In the era of profound global economic transformation and rapidly evolving technological revolution, new quality productive forces have become the core engine driving high-quality socioeconomic development. The cultivation and development of these forces urgently require professionals with innovative thinking and digital literacy. Integrating the concept of new quality productive forces with intelligent digital technologies, while comprehensively improving the quality of higher-education talent cultivation and learning efficiency, has become a crucial theoretical and practical subject in current educational reform. As a faculty member deeply engaged in electric vehicle design education, I have witnessed firsthand the transformative impact of artificial intelligence on course design, instructional methodology, and student assessment.
Artificial intelligence has become increasingly prevalent in higher education, presenting both opportunities and challenges. In terms of optimizing instructional content, AI can deeply analyze existing teaching materials, assist educators in planning structured course content, and create instructional presentations that improve teaching efficiency. Furthermore, AI can generate customized learning pathways and resources according to learners’ progress and preferences, encouraging autonomous and deep learning. In addition, through analyzing student learning data, AI offers accurate evaluation of learning outcomes, which provides a robust foundation for teaching improvements.
Electric vehicle design courses involve multiple disciplines, including mechanics, automation, electrical engineering, computer science, and materials science. These courses typically encounter problems such as strong interdisciplinary crossover, slow updating of teaching content, monotonous pedagogical methods, and one-sided assessments. To address these challenges, I have attempted to deeply integrate digital resource development with smart course platform construction, while practicing a SPOC (Small Private Online Course) blended teaching model, in order to establish a new paradigm of smart course construction for electric vehicle design under the perspective of new quality productive forces.
Overall Architecture of the New Paradigm
The new paradigm for smart course construction in electric vehicle design, oriented toward nurturing new quality talents, reveals four fundamental dimensions: digitalization, intelligence, precision, and personalization. Through digital methods, I have enriched learning resources and achieved precise content delivery and personalized learning path development. Intelligent technologies enable an interactive teaching framework involving teacher, student, and machine. This framework builds a diversified assessment system so as to comprehensively enhance students’ creative capabilities and digital literacy, thereby forming innovative and persistent talents equipped with the technical attributes of the digital era.
The architecture can be expressed mathematically as a system of integrated components:
$$
S_{\text{smart-course}} = f(D_{\text{digital-resources}}, T_{\text{interactive-design}}, P_{\text{precise-delivery}}, A_{\text{assess-system}})
$$
where \(D_{\text{digital-resources}}\) represents the rich, structured resources aided by knowledge graphs, \(T_{\text{interactive-design}}\) represents the teacher-student-machine interaction using AI, \(P_{\text{precise-delivery}}\) denotes the adaptive learning path recommendation and content delivery, and \(A_{\text{assess-system}}\) represents the multifaceted evaluation mechanism.
Digital Resource Integration Based on Knowledge Graphs and Modular Design
In accordance with graduation requirements and course objectives, I have divided course content into fundamental theory modules, design modules, and frontier technology modules. By using large language models, I deeply explored and analyzed existing teaching resources, extracted keyword information, constructed thematic models, and identified relations among key information and knowledge points. The extracted knowledge points and structured information are subsequently used as nodes in the course knowledge graph. Logical connections among knowledge points are clarified to create a tree-like or network-like structure. The knowledge graph ensures that every point is represented independently while the entire system reflects its systematic and cutting-edge nature.
For the course “Fundamentals of Electric Vehicle Design”, the modular structure can be summarized in the next table:
| Module | Core Content | Representative Knowledge Points | Relation Type |
|---|---|---|---|
| Basic Theory | Vehicle dynamics, energy sources, driving cycles | Rolling resistance, aerodynamic drag, grade resistance, acceleration force | Prerequisite |
| Battery System | Lithium-ion cells, battery packs, BMS | Capacity, C-rate, state of charge, thermal management | Dependent on Basic Theory |
| Motor System | Permanent magnet motor, induction motor, switched reluctance | Torque-speed characteristics, efficiency map, motor parameters | Dependent on Basic Theory |
| Electric Control | Motor controllers, energy management, regenerative braking | PID control, state observers, control strategies | Integrates battery and motor |
| Vehicle Integration | Powertrain matching, chassis, thermal systems | Transmission ratio, final drive ratio, vehicle controller | Top-level integration |
| Frontier Topics | Solid-state battery, blade battery, hairlpin motor, smart energy management | Solid electrolyte, cell-to-pack, high fill factor windings | Extended learning |
Construction of an AI-enhanced Online Learning Platform
Based on the knowledge graph, I distilled a matrix of core knowledge points and recorded micro-lecture videos for each essential element. These videos were uploaded to the online education platform, so students could preview and review the course content. With AI technology, I built diverse question banks, enabling pre-class and post-class assessment of students’ mastery of essential knowledge nodes. Large language models empowered project-based teaching by generating project topics tailored to students’ diverse learning interests, promoting all-round development.
The digital resources include lecture slides, knowledge maps, core knowledge videos, frontier technology optional videos, after-class readings, exercise banks, and project files. The platform enables me to conduct course management and teaching actions: assigning mandatory and optional tasks, initiating group discussions, guiding students to make personalized design choices, and running periodic online quizzes. Learning analytics are recorded continuously, and I can inspect students’ interests, participation patterns, and progress. This real-time data allows rapid adjustments to instructional content.
Interactive Teacher–Student–Machine Design Based on SPOC
Using an improved BOPPPS model (Bridge-in, Objective, Pre-assessment, Participatory Learning, Post-assessment, Summary), I divided the blended teaching process for the electric vehicle design course into three phases: introduction, interactive learning, and post-learning detection. The design relies heavily on AI-assisted data analytics to refine guidance and personalize learning pathways.
In the introduction phase, my role is to supervise the learning process. I generate digital resources, publish pre-class learning materials, and answer students’ questions on the platform. Students independently explore the knowledge graph, complete required preparation activities, and optionally engage with frontier content. In the interactive phase, I lead the discussion, deliver the lecture based on the syllabus, provoke questions, guide students’ participation, and help synthesize the key structure of knowledge graph. Students are active participants: they respond to random in-class questions, work in groups on project topics, and respond to in-class surveys. In the final phase, I monitor progress by designing quizzes, homework assignments, and project task booklets. Students complete online quizzes within a limited time, submit homework, and write group project reports.
This three-phase model can be algebraically formulated as a cyclical process:
$$
L_{t+1} = g\left(L_t, \sum_{i=1}^{n} \alpha_i K_i + \sum_{j=1}^{m} \beta_j I_j + \sum_{k=1}^{p} \gamma_k P_k \right)
$$
where \(L_t\) indicates the learning state in cycle \(t\), \(K_i\) denotes knowledge graph nodes, \(I_j\) denotes AI-driven interactions and feedback, \(P_k\) denotes personalized project tasks, and \(\alpha,\beta,\gamma\) are weights computed from analytical data about student behavior, interests, and performance. In practical implementation, this iterative model ensures timely intervention when the personalized learning curve deviates from the expected trajectory.
Personalized Project Topic Selection
One of the most significant aspects of my course is personalized project selection under the guidance of AI-driven analysis. For example, based on students’ learning records in the platform, the AI recognizes high-interest topics such as solid-state batteries or advanced electric motor control. The system then generates a matrix of topic recommendations that suit students’ preferences while keeping curriculum requirements.
The next table demonstrates the actual topic choices and their corresponding driven areas:
| Direction | Project Topic Examples | AI Analysis Feature |
|---|---|---|
| Battery | Battery pack lightweight structure design, thermal management, battery equalization strategy, state-of-health prediction | Tracks student attention to cell chemistry and thermal safety |
| Electric Motor | Permanent magnet synchronous motor optimization, induction motor rotor resistance, reluctance torque optimization, drive control | Identifies student preference for electromagnetic analysis tools |
| Electric Control | Vehicle controller design, energy recovery, DC-DC converter efficiency, energy management strategy | Detects prior programming and control project competencies |
| Powertrain Integration | HEV powertrain parameters, pure EV powertrain parameters, range extender architecture, fuel cell powertrain | Highlights system-level integration learning patterns |
By offering such diversified topics, the course stimulates creativity and encourages in-depth exploration of design problems. Students are not forced into a single uniform assignment; instead, they choose or negotiate with me according to their interests. Because AI can predict which topic areas may generate more engagement for a particular student cohort, I constantly update the suggested list; after the course ends, the final reports demonstrate higher originality and technical depth than conventional assignments.
Digital Teaching Resource Construction in Practice
To implement the above paradigm, I used “Fundamentals of Electric Vehicle Design” as the pilot course. The initial step was careful alignment with the professional training plan. After defining the course aims, I used large language models to process the syllabus, digital textbooks, previous exams, and engineering standards. The models identified core concepts and segmented the content into four major teaching modules: battery, motor, electronic control, and full vehicle system. Those modules were then integrated into a networked structure with explicit dependencies. Each module includes basic concept presentations, working-principle animation displays, and structural analysis videos for representative components.
Meanwhile, through the online education platform, I structured a complete collection of resources in hierarchical fashion. Resources are associated with knowledge points in the graph; every micro-video, quiz question, reading, and project file is tagged with the corresponding knowledge point IDs. This tagging process enables two important functions: first, automatic recommendation of remedial resources when a student fails a section quiz; second, course-level analytics that show me which knowledge points are hardest to master.
Educational Data Analytics
Learning analytics can be visualized by a set of engagement metrics. Let \(C_{n}\) be the total number of student clicks in the online platform for a given knowledge node \(n\). Define \(M_{n}\) as the number of students who accessed the relevant material, and \(Q_{n}\) as the average quiz score associated with node \(n\). Then the engagement–performance index \(E_n\) can be expressed as:
$$
E_n = \frac{C_n}{M_n} \times \log\left(\frac{100}{100-Q_n + 1}\right)
$$
High values of \(E_n\) indicate that students attempted many accesses but still struggled with the knowledge point, meaning that the material should be simplified or supplemented. In my course, I routinely calculate such indices to refine teaching resources and identify students requiring additional support.
The table below illustrates representative engagement data observed in the course:
| Learning Resource Category | Number of Students Accessing | Average Clicks per Student | Quiz Accuracy (%) | Resource Type |
|---|---|---|---|---|
| Battery knowledge graph video | 82 out of 90 | 7.6 | 81% | Required |
| Motor efficiency map video | 75 out of 90 | 6.1 | 79% | Required |
| Solid-state battery frontier video | 86 out of 90 | 4.2 | 86% | Optional |
| Regenerative braking case study | 79 out of 90 | 3.8 | 84% | Optional |
| Energy management strategy quiz preparation | 70 out of 90 | 9.3 | 68% | Required |
This information clarifies not only the popularity of optional modules but also where the required modules are struggling in terms of quiz accuracy. For example, the energy management strategy section had a lower average quiz accuracy and a high number of clicks per student, indicating that additional videos or explanatory materials are needed.
Blended Learning Process Based on SPOC
The SPOC blended approach gives students a stronger sense of accountability. Every lesson module contains a sequence of online and offline events. Before each face-to-face session, students are expected to watch two or more micro-lecture videos and complete several pre-class tasks. During the class, I do not simply repeat the content. Instead, I use interactive tools to gauge misconceptions and deepen comprehension through discussions, small problem-solving workshops, and case-based reasoning. After class, students master independent scheduling, complete assignments, and participate in discussion forums.
For the course “Fundamentals of Electric Vehicle Design”, I used the following timeline, which represents one complete teaching cycle for a selected topic.
| Phase | Time Frame | Student Activity | Instructor Activity | AI Tool Assistance |
|---|---|---|---|---|
| Online introduction | Before class (30-45 min) | Watch knowledge graph videos; complete pre-quiz; answer open questions | Publish tasks on platform; review pre-class analytics | Auto-grading, identification of students with low preparation |
| Classroom engagement | 2×45 minutes | Participate in concept mapping; solve numerical design cases; debate trade-offs | Explain difficult points; moderate team discussions; provide real-world examples | Real-time polls, instant feedback collection, recommendation of questions based on weak points |
| Post-class consolidation | After class (1-2 hours) | Complete online assignment; attempt chapter assessment; refine project report | Assess assignments, post annotations; update material based on performance | Plagiarism checking, automatic categorization of errors |
This design explicitly connects the digital resources with classroom pedagogy. The initial online assessment can be treated as a pre-test \(P_{pre}\). The classroom activities produce a set of interactive response signals \(R\), and the post-class assessment yields \(P_{post}\). A straightforward effectiveness evaluation is given by the normalized learning gain:
$$
G = \frac{P_{post} – P_{pre}}{100 – P_{pre}} \times 100\%
$$
In my observed class the average pre-class quiz score was 62.2%, and the post-class assessment improved to 84.5%, yielding a learning gain of about 59.3%. Without controlling for all variables, this suggests that the AI-empowered SPOC design produced substantial advantages.
BOPPPS Model Modification and AI Integration
The BOPPPS model provides a strong instructional skeleton. However, AI integration allows me to adapt every BOPPPS stage based on real-time data. I have introduced an “adaptive bridge-in” where the AI reads recent science news and suggests thematic questions relevant to electric vehicle design. The objective stage is explicitly anchored to the course knowledge graph; I show students the associated concept map and illustrate where they currently stand. Pre-assessment tasks are generated dynamically based on previous item difficulty and student’s mastery level.
During the participatory learning stage, my teaching assistants and I facilitate project-oriented activities. For example, when teaching energy management strategies, the AI presents a set of simulated driving cycles and asks students to decide the appropriate power split between a battery and a range extender. Student teams can run small computational simulations and compare results to a benchmark. The AI monitors the process, flags students who are stuck, and offers hints. These actions promote a rich teacher-student-machine interaction: the teacher supervises the environment, the machine gives instructional scaffolds, and the student takes active cognitive responsibility.
For post-assessment, the system uses automatic question generation to adapt the difficulty of quizzes. Students who answer correctly receive more challenging questions, while those who struggle see more foundational questions. The final summary phase is visualized as a knowledge graph overlay: each student sees green, yellow, or red nodes representing their level of mastery. I spend the last five minutes of the session summarizing the connections between concepts and making a personalized study plan.
Multi-dimensional Process Assessment
Traditional final exams are insufficient to evaluate creativity and digital literacy. In my proposed paradigm, process assessment is designed to encourage continuous participation throughout the entire semester. The weights are: online tasks (required plus optional) 5%; after-class homework 5%; classroom participation 5%; chapter tests 10%; personalized project report 25%; final closed-book examination 50%. Both the online components and the face-to-face components make up half of the course grade. Each dimension is supported by AI analytics with clear rubrics.
The personalized project report is the most demanding element. It requires students to choose a design topic, carry out a literature review, perform calculations, develop system architecture, and discuss possible innovations. AI recommends a starting point, but the student must make design decisions. In my assessment rubric, I assign points to innovative thinking, technical depth, use of digital tools, quality of written communication, and teamwork. The AI platform allows peer evaluation and self-evaluation, which enrich the assessment perspectives.
The table below illustrates the complete assessment structure for my course:
| Assessment Component | Percentage | Method | Learning Objective Addressed |
|---|---|---|---|
| Online preparation tasks | 5% | Auto-graded quizzes; completion of optional videos | Autonomous learning, self-discipline |
| Homework assignments | 5% | Problem solving; CAD and simulation models | Ability to apply engineering principles |
| Classroom participation | 5% | In-class polling; group discussion records | Communication and collaboration |
| Chapter tests | 10% | Online timed tests, adaptive difficulty | Mastery of key knowledge nodes |
| Personalized project report | 25% | Written report; oral defense; peer review | Innovation, design thinking, digital literacy |
| Final exam | 50% | Closed-book, problem-based | Comprehensive understanding and analytical reasoning |
A significant advantage of this system is that it provides continuous feedback. The AI dashboard at mid-semester indicates which students are at risk due to frequent missed online tasks or low quiz performance. I schedule one-to-one office hours for those students and help them devise better learning plans. Additionally, because I can see which item types cause the most difficulty (e.g., deriving torque equations, understanding state-of-charge dynamics, drawing hybrid powertrain architectures), I can address these topics in later lectures or create targeted supplementary resources.
Implementation Outcomes and Discussion
I collected one full semester of student behavior data from the smart teaching platform. By analyzing these data, I identified individualized learning hotspots and common misconceptions. The table shows the top hotspot frontier optional modules for “Fundamentals of Electric Vehicle Design”.
| Frontier Optional Module | Click Rate (%) | Number of Questions in Forum |
|---|---|---|
| Solid-state battery | 95.3 | 53 |
| Blade battery | 94.5 | 41 |
| Hairpin winding motor | 91.2 | 87 |
| Fuel cell technologies | 88.6 | 36 |
| Regenerative braking control | 87.9 | 48 |
These high click rates demonstrate that students are highly eager to explore cutting-edge technologies. I therefore enriched course content with discussions and design tasks around such frontier topics. In the next semester, I added a set of small group debates on the pros and cons of solid-state batteries compared with conventional lithium-ion cells. The students’ project topics related to solid-state battery thermal management increased by 40%.
In terms of misconceptions, the following table lists the most troublesome points and corresponding rates based on chapter tests.
| Key Point in Chapter Test | Correct Answer Rate (%) | Difficulty Classification |
|---|---|---|
| Battery-related calculations | 80.0 | Medium-high |
| Motor parameter calculations | 82.0 | Medium-high |
| HEV architecture drawing | 83.0 | Medium |
| Powertrain system matching | 83.0 | High |
| Energy management strategy | 85.0 | High |
The battery-related calculations reveal a correct rate of 80%, lower than expected. After diagnosing the root cause, I realized that the required calculation was conventionally presented as a single long video. I subsequently decomposed it into smaller micro-units, each focusing on a specific equation, and embedded interactive practice within the videos. For example, a five-minute video explaining the Shepherd model was followed by a short check question. In the second administration, the average score on the battery calculation questions increased from 80% to 88%.
Another common misconception was the accurate drawing of hybrid electric vehicle architectures. About 17% of students could not correctly draw powertrain configurations, identify the positions of clutches, or label power-flow paths. In response, I added more dynamic examples in classroom discussions and provided custom-tailored learning materials through the AI assistant. The assistant could answer questions such as “What is the difference between series and parallel hybrid?” with visual animations. In the later exam, only 8% of students still mislabeled the architecture, showing that the intervention was effective.
Role of AI in Empowering Teachers and Students
The new teaching model requires teachers to shift from information transmitters to learning architects and facilitators. In my personal experience, AI has not replaced the teacher. Instead, it amplifies my ability to understand each learner’s needs. Large language models assist me in generating new assessment items, developing concept explanations, and proposing design scenarios. The AI also reduces the administrative burden of grading routine assignments, freeing more time for high-level feedback on project designs.
For students, AI-powered learning platforms provide immediate assistance. When students are stuck on a calculation in the middle of the night, they can consult an AI tutoring agent embedded in the course platform. This agent is trained with the course syllabus, knowledge graph, and lecture videos; it provides explanations consistent with the course terminology and can connect the question to prior knowledge modules. By using the agent, students gain a sense of autonomy and confidence. The platform logs their queries, which I use to update the frequently asked questions section and improve the course design.
One important observation is that AI-empowered learning requires careful digital education. Students may overly rely on the AI answer rather than struggling with the problem themselves. To avoid this, I require students to show step-by-step reasoning in online assignments, and the AI tool is programmed to offer hints rather than full solutions. In project-based learning, the AI generates proposals but not final design reports; students must develop calculation tables, simulation models, and engineering drawings from first principles. This balance improves both efficiency and intellectual rigor.
From a systems perspective, the course can be modeled as a control loop. Suppose \(x_t\) is the vector of student knowledge states at week \(t\), \(u_t\) is the vector of instructional interventions chosen from the set of digital resources, and \(y_t\) is the observed assessment output. Then:
$$
x_{t+1} = A x_t + B u_t + w_t
$$
$$
y_t = C x_t + v_t
$$
with \(A\) representing the prior knowledge transition, \(B\) the efficacy of interventions, \(C\) the assessment mapping, and \(w_t, v_t\) process and measurement noises. AI analytics support the estimation of \(x_t\) with Kalman-filter-like methods. The teacher embodies the controller, updating \(u_t\) based on observed deviations from the desired mastery level. This control-oriented interpretation gives a quantitative foundation to explain why timely, personalized interventions are essential in smart courses.
Recommendations for Continuous Improvement
Continuous improvement is inherent in any curriculum innovation. The following feedback loop is implemented after every semester:
- Collect platform data on online engagement, quiz scores, forum activity, and project quality.
- Identify knowledge points with low mastery and high access rates.
- Analyze student feedback and course evaluation questionnaires.
- Revise the knowledge graph and teaching resources, especially by adding targeted micro-videos or interactive simulations.
- Update the AI assistant’s training data and question bank.
- Retrain large language models with new course materials and hold faculty workshops on emerging educational technology.
Through one teaching cycle, I noticed that optional video viewing did not automatically translate into course performance. Students who viewed frontier videos performed slightly better on project reports but not significantly on quizzes. This observation encourages me to design optional activities that explicitly ask students to connect frontier topics to foundational course content. For example, after watching a video on solid-state batteries, students must write a half-page memo explaining how solid electrolytes affect battery pack design and thermal management. This assignment not only reinforces their conceptual understanding but also improves digital writing and synthesis abilities.
Another improvement involves teacher development. As an educator, I need to become conversant with the capabilities and limits of AI. Faculty should experiment with prompt engineering to create high-quality educational content, analyze outputs critically, and design lessons that integrate AI without undermining critical thinking. Institutional support is necessary, including a reliable online infrastructure, AI usage training, and technical assistance. In my case, the course-level collaboration among teaching faculty and an educational technology expert was crucial for building the knowledge graph and fine-tuning for our course goals.
Let me also discuss the scalability of the paradigm. The method of constructing knowledge graphs from course materials can be transferred to other engineering disciplines. The main formula for the interconnection of knowledge points is general:
$$
y = \sigma\left(\sum_{i=1}^{n} w_i \phi_i(x) + b\right)
$$
where \(x\) represents the prerequisite knowledge, \(\phi_i\) are learning activities, \(w_i\) are the relevance weights extracted from AI analytics, and \(y\) is the intended learning outcome. For automotive design courses, \(x\) can be mathematics and physics foundations, \(\phi_i\) can be simulation tasks, and \(y\) can be performance on a design review. The same template works for mechanical engineering, electrical engineering, and computer science courses.
The table below compares traditional course construction with the proposed smart course paradigm in terms of several pedagogical dimensions.
| Dimension | Traditional Course | New Smart Course Paradigm |
|---|---|---|
| Content organization | Linear chapters | Networked knowledge graph |
| Resource updating | Slow, static textbooks | Rapid, AI-assisted frontier modules |
| Learning path | Uniform for all students | Personalized adapted to learning status |
| Teacher-student interaction | Mostly one-way instruction | Three-way teacher-student-machine interaction |
| Assessment | Summative final exam | Multi-dimensional process assessment |
| Data usage | Limited to grade records | Full learning analytics and rapid feedback |
| Student competence | Knowledge memory | Innovation and digital literacy |
Clearly, the transformation to a smart education model is not merely cosmetic. It changes the fundamental relation between teaching and learning. From the perspective of new quality productive forces, universities are responsible for cultivating new quality talents; hence, they must develop new paradigms correspondingly. The curriculum should not only include digital tools but also develop an integrated ecosystem that supports lifelong learning and innovation.
Future Perspectives and Research Directions
Because electric vehicle technology is evolving rapidly, course content must remain responsive to industrial changes. Knowledge graphs allow me to add new nodes and edges as soon as a technology matures. For example, when the production of battery electric vehicles with 800-volt architecture began to gain market share, I could add voltage platform nodes, SiC inverter nodes, and charging infrastructure nodes. The graph can be updated locally without restructuring the entire course. This agility is crucial in the field of electric vehicles.
Artificial intelligence holds promise for automatic knowledge graph extraction from scientific literature and engineering standards. In the future, I plan to develop machine-learning models that compare curriculum knowledge graphs with job market skill demands. Such models could recommend which emerging topics should be inserted into the course to ensure graduates are industry-ready. Additionally, generative AI can generate realistic design challenges based on recent automotive failure cases or competitive benchmark data, making project-based learning richer and more relevant.
Another promising direction is campus-wide integration. The smart course platform can be expanded across a program, allowing instructors to map all courses into one multi-layer knowledge graph. Students will be able to see how an electric vehicle design course depends on a prior control theory course and a materials science course. This transparency boosts motivation and enables global curriculum optimization. For instance, if a student is weak in calculus, the platform might recommend review materials from an earlier mathematics course and relate them to current electric vehicle dynamics equations.
Educational research must investigate the effects of smart courses on student outcomes. Controlled experiments can compare the learning gains from standard SPOC and AI-enhanced SPOC. The variables to study include student engagement, retention, self-regulation, and creativity. Sophisticated natural language processing can analyze the novelty and technical correctness of student project reports. In addition, qualitative studies can explore how students perceive the AI assistant, the university trust policies, and the ethics of using learner data. Addressing these topics will further refine the design framework I have presented here.
As an engineering educator, I also recognize the importance of designing courses for sustainability. Smart technologies rely on servers and data centers, which have carbon footprints. A truly new quality productive force should include considerations of social responsibility and environmental sustainability. I therefore introduced small assignments in which students calculate the energy consumption of cloud-based simulation tasks, compare battery material footprints, and discuss the life-cycle environmental impact of electric vehicles. This integration mirrors the larger goal of engineering education: producing not only skilled designers but also responsible global citizens.
The role of the university is not to train students solely for existing job positions but to prepare them to create future positions. In the context of electric vehicle design, new quality productive forces mean more than making cars; they imply designing intelligent, connected, and sustainable mobility systems. The curriculum of the future must teach students to adapt to technological change, to learn how to learn, and to combine digital skills with deep domain expertise. The smart course paradigm I have established provides an operational route to achieve such ambitions.
I adopted the following mathematical formula to frame the intended student competence:
$$
C_{\mathrm{student}} = w_1 K_{\mathrm{core}} + w_2 D_{\mathrm{digital}} + w_3 I_{\mathrm{innovation}} + w_4 S_{\mathrm{sustainability}}
$$
Here \(K_{\mathrm{core}}\) is assessed by exams and quizzes, \(D_{\mathrm{digital}}\) by use of simulation tools and online learning, \(I_{\mathrm{innovation}}\) by project reports and design proposals, and \(S_{\mathrm{sustainability}}\) by assignments reflecting environmental consciousness. The weights \(w_i\) are calibrated to align with graduate attributes defined by the professional accreditation body. This simple additive model helps stakeholders understand the multiple objectives of curriculum transformation.
Conclusion
In this article, I presented a new paradigm for developing smart courses in electric vehicle design under the broader perspective of new quality productive forces. By building a digital resource system based on knowledge graphs and modularization, establishing a blended learning platform empowered by AI and SPOC, implementing an interactive teacher-student-machine instructional design, and adopting a multi-dimensional process assessment system, I have achieved notable progress in enhancing student engagement and learning efficacy.
The digital construction supplies precision: each educational resource is linked to an explicit node in the knowledge graph, allowing flexible paths for learners at different levels. The SPOC model makes teaching adaptable: online preparation, face-to-face interaction, and after-class consolidation form a coherent loop with data-driven interventions. AI supports us in interpreting large-volume learning data and generating timely feedback. As a result, students receive individualized recommendations that respect their pace, interests, and capacities. They learn not only the technical principles of electric vehicles but also how to harness digital tools to solve complex engineering problems.
Since new quality productive forces require new quality talent, higher education must redesign curriculum content, delivery modes, and evaluation schemes. The course “Fundamentals of Electric Vehicle Design” serves as a successful reference: it illustrates that the path to innovative education is feasible and powerful. My students’ abilities to conduct design innovation and their digital literacy have visibly improved through this smart-course experience. Future research will further refine the model, explore scalable institutional implementations, and incorporate ever-evolving AI breakthroughs. I believe this paradigm can be widely applied in higher education, contributing to the cultivation of competitive, ethical, and creative engineers prepared for the digital and sustainable economy.