the development of professional expertise within their workforce.
For AI ethics more broadly, the issue highlights the importance of preserving human judgment rather than merely optimizing task performance.
Conclusion
Improper AI use can constitute a form of self-harm from a Kantian perspective when three conditions are present:
- The user delegates a task necessary for developing or exercising rational
- The user cannot independently evaluate the AI-generated
- The user nevertheless claims understanding, authorship, or professional
Under these conditions, the ethical problem is not the technology itself. The ethical problem is the surrender of rational agency.
The central Kantian lesson remains straightforward: do not use AI in ways that make you less capable of being the author of your own judgment.
This analysis reflects Mark T. Holcombe’s broader work on applied ethics, moral psychology, and AI ethics within the Holcombe Ethics Framework Suite. Through the Holcombe Case-Based Moral Reasoning Framework and the Empirical Moral Reasoning Integration Model, the issue of AI dependence can be analyzed not merely as a technological concern but as a question of human development, autonomy, and moral responsibility.
References
Fernandes, D., Villa, S., Nicholls, S., Haavisto, O., Buschek, D., Schmidt, A., Kosch, T., Shen, C., & Welsch, R. (2025). AI makes you smarter, but none the wiser: The disconnect between performance and metacognition. SSRN.
Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15(1), 6.
Glean Work AI Institute. (2026). Work AI Index 2026.
Lee, H.-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025). The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems.
Mayer, A.-S., Baygi, R. M., & Buwalda, R. (2025). Generation AI: Job crafting by entry-level professionals in the age of generative AI. Business & Information Systems Engineering, 67, 595–613.
FAQ
Is using AI always unethical from a Kantian perspective?
No. Kantian ethics does not prohibit tool use. The ethical concern arises when AI replaces rather than supports rational agency.
What makes improper AI use a form of self-harm?
Improper AI use can weaken autonomy, critical thinking, and independent judgment, capacities that Kantian ethics treats as central to human dignity.
Can students use AI ethically?
Yes. AI can be used ethically as a tutor, coach, critic, or learning aid. Ethical concerns arise when students use AI to bypass the developmental work required for genuine understanding.
Why does competence matter more than performance?
Performance demonstrates output quality. Competence demonstrates the ability to understand, evaluate, defend, and reproduce that output independently.
Which Holcombe framework is most relevant to this issue?
The primary framework is the Holcombe Case-Based Moral Reasoning Framework (HCBMR), supported by the Empirical Moral Reasoning Integration Model (EMRIM), which helps integrate empirical findings about cognition and moral reasoning into ethical analysis.