Beyond Tutoring: Can AI Truly Coach the Next Generation of Engineers?

by

Author: Denis Avetisyan

A new review examines the potential of AI chatbots to move beyond simple information delivery and provide meaningful coaching in engineering education.

Research suggests AI excels at technical problem-solving but requires human collaboration to address nuanced skills like ethical reasoning and interpersonal dynamics.

Traditional engineering education, reliant on apprenticeship models fostering tacit skill, faces disruption concurrent with the rise of generative AI as an informal coaching tool. This convergence prompts a re-examination of the limits of computation and the nature of wisdom, questions addressed in ‘Can AI Chatbots Provide Coaching in Engineering? Beyond Information Processing Toward Mastery’. Our mixed-methods study reveals that while participants readily accept AI for technical problem-solving, skepticism remains regarding its capacity for nuanced judgment requiring emotional and ethical reasoning. Can a ‘multiplex coaching’ framework, integrating human mentorship with AI scalability, effectively bridge this gap and redefine the future of engineering education?

The Erosion of Experiential Expertise

Engineering education once heavily relied on the traditional apprenticeship model, a system where aspiring engineers learned not just what to do, but how to do it through prolonged, hands-on experience alongside seasoned professionals. This immersive approach fostered what is known as ‘tacit knowledge’ – the unwritten, intuitive understanding of materials, processes, and problem-solving that extends beyond textbook learning. Apprentices didn’t simply memorize formulas or procedures; they absorbed the subtle cues, practical shortcuts, and nuanced judgments developed over years of practice. This tacit knowledge was crucial for navigating the unpredictable realities of engineering projects, allowing them to adapt to unforeseen challenges and make informed decisions even when faced with incomplete information. The system prioritized learning through doing, observation, and mentorship, creating a robust foundation of practical skill and instinctive understanding that is increasingly difficult to replicate in modern, classroom-centric educational environments.

The longstanding model of engineering education, once characterized by immersive, practical experience, is demonstrably diminishing. This erosion isn’t simply a shift in pedagogical approach; it creates a critical deficit in the development of nuanced judgment essential for tackling ‘divergent problems’ – those lacking a single, clear solution. Contemporary curricula, while robust in conveying explicit knowledge and refining skills for ‘convergent problems’ with defined answers, frequently overlook the vital capacity to navigate ambiguity, assess incomplete information, and formulate innovative strategies when faced with ill-defined challenges. The consequence is a potential disconnect between academic proficiency and the complex, real-world demands requiring engineers to synthesize knowledge, exercise critical thinking, and make informed decisions under conditions of uncertainty.

Contemporary engineering education frequently emphasizes the acquisition of explicit knowledge – established facts, formulas, and procedures – and focuses on solving convergent problems, those with clearly defined solutions. While proficiency in these areas remains vital, this approach often inadvertently diminishes the development of crucial skills needed to address the ambiguity inherent in real-world challenges. Complex engineering endeavors rarely present themselves with neat, single answers; instead, they demand the ability to navigate uncertainty, assess incomplete information, and formulate solutions amidst conflicting constraints. Consequently, a curriculum heavily weighted towards readily quantifiable knowledge may leave graduates less prepared to tackle divergent problems – those with multiple plausible solutions, requiring innovative thinking and sound judgment in the face of incomplete data.

Scaling Guidance with Intelligent Systems

AI Coaching represents a significant expansion of Coaching (Engineering) principles by leveraging automation to reach a larger student population. Traditional coaching models, often limited by resource constraints and individual coach capacity, can be effectively scaled through AI-driven systems. This scalability is achieved by automating key components of the coaching process, such as goal setting, action planning, and progress monitoring, while maintaining a focus on individualized support. The core benefit lies in providing personalized guidance to a broader range of students who might not otherwise have access to such resources, thereby improving student outcomes and overall institutional support capabilities.

Intentionality, as it applies to AI coaching, refers to the ability of a student to effectively convert desired outcomes into a series of planned and executed steps. This process is not simply goal-setting, but a dynamic translation of objectives into concrete actions. Frameworks such as ‘Conversations for Action’ provide a structured methodology for facilitating this translation, guiding students through iterative cycles of planning, execution, and reflection. These frameworks enable AI coaching systems to move beyond providing information and toward actively supporting students in the process of enacting their goals, focusing on the ‘how’ as much as the ‘what’ of achievement.

AI coaching leverages generative AI technologies to deliver scalable guidance; however, successful implementation necessitates a focus on AI literacy. Recent studies indicate a generally positive perception of utility among both students and faculty, with average scores of 3.83-3.84 on a 5-point scale. Despite this perceived benefit, significant disparities exist in risk assessment – a Cohen’s d of 1.34 (p < 0.003) demonstrates that faculty members express substantially higher concerns regarding potential risks associated with AI coaching compared to students, highlighting the need for targeted training and awareness initiatives.

Analysis of perceptions regarding AI coaching reveals a statistically significant difference in risk assessment between student and faculty groups. A Cohen’s d of 1.34, accompanied by a p-value of less than 0.003, demonstrates a large effect size indicating substantially higher risk concerns among faculty compared to students. This suggests that while both groups perceive some level of risk associated with AI coaching, faculty members exhibit considerably greater apprehension, potentially related to concerns about academic integrity, data privacy, or the efficacy of AI-driven guidance relative to traditional methods.

A Multiplex Approach: Optimizing Human and Artificial Expertise

The Multiplex Coaching Model utilizes a hybrid approach to learning by assigning problem types to the most appropriate solver. ‘Convergent Problems’ – those with a single, definable correct answer – are efficiently addressed through Artificial Intelligence. This allows for automated assessment, personalized practice, and scalable support for foundational skills. By offloading these tasks to AI, human coaches are then freed to focus on more complex challenges, specifically those requiring subjective judgment, creative solutions, and nuanced understanding that current AI systems cannot provide. This division of labor aims to maximize both efficiency and the development of higher-order thinking skills.

Human mentorship is essential when learners encounter ‘Divergent Problems’ – those lacking a single correct solution and requiring exploration of multiple possibilities. These problems necessitate the development of critical thinking skills, including analysis, evaluation, and synthesis of information, which are best facilitated through guided discussion and personalized feedback from a human mentor. Furthermore, divergent problems often involve complex ethical considerations; mentors provide crucial guidance in ‘Ethical Discernment’, helping learners navigate ambiguous situations, identify potential biases, and make responsible decisions that align with established principles and values. The subjective and nuanced nature of these challenges makes AI-driven solutions less effective than the interpretive capabilities of a human mentor.

The Multiplex Coaching Model intentionally incorporates periods of ‘Productive Struggle’ – defined as cognitively demanding effort that leads to learning – by strategically delaying complete solutions. Rather than immediately providing answers, the model utilizes AI and human mentors to deliver targeted, just-in-time support and feedback. This approach allows learners to actively grapple with challenges, strengthening neural pathways and improving problem-solving skills. The timing of these interventions is crucial; support is offered when a learner is demonstrably stuck, preventing frustration, but not before they have engaged in meaningful effort. This balance optimizes learning by maximizing cognitive engagement and minimizing dependence on external assistance.

The Five Shifts Framework structures the Multiplex Coaching Model around five key areas of development: from knowledge acquisition to skill application; from task completion to problem solving; from extrinsic to intrinsic motivation; from dependence on authority to self-directed learning; and from cognitive understanding to embodied experience. This framework posits that effective learning requires a progression through these shifts, with reflection and the integration of learned concepts into practical application – embodiment – being crucial for sustained growth. The framework is not intended as a linear progression, but rather as interconnected dimensions to be considered when designing learning experiences and providing targeted support.

Cultivating Future-Ready Engineers: A Symbiotic System

The escalating complexity of modern engineering challenges demands a shift in educational approaches, and integrating artificial intelligence coaching with established human mentorship offers a promising solution. This blended learning model doesn’t aim to replace experienced guidance, but rather to augment it by providing engineers with personalized, on-demand support for tackling ill-defined problems. AI can facilitate iterative exploration, offering feedback on proposed solutions and prompting consideration of alternative perspectives, while human mentors contribute crucial contextual understanding, ethical considerations, and nuanced judgment. By repeatedly engaging with complex scenarios under this dual guidance, engineers develop not just technical proficiency, but also the critical thinking, adaptability, and problem-framing skills essential for navigating ambiguous and evolving challenges – ultimately preparing them to innovate responsibly in a rapidly changing world.

Recent advancements explore the integration of Motivational Interviewing (MI) principles into artificial intelligence coaching systems to bolster learner agency and intrinsic motivation. This approach moves beyond simply delivering information; the AI is designed to employ techniques such as reflective listening, affirming statements, and open-ended questioning – mirroring the core tenets of human MI practice. By prompting self-exploration and highlighting the learner’s own reasoning for change, the AI fosters a sense of autonomy and ownership over the learning process. Studies suggest that this can lead to increased engagement, improved problem-solving skills, and a stronger commitment to lifelong learning, as individuals become adept at identifying their goals and navigating challenges independently, guided by an AI that prioritizes their internal drive.

The integration of AI coaching into engineering education isn’t simply about skill acquisition; it’s deliberately designed to cultivate a nuanced comprehension of artificial intelligence itself. By routinely interacting with AI systems that offer guidance and feedback, future engineers gain firsthand insight into both the remarkable capabilities and inherent constraints of these technologies. This experiential learning moves beyond theoretical understanding, fostering critical evaluation of AI-driven solutions and promoting a proactive approach to identifying potential biases or limitations. Consequently, engineers are better prepared to innovate responsibly, developing and deploying AI systems that are not only technically advanced but also ethically sound and aligned with societal needs – a crucial attribute in a rapidly evolving technological landscape.

The evolving landscape of technology demands more than just technical skill from future engineers; it necessitates a holistic development encompassing ethical reasoning and a commitment to continuous learning. Contemporary engineering education is shifting towards cultivating professionals equipped to navigate not only the ‘how’ of innovation, but also the ‘why’ – understanding the societal impact of their creations and acting responsibly. This approach emphasizes adaptability, fostering the ability to acquire new knowledge and refine existing skills throughout a career, crucial in a field defined by rapid advancements. The ultimate aim is to produce engineers who are not simply problem-solvers, but thoughtful innovators capable of shaping a sustainable and equitable future through technology.

The exploration of AI’s capacity to coach engineering students reveals a fundamental principle: scalable systems rely on clarity of purpose, not simply computational power. As Vinton Cerf observed, “What scales are clear ideas, not server power.” This resonates deeply with the study’s findings; while AI excels at convergent problem-solving – the technical aspects of engineering – its limitations in handling divergent thinking, ethical reasoning, and the nuances of tacit knowledge demonstrate that true mastery requires a holistic approach. The proposed ‘multiplex coaching’ model acknowledges this, framing AI as one component within a larger ecosystem of support, much like interconnected systems where each element’s function impacts the whole.

The Road Ahead

The pursuit of artificial coaching in engineering, as explored within this work, reveals a familiar pattern: readily automating the convergent, while the divergent stubbornly resists. The capacity of current AI to assist with technical problem-solving is demonstrable, yet it highlights a more fundamental question. Engineering is not merely applied mathematics; it is a deeply human endeavor, reliant on contextual awareness, ethical considerations, and the navigation of complex interpersonal dynamics. To assume these can be readily codified – or even adequately simulated – is to mistake the map for the territory.

Future research should not focus solely on refining AI’s ability to mimic human responses, but on understanding the inherent limitations of such an approach. The concept of ‘multiplex coaching’ – a synergistic blend of artificial and human mentorship – offers a pragmatic path forward, but demands careful consideration of where each modality excels. Simply layering AI atop existing structures will not suffice; a fundamental re-evaluation of engineering pedagogy, recognizing the inseparability of technical skill and human judgment, is essential.

Ultimately, the true test will not be whether AI can teach engineering, but whether it can help cultivate engineers capable of adapting to a world defined by increasing complexity and ethical ambiguity. The elegance of any system, artificial or otherwise, lies not in its ability to solve existing problems, but in its capacity to anticipate – and gracefully accommodate – those yet to emerge.

Original article: https://arxiv.org/pdf/2601.03693.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/

See also:

2026-01-09 01:59