An Interdisciplinary AI Curriculum System from the Perspective of New Liberal Arts

In 2025, the Ministry of Education issued the Guidelines for Artificial Intelligence Curriculum Teaching in Higher Education Institutions, which explicitly required that AI general education achieve full coverage across all disciplines and all majors and accelerate the establishment of an “AI+X” interdisciplinary curriculum landscape. This policy signal conveys a clear direction: AI literacy is evolving from a mandatory technical course for science and engineering majors into a foundational competency requirement for students across all academic disciplines.

For higher vocational colleges whose programs are predominantly in the humanities and social sciences, this trend poses a particularly profound challenge. How can an AI curriculum system that combines technical depth with industry relevance be built within a disciplinary context that lacks a foundation in science and engineering? How can AI general education avoid degenerating into superficial concept popularization or mechanically imposed technical training? How can the cultivation of technical application skills be balanced with the value-driven care and ethical reasoning tradition inherent in the humanities?

The curriculum development practice of Nanjing Institute of Tourism and Hospitality offers a valuable reference solution to the questions above.

The Starting Point: Structural Challenges Facing a Liberal Arts College

Nanjing Institute of Tourism and Hospitality is a nationally renowned higher vocational college in tourism. With a long history of education, its program portfolio focuses on tourism management, hotel management, culinary arts and nutrition, and cultural creative design, forming a distinctive cluster of liberal arts and application-oriented majors. In recent years, the penetration of AI technology into the culture and tourism industry has been accelerating and deepening—from passenger flow forecasting and dynamic scheduling in smart scenic areas, to intelligent customer service and personalized service recommendations in the hotel industry, to AI-assisted dish development and supply chain optimization in the catering sector, to AI content generation and precision advertising in cultural tourism marketing. The rapid emergence of these application scenarios is reshaping the industry’s fundamental expectations of the talent competency structure.

Changes on the industry side transmit directly to the education side. Employers’ competency requirements for culture and tourism graduates have expanded from traditional service skills and management knowledge to composite competencies such as data-driven thinking, intelligent tool application, and even AI solution design. In the face of this shift, if the college fails to systematically integrate AI education into its talent development programs, its graduates’ career competitiveness risks structural weakening.

However, advancing AI education in liberal arts colleges is not simply a matter of adding courses. A deeper analysis reveals three structural contradictions in this process that must be systematically addressed at the curriculum design level.

First, the gap between disciplinary foundations and technical thresholds. Students in liberal arts programs generally lack programming training and mathematical foundations, and their understanding of concepts such as algorithms, models, and data structures is relatively weak. If the AI course framework of computer science programs is applied directly, not only is teaching effectiveness hard to guarantee, but it can easily create severe frustration in the early stage of learning, which in turn deepens students’ alienation from and resistance to technology. Curriculum design must find an appropriate cognitive entry point—one that effectively lowers the barrier to entry and provides students with positive learning feedback early on, while avoiding the risk of reducing the teaching content to a shallow tool introduction that loses genuine capacity-building value.

Second, the mismatch between general courses and professional needs. Most of the AI general courses currently available on the market are organized around technical principles and general-purpose tools, lacking deep integration with specific industry scenarios. For students in majors such as tourism management, hotel operations, and culinary arts, such courses neither establish a direct sense of career relevance nor effectively translate into transferable industry application skills. Students may learn in class how to call a generic API, yet have no idea how to apply it to scenic area passenger flow analysis or hotel energy management—this disconnect between “learning” and “applying” is precisely the fundamental limitation of generalized course design.

Third, the rupture between technical capability and ethical literacy. The culture and tourism industry naturally involves large volumes of user privacy data and public-facing content dissemination. A single AI-generated fake travel review can mislead consumer decisions, an unauthorized use of a tourist’s personal information can trigger legal risk, and a deepfake promotional image of a scenic area can constitute commercial fraud. For culture and tourism practitioners, AI ethics governance is not a distant academic topic but a real-world issue they must confront in their daily work. Yet in most AI course systems, ethics education is often compressed into a brief overview in the final chapter, or even touched upon only through classroom discussion, falling far short of a systematic cultivation mechanism.

It was precisely based on an accurate assessment of the contradictions above that Nanjing Institute of Tourism and Hospitality decided to start from top-level design, abandon the path of simply importing general courses, and instead build an AI general education curriculum system truly suited to the disciplinary characteristics of a liberal arts college and the needs of the industry.

Top-Level Design: A Three-Dimensional Integrated Curriculum Architecture

After a systematic analysis of the college’s positioning, program structure, faculty resources, and student competency profiles, the project team designed for Nanjing Institute of Tourism and Hospitality a three-dimensional integrated AI general education curriculum architecture comprising “technology application—industry integration—ethics governance.”

The core design logic of this architecture is that the three dimensions are not three independent course tracks but mutually supportive, organically connected layers of competency development. The technology application dimension addresses the question of “whether it can be used,” building students’ basic competence in using AI tools; the industry integration dimension addresses the question of “where it is used,” anchoring technical capabilities to specific occupational scenarios and industry needs; and the ethics governance dimension addresses the questions of “whether it should be used” and “how to use it responsibly,” defining value boundaries and behavioral norms for technology application. Together, the three dimensions point toward a unified educational goal: to equip students with the comprehensive competency to understand, apply, and govern artificial intelligence technology within their own professional fields.

At the implementation level, the college can independently select and configure course modules across the three dimensions based on the training programs and teaching conditions of different program clusters, flexibly organizing teaching content. This modular organization both respects the objectively differentiated needs across programs and reserves ample flexibility for the curriculum system to dynamically iterate in response to technological development and industry change.

Technology Application Dimension: Rebuilding the AI Learning Path for Liberal Arts Students

The curriculum design of the technology application dimension must first answer a fundamental question: What is a sensible path for liberal arts students to learn AI technology?

The project team’s judgment is that the AI learning path for liberal arts students should not be a simplified, downscaled version of the science and engineering path, but rather a course logic independently constructed based on the cognitive characteristics of the liberal arts. This means the starting point of the course is not programming syntax or mathematical formulas, but the cognitive experience and competency advantages students already possess—content creation, visual expression, project planning, and communication and coordination—which are then enhanced and extended through AI tools, building an understanding of technical principles progressively through the process of “using” them.

Based on this logic, the courses in the technology application dimension unfold across four levels: “cognition building—tool mastery—scenario application—project practice.”

The cognitive enlightenment level focuses on resolving the basic cognitive question of “what AI is and what it can do,” while dispelling the common fear liberal arts students feel when facing technology. At this stage, students do not need to write any code; instead, they build agents through visual interfaces, create their own dedicated AI knowledge bases, and construct automatic question-answering systems capable of responding to questions in specific domains. When students discover that they can get an AI assistant working within half an hour, the mystery of technology quickly dissolves and is replaced by interest and confidence in further exploration. In addition, an experimental exercise that uses AI tools to convert natural language descriptions into 3D models gives students direct, hands-on experience of AI’s transformative boost to the efficiency of creative expression—this kind of instant feedback, where “what you imagine is what you get,” is precisely what sustains liberal arts students’ motivation for continuous learning.

The tool application level establishes two separate teaching paths tailored to the distinct characteristics of liberal arts and science students, a differentiation worth noting in the curriculum design.

The large language model practice course for liberal arts majors takes “using AI to create professional content” as its core thread. The course covers three progressive hands-on modules: the content creation lab guides students in using AI tools to generate structured content frameworks, such as travel guides, hotel marketing copy, and event planning proposals, enabling them to master prompt engineering methods for effective collaboration with large language models; the visual design workshop lets students produce professional-grade visual works through natural language instructions, including tourism promotional posters, dish photography style images, and cultural creative product concept illustrations, helping them appreciate AI’s potential in visual communication; and the lesson plan development system targets students oriented toward education management, training them to use AI to quickly generate complete course outlines and classroom activity plans based on teaching objectives and disciplinary characteristics. The design principle running throughout the course is to always conduct technical training within professional content scenarios familiar to the students, making AI tools an extension of professional competence rather than an extra burden detached from their majors.

The science-track path for information technology programs focuses on more engineering-oriented skill development. The industrial data governance practice requires students to process real sensor datasets and systematically master foundational yet critical data preprocessing skills such as outlier detection, missing value imputation, and data format standardization; the intelligent application deployment practicum guides students in packaging trained AI models into standalone applications, bridging the technical gap from experimental environment to production environment; and the workflow design workshop focuses on methods for building automated business processes, helping students understand how AI technology is systematically integrated into enterprise operations.

The scenario application level further extends technical capabilities into specific scenarios of daily work through the AI assistant application practice course. The intelligent office system module guides students in creating work platforms with automatic task assignment and document management functions; the creative design platform module trains students to use AI tools to complete the full visual creation process from idea conception and asset generation to proposal presentation; and the teaching management assistant module requires students to build a teaching resource management system capable of automatic courseware archiving and intelligent analysis of student learning situations. These scenario-based practicum projects enable students to clearly see the concrete points where AI technology will take effect in their future professional lives.

The project practice level opens an advanced track to students with strong willingness to learn in depth and high self-motivation. This level provides a complete learning path that starts with programming fundamentals and data processing foundations, progresses through the understanding of core AI mathematical principles, visual analytics and automated modeling training, and enterprise-grade programming framework practice, and culminates in independently building mobile AI applications and developing industrial-grade AI solutions. The intent behind this level is to ensure the curriculum system’s vertical extensibility, giving students who discover a strong interest in AI technology during the general education stage the space and ladder to go deeper. The starting point can be low, but the ceiling for growth should not be artificially limited.

Industry Integration Dimension: Reorganizing AI Teaching Content Anchored in Professional Scenarios

The industry integration dimension is the most distinctive and original design component of this curriculum system. It directly responds to a question that has been repeatedly discussed in the field of AI general education yet never satisfactorily answered: how can general courses truly achieve deep integration with professional education rather than stopping at superficial case embellishment?

The solution proposed by the project team is to abandon the traditional organization of “explaining technology first, then citing industry cases,” and instead take typical industry business scenarios as the teaching starting point, organizing and allocating AI technical content in reverse. In other words, the first step of curriculum design is not to map out the AI technology knowledge graph, but to deeply analyze the core business processes, key pain points, and intelligent transformation needs of the industries corresponding to each program cluster, and then precisely match the AI technology modules that can effectively respond to these needs. This reverse design method of “from scenario to technology” ensures that every knowledge point has a clear professional affiliation and an explicit career orientation.

The industry AI core technology modules organize teaching around the high-frequency business scenarios of each program cluster. The tourism management track focuses on scenarios such as passenger flow forecasting, tourist behavior analysis, and tourism public opinion monitoring in smart scenic areas, guiding students to understand the application logic of natural language processing and time series forecasting in tourism operations; the hotel management track centers on building intelligent customer service dialogue systems and designing personalized service recommendation engines, enabling students to master the basic working principles of conversational AI and recommendation algorithms; the culinary arts track focuses on AI application scenarios in assisted dish development, nutritional ratio analysis, and intelligent management of ingredient supply chains; and the cultural creativity track explores the technical possibilities of generative AI in cultural creative product design, digital cultural tourism content creation, and immersive experience development. Each scenario is paired with a complete teaching loop of “business logic analysis—technical principle understanding—hands-on tool training—solution output,” allowing students to construct their AI capabilities while solving real problems in their professional fields.

The industry data governance module pays particular attention to the uniqueness of the data ecosystem in the culture and tourism industry. Compared with industrial manufacturing or finance, the culture and tourism industry has more dispersed data sources and more diverse data types—transaction data from online travel platforms, consumption records from offline stores, user review texts on social media, and environmental and foot-traffic information collected by IoT devices at scenic areas. The effective integration and governance of these multi-source heterogeneous data itself constitutes a professional capability challenge of considerable complexity. The course systematically guides students to master the basic methodology of multi-source data fusion, standardized processes for data quality assessment and cleaning, and the specific application norms of privacy protection technologies in scenarios involving the processing of tourist personal information. The cultivation of these skills serves not only the upstream data preparation stage of AI applications but is also directly related to students’ core competence in future industry digital roles.

As the output terminal of the curriculum system’s capability development, the industry AI solution design module shoulders the important function of testing and integrating all prior learning outcomes. In this stage, students must independently complete the entire workflow from requirements research and solution design to prototype development, targeting a real business pain point or intelligent transformation need in the industry corresponding to their major. For example, tourism management students may need to design an integrated smart guided tour and emergency evacuation solution for a specific scenic area; hotel management students may need to develop a prototype of an intelligent guest room energy management system based on occupancy data analysis; and culinary arts students may need to build an AI-driven smart menu scheduling and inventory-linked management solution for restaurants. Through this stage of practical training, students undergo a role transformation from “learners of AI knowledge” to “designers of industry AI solutions”—which is precisely the core value that distinguishes interdisciplinary AI education from traditional technical education.

Ethics Governance Dimension: Bringing Value Judgment into the Core Framework of Competency Development

In the design of many AI curriculum systems, ethics education has long occupied an awkward position—everyone acknowledges its importance, yet few curriculum systems truly grant it cultivation weight and teaching hours equal to those of technical education. In most cases, AI ethics is scheduled into the last one or two class sessions and hurriedly covered in an “overview-style” manner, lacking both a systematic knowledge framework and supporting practical training. Students may be able to recite a few ethical principles on exams, yet when faced with real ethical dilemmas, they often lack the ability to make well-founded judgments and respond effectively.

This project made an explicit decision in its curriculum architecture design: to establish ethics governance as an independent curriculum dimension with weight equal to technology application, rather than an appendage or supplement to technical courses. Behind this decision lies a deeper educational philosophy. Students in humanities and social sciences programs, through their disciplinary training, already possess sensitivity to value issues and critical reflection on social impact. The establishment of an AI ethics governance course is not merely about filling the ethical gap in technical education, but about leveraging the inherent strengths of liberal arts education—making humanistic care an internal constraining force on technology application rather than an external passive restriction. In this sense, liberal arts colleges not only need to carry out AI ethics education but also possess a unique disciplinary foundation for delivering high-quality AI ethics education.

The ethics foundation course builds a systematic cognitive framework for AI ethics. Course content covers the identification and analysis of AI abuse scenarios, including the fraud risks of deepfake technology in tourism marketing and social communication, the distorting effects of AI-generated fake reviews and fake travel guides on consumer decisions, and the issues of information cocoons and price discrimination implicit in recommendation algorithms. Building on case analysis, the course further guides students to understand the basic methods of technical fairness assessment and the ethical principle framework for responsibility attribution—when an AI tourism recommendation system offers differentiated pricing to users with different spending capacities, should responsibility rest with the algorithm designer, the platform operator, or the regulator? Open-ended ethical questions of this kind are precisely the high-quality teaching material that cultivates students’ capacity for deep critical thinking.

The governance technology course translates ethical cognition into operable technical capabilities, bridging the practical gap between “knowing what should be done” and “knowing how to do it.” The course requires students to master the application methods of data flow analysis tools and systematically trace the complete data lifecycle of tourist personal information from front-end collection, back-end storage, and algorithm invocation to result output; to learn the fundamental technical means of data privacy protection, including understanding technical concepts such as data masking, differential privacy, and federated learning, and using simple tools; and to understand the nature of the AI “black box” problem and its potential harm in high-risk decision-making scenarios, as well as to use mainstream model decision explanation tools such as LIME and SHAP to improve the transparency and auditability of AI systems. The cultivation of these technical capabilities enables students to conduct basic technical assessment and risk prediction of AI system compliance in their future professional practice, rather than stopping at perceptual cognition at the value level.

The industry compliance practice course requires students to turn the ethical cognition and technical capabilities built in the first two modules into compliance solutions and normative documents for real industry scenarios. Specific practical tasks include: writing a compact technical compliance manual for specific business scenarios (such as hotel customer data management, AI recommendation algorithms on travel platforms, and copyright ownership of AI-generated marketing content) against the backdrop of the culture and tourism industry; and designing AI application compliance solutions aligned with industry characteristics, with reference to regulatory requirements such as the Personal Information Protection Law, the Data Security Law, and the Interim Measures for the Administration of Generative Artificial Intelligence Services. These practical tasks ground ethics education in verifiable, assessable, and deliverable professional outcomes rather than abstract discussions of principles.

Design Methodology: Several Conceptual Choices Worth Noting

Looking back on the design process of the entire curriculum system, several layers of conceptual choices deserve further discussion. They apply not only to the case of Nanjing Institute of Tourism and Hospitality but may also offer methodological reference value for the AI education development of similar institutions.

On the question of “who adapts to whom.” In advancing AI general education, a common practice is to require students to adapt to the existing technical course framework—making up for mathematics foundations, learning programming syntax, and following the same learning route as science and engineering students. The implicit assumption behind this practice is that there is only one learning path for AI, and the only difference lies in how fast or slow one walks it. The design logic of this project is different. We believe that the curriculum system should proactively adapt to students’ cognitive characteristics and competency advantages, rather than requiring students to cram their feet into ill-fitting shoes to fit the curriculum. Liberal arts students possess unique talents in language expression, creative ideation, project planning, and ethical reasoning. Curriculum design should take these strengths as its starting point and fulcrum, using AI tools to enhance and extend them, naturally bringing in an understanding of technical principles through the process of “using” them. This is not a compromise or downgrade of technical education, but a respect for and development of different learning paths.

On the relationship between “general” and “professional.” The term “general education” can easily create the impression that general courses are shallow, introductory teaching unrelated to any specific major. This understanding is especially harmful in the field of AI education. When general courses are completely decoupled from students’ professional studies, they become an extra module on the schedule that needs to be “dealt with,” and can hardly generate genuine educational value. The design strategy of this project is to embed the “professional” core within the “general” framework. The curriculum system maintains the universality and systematicity of general education in its overall architecture, while deeply connecting with the industry scenarios and career needs of different majors at the level of specific teaching content. What students learn is not abstract technical knowledge divorced from context, but specialized AI application capabilities that can be directly transferred to their future professional practice. The value of general education lies precisely in its ability to provide a shared competency foundation for students of all majors while reserving differentiated content interfaces for each professional direction.

On the distinction between the levels of “ability” and “literacy.” Although technical application ability and ethics governance literacy both belong to the educational goals of the curriculum system, the two differ importantly in nature. Technical application ability is instrumental; it addresses questions of efficiency and effectiveness—whether students can use AI tools to complete a specific task. Ethics governance literacy, by contrast, is value-based; it addresses questions of direction and boundary—whether students can judge under what conditions an AI application is appropriate and under what circumstances it should be restricted or prohibited. In curriculum design, these two levels of educational goals should not be conflated, nor should one be subordinated to the other. Technical capability determines what students “can do,” while ethical literacy determines what students “should do” and “should not do.” Only the synchronized cultivation of both can produce truly responsible AI practitioners. The project’s decision to establish ethics governance as an independent dimension at the same level as technology application, rather than an appendage of the latter, is a practical expression of this educational philosophy.

On the sustainability of the curriculum system. The iteration speed of AI technology far exceeds the update cycle of educational courses, a realistic constraint that all AI education projects must confront. The tools and frameworks taught today may well be outdated or replaced within two or three years. Therefore, curriculum design must strike a balance between “stability” and “flexibility.” The modular architecture of this project was designed precisely for this purpose: the overall framework and educational goals of the three dimensions are relatively stable—they reflect the underlying logic of AI education and will not become invalid with the replacement of specific technical tools—while the specific course modules and teaching content under each dimension are replaceable and updateable, able to adjust flexibly as technology develops and industry needs change. A good curriculum system is not a static product but a dynamic framework with the capacity for self-iteration. This design principle provides structural assurance for the long-term sustainable operation of the courses.

The Significance of the Project: From a Case Study to a Paradigm

The value of this curriculum development practice at Nanjing Institute of Tourism and Hospitality lies not only in solving a specific teaching need for one institution, but more importantly in providing a systematic response to a universal educational issue: under the contemporary context of New Liberal Arts development, how can humanities and social sciences higher vocational colleges build AI talent development programs suited to their own disciplinary endowments?

This question matters because it relates to a broader proposition of educational equity. If AI literacy education is effectively implemented only in science and engineering institutions while students in humanities and social sciences colleges are excluded, then as AI technology deepens its penetration across industries, this imbalance in the allocation of educational resources will gradually translate into a capability gap and inequality of opportunity in the labor market. From this perspective, designing appropriate AI curriculum systems for liberal arts colleges is not merely the technical work of course development, but a systematic undertaking concerning educational equity and social justice.

The answer provided by this project can be summarized in three core propositions.

First, AI education for liberal arts students requires an independently designed learning path. This path should start from the cognitive advantages students already possess, be driven by industry application scenarios, and replace one-shot technical indoctrination with progressive competency construction.

Second, the true value of AI general courses lies not in their “generality,” but in being “general yet capable of specialization”—achieving deep alignment with each professional direction within the general framework, allowing AI capabilities to genuinely take root in the soil of students’ majors and become an organic component of their career competitiveness.

Third, ethics governance should not be regarded as an ethical patch on technical education, but should be established as a core competency dimension of AI education. Particularly for humanities and social sciences institutions, integrating humanistic care and value judgment into AI education is both an expression of educational responsibility and a leveraging of disciplinary strengths.

When students of Nanjing Institute of Tourism and Hospitality enter the workforce, they will be not only competent tourism managers, hotel operators, culinary professionals, or cultural creative designers, but also a new generation of culture and tourism talent equipped with AI technology comprehension, industry application capability, and ethical judgment. This is precisely the educational outcome pursued by the interdisciplinary AI curriculum system, and the proper embodiment of innovation in talent development models from the perspective of New Liberal Arts.