top of page

The Tractor Is Already in the Field: Why Doctoral Education Must Stop Banning AI and Start Reimagining Itself



At the turn of the twentieth century, the mechanical tractor arrived on American farms. Farmers who embraced it multiplied what they could produce. Those who clung to hand-plowing in defense of "proper farming" fell behind, not because they were less diligent, but because they were measuring the wrong thing. We are at an identical crossroads in doctoral education today.


Generative AI, ChatGPT, Claude, Gemini, and their descendants are not a passing disruption. It is a categorical shift in the cognitive infrastructure available to every knowledge worker, including every doctoral student on the planet right now. And the academy's dominant response so far has been: ban the tractor.


I've spent over two decades in enterprise software engineering and nearly as long in graduate-level instruction, designing and teaching doctoral courses in Generative AI, Advanced Algorithms, and Software Engineering. What I'm witnessing in online doctoral programs today represents one of the clearest cases of institutional failure I have seen in my career. We are not defending intellectual rigor. We are defending proxies for it that AI has exposed as hollow.

The Discussion Board Is Already Dead

Let me describe something that is happening at a significant scale right now, and that almost no one in higher education is willing to say out loud: AI-to-AI interaction.


A doctoral student, often a working professional, squeezing in coursework late on a Tuesday night, prompts a language model to generate a 300-word discussion board response citing two peer-reviewed sources. Their classmates do the same. Their classmates' "substantive responses" to each other's posts are also AI-generated. The instructor reviews ten sections of sixteen students each, applies a standardized rubric, and returns automated-sounding feedback. In some cases, the feedback itself was drafted with AI assistance.


The learning outcome of this entire exchange is approximately zero. The intellectual labor of every human party is approximately zero. And we are conferring doctoral credentials for it.


"This is not a consequence of AI. It is a consequence of an assessment design that had no authentic intellectual purpose from the outset and that AI has simply exposed."

The solution is not to ban AI from discussion boards. The solution is to abandon the discussion board, or to radically reconceptualize it as live oral defenses of written positions, structured argumentation with real-time challenge, and collaborative problem documentation. These formats are harder to assess at scale. That is precisely why they have not been adopted. And precisely why they represent the legitimate challenge of doctoral education.

In commercial aviation, the copilot is a fully qualified aviator who manages substantial flight operations, assists in complex decisions, and operates sophisticated automated systems, but does not replace the pilot's judgment, situational awareness, or ultimate command authority. The copilot extends the pilot's capacity; it does not substitute for it.


Generative AI, positioned correctly, operates analogously. It can survey literature, synthesize sources, draft preliminary texts, generate code scaffolding, and check logical consistency. What it cannot do and what doctoral education should be evaluating is the student's ability to direct, evaluate, debug, and build upon what the AI produces in service of a genuine intellectual purpose.


Consider the trajectory of tools in professional knowledge work:


The developer who uses Copilot brilliantly, maintaining architectural clarity, debugging intelligently, ensuring code meets quality and security standards, is more skilled than the developer who writes every line by hand but can't see the system-level picture. The doctoral student who can direct an AI copilot toward a genuine knowledge contribution while understanding its limitations is operating at a higher cognitive level, not a lower one.

In 1969, Herbert Simon published The Sciences of the Artificial, formalizing a distinction that doctoral education in computing, information systems, and engineering has never fully absorbed: natural science asks how things are; design science asks how things ought to be. Design science is not a lesser form of inquiry; it is a rigorous, systematic discipline with its own epistemological foundations.


The formal frameworks followed: March & Smith (1995) distinguished design artifacts from natural science findings; Hevner, March, Park & Ram (2004) articulated seven guidelines for rigorous Design Science Research (DSR) in MIS Quarterly; Peffers et al. (2007) gave us the DSRM process model; Gregor & Hevner (2013) defined the knowledge contribution types.


Here is the problem. When a doctoral student has designed a novel federated learning architecture for secure health data sharing, some dissertation committees still require them to conduct semi-structured interviews with hospital administrators to justify the research's "qualitative grounding." The interviews add nothing to the evaluation of the architecture's novelty, efficiency, or impact. They exist to satisfy the methodological comfort of faculty trained in a different paradigm.


"Forcing natural-science methods onto design-oriented research is not a guarantee of rigor. It is a distortion methodological ornamentation in service of faculty comfort, not intellectual value."

The most impactful doctoral contributions in AI today are design contributions to the transformer architecture, reinforcement learning from human feedback, federated learning systems, domain-specific AI applications in healthcare, and law. These are artifacts, not hypotheses. DSR is the framework built to evaluate them rigorously. And DSR's iterative design cycle maps naturally onto AI-augmented development, making it the paradigm most coherently aligned with doctoral innovation in our moment.

The most persistent objection to AI integration in doctoral education is the sentiment that it makes everything "too easy" so that students should struggle through tasks by hand to develop discipline and rigor. This sentiment reflects a failure to distinguish between productive struggle and performative struggle.


The farmer's manual plowing did not produce better crops than tractor-plowing it produced more exhausted farmers who plowed less land. The intellectual value of struggle is not uniformly distributed across all types of challenge. Struggling with a genuinely hard conceptual problem develops intellectual depth. Struggling to produce a syntactically correct literature review in a domain where you have genuine expertise is an exercise in performance, not cognition.


Doctoral education should be hard in the ways that matter:


None of these challenges is mitigated by AI. If anything, AI raises the bar on them by compressing the mechanical components of research, making the intellectual components more visible and more central to evaluation.



Institutional inertia is real, but it is not an excuse. Based on two decades of practice and years of graduate instruction, here are six concrete steps doctoral programs can take immediately:

Every doctoral student has a smartphone in their pocket with access to AI that can produce a passable discussion board post in under thirty seconds. This cannot be prohibited through honor codes, Turnitin AI-detection modules with documented false-positive rates, or proctoring software that turns learning environments into surveillance environments.


The question is not whether doctoral students will use AI. They will. The question is whether doctoral programs will train them to use it well, rigorously, transparently, critically, and in the service of genuine contributions to human knowledge or whether programs will spend their institutional energy catching students who have learned to hide their AI use rather than reflect on it.


The farmer who embraced the tractor did not become a lesser farmer. They became capable of farming land that hand-plowing would never have reached. That is exactly what well-designed AI integration in doctoral education should produce: scholars who can operate at a scale, depth, and ambition that the pre-AI era could not have imagined.


The tractor is in the field. The question is who will learn to drive it and who will teach them how.


Read the full paper on Academia



Dr. Moody Amaboke an accomplished IT leader and educator with over two decades of experience leading digital transformation across regulated industries.

As the Founder of the Global Data Science Institute and an adjunct professor, he brings together academic depth and real-world application - translating complex technological shifts into practical, impactful solutions. His work sits at the intersection of innovation, data science, and industry transformation, with a strong focus on building future-ready systems and talent.



Comments

bottom of page