The Higher Education Curriculum Crisis: Why “AI Degrees” are Already Obsolete

by Prof. (Dr.) Rangamini Werawatta

Across global higher education and throughout Sri Lanka’s state and non-state university systems, a frantic race is underway. Driven by market demand and eager to capture student enrollments, tertiary institutions are rapidly launching brand-new specialised AI degree programmes: BSc in Artificial Intelligence (AI), BSc in Data Science and AI, and BSc in Applied Machine Learning.

To university marketing departments and anxious parents, these programmes sound like the ultimate hedge against future unemployment. To corporate leaders and technology strategists, however, they reveal a profound structural delusion.

Rushing to package artificial intelligence into a static, traditional four-year degree structure is a fundamental mistake. In an era where technological paradigms shift every six months, the traditional academic engine is simply too slow to deliver relevant training. Unless we fundamentally rethink how university curricula are designed, named, and accredited, today’s shiny new “AI degrees” will produce graduates whose qualifications are obsolete before they even walk across the graduation stage.

The Institutional Velocity Paradox

To understand why specialised AI degrees are failing, one must look at the mechanical reality of higher education governance.

In the Sri Lankan higher education system—governed by Quality Assurance Council (QAC) guidelines, Institutional Review Frameworks, and strict Faculty Board approvals—a university curriculum is typically revised only once every five years. From initial curriculum design, stakeholder consultations, and senate approvals to final UGC endorsement, launching or updating a degree programme is a multi-year administrative endeavor.

Now, contrast that institutional velocity with the rate of artificial intelligence development:

This structural disconnect creates the Institutional Velocity Paradox:

1. Year 1: A student enters a newly approved BSc in AI and learns foundational syntax, specific Python libraries, and basic model architecture based on a syllabus written two years prior.
2. Year 2 & 3: Autonomous AI coding agents, multi-modal systems, and advanced reasoning models render the manual coding techniques and specific frameworks taught in Year 1 completely obsolete.
3. Year 4: The student graduates with deep, evaluated expertise in toolchains that the industry abandoned three years ago, while the university remains locked in its mandatory five-year review cycle.

Teaching students specific AI tools, libraries, or prompt frameworks within a rigid multi-year syllabus is like teaching navigation using a map of a city whose streets are redrawn every Monday.

The Rebranding Trap: Syntax in a New Wrapper

A rigorous audit of many newly minted “AI curricula” reveals another uncomfortable truth: most are simply traditional Computer Science or Software Engineering degrees with a superficial facelift.

Core modules in procedural programming, traditional database management, and object-oriented design are retained, supplemented by elective modules like “Introduction to Machine Learning” or “Neural Networks.” They continue to evaluate students on the same legacy metrics—whether code compiles, whether mathematical formulas are memorised, or whether basic syntax is executed without errors.

As addressed in our analysis of industry hiring shifts, writing routine boilerplate syntax and implementing manual logic are no longer high-value market capabilities. Generative models and autonomous developer agents now execute these tasks instantly. When universities grade students on code generation, they are evaluating capabilities that AI has already commoditised.

The K-SAM Imperative: Shifting from Tool Mastery to Cognitive Resilience

If specific AI toolchains cannot be taught within a four-year window without becoming obsolete, what should universities actually teach?

The answer lies in grounding higher education firmly within the K-SAM Paradigm (Knowledge, Skills, Attitudes/Values, Mindset/Paradigm)—the holistic competency framework embedded within Outcome-Based Education (OBE) and the Sri Lanka Qualifications Framework (SLQF).

Instead of chasing fleeting software tools, higher education must divide its curriculum into two distinct layers:

1. The Immutable Core (Knowledge & Skills)

Universities must stop trying to be trade schools that teach temporary software frameworks. The core university syllabus must double down on foundational, timeless principles that do not spoil with age: discrete mathematics, computational complexity, logic, systems architecture, data structures, and probability theory. A student who deeply understands first-principles system design can master any new AI paradigm in a weekend.

2. The Behavioral & Perceptual Layer (Attitudes & Mindset)

The greatest gap facing modern tech employers is not a lack of technical information access, but a deficiency in the Attitudes and Mindset Paradigm. Higher education must explicitly train and evaluate:

• Output Auditing & Critical Verification: Teaching students how to systematically interrogate, stress-test, and audit AI-generated code and system outputs for edge-case vulnerabilities.
• Architectural Orchestration: Transitioning student assignments from writing raw syntax to orchestrating multi-agent systems, managing cloud resilience, and aligning technical solutions with enterprise strategy.
• Cognitive Agility & Unlearning: Developing a psychological mindset that views technological disruption not as a threat, but as an operational baseline.

Re-Engineering Academic Governance: A Blueprint for Sri Lanka

To prevent our university system from producing waves of unemployable graduates holding impressive-sounding AI degrees, university leadership, Deans, Senate Academic Boards, and the University Grants Commission (UGC) must adopt three immediate structural reforms:

1. Decouple Core Degrees from Fast-Fading Tooling and Align with SLQF/ACM Standards

Universities must stop creating reactionary, narrow degree titles whenever a new technology trends in the market. Degree nomenclature must strictly align with internationally established standards—specifically the ACM/IEEE Computing Curricula 2020 (CC2020) framework—while adhering to the qualification naming protocols enforced by Page 7 of the SLQF Guidelines.

Under the ACM/IEEE CC2020 framework, computing higher education is built around 6 core recognized disciplines:

1. Computer Engineering (CE) – Hardware-software integration, embedded systems, and digital architecture.
2. Computer Science (CS) – Theoretical foundations, algorithms, and computational science.
3. Cybersecurity (CSEC) – System security, threat mitigation, digital forensics, and privacy governance.
4. Information Systems (IS) – Strategic alignment of technology, data, and workflows within business organizations.
5. Information Technology (IT) – System administration, cloud infrastructure, network deployment, and operational technology.
6. Software Engineering (SE) – Systematic design, software architecture, verification, and enterprise lifecycle management.

In parallel, Page 7 of the SLQF Guidelines explicitly governs degree naming conventions:

• Qualification Type & Level: A 4-year Honours degree is designated as Bachelor of Science Honours or Bachelor of Computing Honours at SLQF Level 6.
• Main Discipline / Field of Study: The qualification title must be constructed using the broad discipline (e.g., Bachelor of Science Honours in Computer Science).
• Specializations / Majors: SLQF stipulates that a specific area of concentration must be indicated as a specialisation within brackets or appended to the core qualification (e.g., Bachelor of Science Honours in Computer Science in Artificial Intelligence or Specialising in Artificial Intelligence), provided that a minimum required proportion of credits in that specific area are fulfilled.

The Academic Reform:

Fabricating non-standard degree titles like “BSc in Generative AI” or “BSc in Prompt Engineering” violates both SLQF naming rules and global accreditation frameworks (such as the Washington or Seoul Accords). Universities must anchor core degree titles directly in the 6 core ACM/IEEE computing disciplines, incorporating specialised areas like Artificial Intelligence, Multi-Agent Systems, or Autonomous Robotics strictly as SLQF-compliant Specialisations or dynamic elective concentrations.

2. Implement the “Core + Micro-Credential” Modular Model (70/30 Rule)

To bypass the rigid five-year curriculum revision lock, universities should establish an adaptable 70/30 Curriculum Model:

• 70% Core Foundation: Fixed academic theory, discrete mathematics, logic, and system architecture governed by standard academic quality assurance and SLQF credit allocations.
• 30% Dynamic Elective Stack: Modular, industry-co-designed micro-credentials updated annually (or semester-by-semester) without triggering a full five-year degree re-accreditation cycle.

3. Transition from Coding Assignments to Architectural Apprenticeships

Replace traditional “write a programme that does X” assignments with Architectural Audits. Grade students on their ability to write precise system specifications, direct autonomous AI agents to build the prototype, and conduct rigorous security, performance, edge-case, and ethical audits on the generated output.

Conclusion

Artificial intelligence has not made higher education irrelevant; it has made obsolete educational models intolerable.

A university degree should not be a four-year vocational course in writing syntax or prompting specific software models. The true purpose of higher education in the post-AI era is to build resilient, highly adaptable minds capable of critical thought, ethical judgment, and complex system orchestration.

It is time for Sri Lankan higher education leadership to look beyond marketing trends. We must stop trying to teach temporary AI tools through static multi-year curricula and start building the foundational K-SAM capacity that will empower our graduates to lead the AI revolution rather than be replaced by it.