1. Introduction
Artificial intelligence has evolved from a specialized research field into a general-purpose technology that is increasingly embedded in software development, information systems, business processes, education, health care, finance, manufacturing, and public administration. Recent advances in generative AI have enabled machines to produce text, computer programs, images, audio, video, and other digital content from natural-language instructions. For the information technology workforce, AI represents both an object of development and a tool for performing professional work. Software developers now use AI assistants to generate source code, explain unfamiliar programs, write tests, identify defects, create documentation, and support system design. Network and security specialists use machine learning to detect anomalies, classify threats, automate alerts, and analyze large volumes of operational data. Database and system administrators use intelligent tools for query optimization, capacity forecasting, incident diagnosis, and configuration recommendations. These developments have led to concerns that AI may reduce the demand for programmers and other IT workers. However, jobs consist of multiple tasks with different technical, social, organizational, and legal characteristics. The International Labour Organization argues that the most probable effect of generative AI is the transformation of job content rather than the complete replacement of most occupations [1]. Human intervention remains necessary for problem definition, requirement analysis, architectural decisions, validation, accountability, communication, and the management of exceptional situations. The consequences of AI adoption may differ substantially between advanced and developing economies. Developing countries often have younger populations and expanding digital service sectors, but they may also face limited computing infrastructure, expensive Internet access, shortages of qualified instructors, weak links between education and industry, and restricted access to high-quality local-language datasets. The ability to benefit from AI therefore depends not only on access to models but also on connectivity, computing capacity, data availability, workforce skills, institutional capability, and appropriate governance [2].
This paper examines major AI trends and analyzes their effects on two groups: professionally trained IT engineers and graduates of college-level or short-cycle tertiary IT programs. The distinction is important because these groups generally enter the labor market with different levels of theoretical preparation, technical depth, and responsibility.
2. Major contemporary artificial intelligence trends
2.1. Generative and multimodal artificial intelligence
Generative AI is currently one of the most influential technological trends. Large language models can generate and transform natural language, source code, structured data, and technical documentation. Multimodal models extend these capabilities by processing combinations of text, images, audio, video, and sensor data.
In information technology work, multimodal systems can analyze screenshots, software interfaces, diagrams, source code, log files, and written requirements within the same workflow. This capability supports activities such as interface prototyping, system documentation, user assistance, and technical troubleshooting. However, generative models may produce incorrect, insecure, outdated, or fabricated results. Their outputs are based on statistical inference rather than guaranteed factual or logical correctness. Consequently, AI-generated code and technical recommendations must be reviewed, tested, and evaluated by qualified personnel.
2.2. AI-assisted software engineering
AI is being incorporated across the software development life cycle. Contemporary tools can assist with requirement clarification, source-code generation, code completion, refactoring, test creation, debugging, documentation, vulnerability identification, and deployment scripting. This trend is shifting the role of developers from writing every line of code manually toward specifying problems, providing context, reviewing generated solutions, integrating components, and verifying system quality. Productivity may increase, especially for repetitive or standardized tasks, but the benefits depend on the developer’s ability to recognize errors and provide appropriate technical constraints.
AI-assisted programming does not eliminate the need for programming fundamentals. A worker who lacks knowledge of algorithms, data structures, software architecture, databases, and security may accept generated code without understanding its behavior. Such dependence can produce unreliable systems and accumulated technical debt.
2.3. AI Agents and process automation
AI agents are systems that can interpret objectives, plan actions, use external tools, retrieve data, and execute multi-step tasks. In IT environments, agents may create service tickets, analyze system logs, generate database queries, update documentation, run software tests, monitor infrastructure, or coordinate parts of a deployment process.
The expansion of agentic AI may automate complete workflows rather than isolated tasks. Nevertheless, reliable deployment requires access control, tool restrictions, logging, human approval, output validation, and mechanisms to stop or reverse incorrect actions. IT engineers will increasingly be responsible for designing these control structures.
2.4. Small, Specialized and Edge AI models
The development of AI is not limited to extremely large cloud-based models. Smaller and domain-specific models are increasingly used because they may require fewer computing resources, provide lower latency, reduce operating costs, and offer greater control over sensitive data.
Edge AI executes models on mobile devices, industrial equipment, cameras, gateways, or local servers. This approach is relevant to developing countries where Internet connectivity may be unstable or expensive. It is also useful for agricultural monitoring, manufacturing, transportation, health services, and public safety.
Small and local models create opportunities for IT professionals who can optimize models, integrate sensors, manage embedded systems, and adapt AI applications to local languages and operational conditions.
2.5. Retrieval-Augmented Generation and Organizational AI
Retrieval-Augmented Generation connects a generative model to external databases, documents, search systems, or organizational knowledge. Instead of relying exclusively on the information learned during model training, the system retrieves relevant content and includes it in the model’s context.
This approach is increasingly used for enterprise search, customer support, technical assistance, internal knowledge management, and educational services. It creates demand for competencies in databases, information retrieval, document processing, application programming interfaces, data governance, and access control.
2.6. Responsible AI, Cybersecurity and Governance
As AI systems are deployed in high-impact contexts, organizations are paying greater attention to transparency, fairness, privacy, security, intellectual property, and human accountability. AI systems can introduce new vulnerabilities, including prompt injection, poisoned training data, insecure plugins, model theft, leakage of confidential information, and automated social engineering.
Consequently, responsible AI is becoming an engineering requirement rather than only an ethical principle. IT professionals must understand model limitations, data provenance, risk assessment, security testing, audit trails, and human oversight.
3. Effects on the Work of Information Technology Engineers
3.1. Transformation of technical tasks
AI is likely to reduce the amount of time engineers spend on repetitive activities such as generating standard code, creating basic tests, searching documentation, formatting reports, and diagnosing common errors. The engineer’s work will increasingly concentrate on system-level decisions and difficult exceptions.
Important responsibilities will include defining requirements, selecting architectures, integrating AI with existing systems, validating model outputs, ensuring cybersecurity, monitoring performance, and evaluating business consequences. The value of an engineer will therefore depend less on the quantity of code produced and more on the ability to design dependable solutions.
This transformation is consistent with international labor research indicating that only a limited number of occupations consist entirely of automatable tasks. Most jobs combine activities that can be automated with tasks that require human reasoning, interaction, contextual knowledge, or accountability [1], [3].
3.2. Changes in software development roles
Junior developers have traditionally learned through relatively simple tasks, such as implementing basic interfaces, correcting minor defects, writing routine tests, and preparing documentation. Many of these tasks can now be accelerated by AI.
This creates a paradox. AI can help beginners learn and become productive, but it may also reduce the number of elementary tasks through which they traditionally gained experience. Employers may expect new graduates to complete more complex assignments earlier in their careers.
Senior engineers will increasingly supervise AI-supported workflows, review generated code, make architectural decisions, and resolve integration problems. Junior professionals must therefore develop code-reading, testing, debugging, and verification skills instead of relying on code generation alone.
3.3. Increasing importance of hybrid competencies
The labor market is likely to reward professionals who combine several areas of competence. Examples include software engineering and AI integration, cybersecurity and machine learning, cloud computing and data engineering, or information systems and business process analysis.
Demand is also expected to increase for AI and machine learning specialists, big-data specialists, fintech engineers, software developers, and cybersecurity professionals [4]. Nevertheless, technical expertise alone will not be sufficient. Creative thinking, analytical reasoning, resilience, communication, and collaboration remain important because AI-supported projects require interaction with users, managers, regulators, and domain specialists [4].
3.4. Productivity, quality andprofessional accountability
AI tools may improve productivity by generating drafts and alternative solutions quickly. However, faster development does not automatically produce higher-quality software. Generated code can contain hidden security vulnerabilities, inefficient algorithms, licensing risks, or incorrect assumptions.
Professional accountability remains with the engineer and the organization deploying the system. Engineers must be able to explain why a solution was selected, how it was tested, what data it uses, and what risks remain. Verification and assurance may therefore become more important as code generation becomes easier.
4. Effects on College-level IT graduates
4.1. Greater exposure of entry-level work
Graduates of college-level IT programs often begin with technical support, website maintenance, data processing, software testing, basic programming, device installation, network administration, and routine system operations. Many components of these jobs are structured and repetitive, making them suitable for partial AI automation.
Chatbots can resolve common support requests; coding assistants can generate standard website components; testing tools can create test cases; and monitoring platforms can classify routine incidents. Entry-level vacancies may therefore require fewer workers or higher skill levels.
This does not mean that college graduates will become unnecessary. Physical installation, local system support, user communication, device maintenance, contextual troubleshooting, and the adaptation of systems to specific organizations continue to require human participation. The challenge is that graduates must be able to work with AI-enabled tools rather than compete against them in routine task execution.
4.2. Opportunities in applied AI implementation
Developing countries need workers who can deploy and maintain practical AI applications rather than only researchers who design new foundation models. College graduates can participate in data preparation, system configuration, API integration, user support, model monitoring, cloud deployment, annotation, quality assurance, and maintenance. They may also support small and medium-sized enterprises that cannot employ large AI research teams. Typical projects may include intelligent customer service, document processing, inventory forecasting, product recommendation, digital marketing, industrial monitoring, and local-language information systems.
Applied competence can therefore offer a realistic employment pathway. Graduates do not necessarily need to develop a large language model from the beginning, but they should understand how to select, integrate, test, secure, and operate existing AI services.
4.3. Risk of skill polarization
AI adoption may increase the difference between graduates who possess strong foundations and those trained only to follow specific software procedures. Workers who understand programming logic, databases, networks, security, and system integration can adapt to new tools. Workers whose competence is limited to repetitive operations may face greater displacement risks.
Digital skills are particularly valuable in low- and middle-income economies because their supply remains limited. International evidence indicates that digital skills can generate stronger wage benefits in such economies than in high-income labor markets [5]. However, these benefits may be distributed unevenly between urban and rural areas, institutions, income groups, and genders.
5. Specific challenges for developing countries
5.1. Infrastructure and access gaps
AI systems require reliable connectivity, computing resources, cloud services, and access to digital devices. These conditions remain uneven in many developing countries. Limited infrastructure may prevent students and smaller institutions from gaining practical experience with modern AI systems.
The AI divide is not determined by model access alone. Effective adoption requires affordable Internet connectivity, relevant local data, digitally skilled workers, and sufficient computing capacity [2]. Without these foundations, developing countries may become consumers of foreign AI services rather than producers of locally relevant solutions.
5.2. Shortage of local-language and contextual data
Many global AI systems perform better in widely represented languages and contexts. They may provide inaccurate or culturally inappropriate responses for local languages, regulations, business practices, and educational content.
This limitation creates both a challenge and an opportunity. IT engineers and graduates can contribute to local datasets, language technologies, domain knowledge bases, and context-aware applications. Data quality, consent, copyright, and community representation must be considered during this process.
5.3. Education–industry mismatch
IT curricula may be updated more slowly than technology and labor-market requirements. Some programs continue to emphasize isolated software tools without sufficient attention to system thinking, project experience, cloud platforms, cybersecurity, data management, and AI-supported work.
The skills gap is already regarded by employers as a major barrier to organizational transformation, while a substantial proportion of occupational skill requirements is expected to change before 2030 [4]. Educational institutions therefore need mechanisms for frequent curriculum review and collaboration with industry.
5.4. Platform dependence and digital outsourcing
Developing countries may benefit from remote work and digital outsourcing. AI can improve the productivity of local professionals and allow smaller teams to participate in international projects. At the same time, routine outsourced services such as basic coding, data entry, transcription, and standardized support may be particularly exposed to automation.
Countries that compete primarily on low labor costs may face increasing pressure. A more sustainable strategy is to move toward higher-value services based on domain expertise, system integration, cybersecurity, quality assurance, local knowledge, and customer relationships.
6. Implications for IT education and workforce development
6.1. Preserve core IT foundations
AI literacy should be integrated into IT programs, but it should not replace foundational subjects. Students still require programming, algorithms, computer architecture, operating systems, databases, computer networks, software engineering, and information security.
These foundations enable graduates to evaluate AI-generated solutions rather than accept them uncritically. For example, students need database knowledge to verify generated queries, security knowledge to identify unsafe code, and software engineering knowledge to assess maintainability.
6.2. Integrate AI across the curriculum
AI should not be confined to a single optional course. It can be incorporated throughout the curriculum:
- Programming courses can teach AI-assisted coding and code verification;
- Database courses can include natural-language querying and vector search;
- Software engineering courses can cover AI-supported testing and documentation;
- Cybersecurity courses can examine both AI-enabled defense and AI-related attacks;
- Networking courses can introduce intelligent monitoring and edge AI;
- Professional practice courses can address ethics, privacy, intellectual property, and accountability.
The objective is not simply to teach students how to write prompts. Students should learn when AI is appropriate, how to provide context, how to test outputs, and how to document the use of AI.
6.3. Adopt project-based and work-integrated learning
IT education should provide realistic projects involving data, APIs, cloud services, security, deployment, and user requirements. Students should be assessed not only on whether an AI tool produces an answer, but also on the quality, reliability, transparency, and security of the completed system.
Internships and cooperation with local enterprises can expose students to actual constraints such as incomplete data, legacy systems, limited budgets, regulatory requirements, and user resistance. Such experience is especially important for college graduates whose programs emphasize applied employment skills.
6.4. Redesign assessment
Traditional assignments based exclusively on producing source code or written reports may no longer demonstrate individual competence because AI can generate both. Assessment should include oral explanation, supervised practice, debugging, code review, system demonstration, design justification, and reflection on AI use.
UNESCO recommends a human-centered approach to generative AI in education, including appropriate regulation, capacity development, privacy protection, and pedagogical validation [6]. Educational institutions should establish clear rules distinguishing legitimate AI assistance from academic misconduct.
6.5. Support lifelong learning and micro-credentials
Because AI technologies develop rapidly, initial education cannot provide all the knowledge required throughout a career. Universities, colleges, employers, and governments should support short courses and micro-credentials in areas such as AI-assisted development, cloud platforms, data engineering, cybersecurity, AI governance, and industrial AI.
Training should be accessible to existing workers as well as full-time students. Otherwise, the adoption of AI may widen inequality between workers who can continuously reskill and those who cannot.
7. Discussion
AI will not affect all IT occupations or all countries in the same manner. The outcome will depend on the structure of local economies, education quality, digital infrastructure, organizational investment, and national policy.
For IT engineers, AI is likely to increase productivity and shift professional value toward architecture, integration, security, verification, and business understanding. For college-level graduates, the effects are more mixed. AI can provide powerful learning and productivity tools, but it can also automate many traditional entry-level activities.
The appropriate response is neither to reject AI nor to reduce IT education to the operation of AI tools. Overreliance on automated generation may weaken fundamental competence and create systems that nobody fully understands. Conversely, prohibiting AI use may leave graduates unprepared for modern workplaces.
Developing countries should concentrate on areas where local knowledge and applied implementation create advantages. These include local-language systems, digital government, agricultural technology, financial inclusion, health information systems, smart manufacturing, cybersecurity, and solutions for small and medium-sized enterprises.
8. Conclusion
Artificial intelligence is transforming information technology work by automating routine tasks, augmenting professional decision-making, and creating new categories of intelligent systems. Generative AI, multimodal models, AI agents, edge intelligence, Retrieval-Augmented Generation, and AI-enabled cybersecurity are among the trends with the greatest influence on IT occupations.
The most probable outcome is not the disappearance of IT engineers but a restructuring of their responsibilities. Engineers will spend less time on repetitive production and more time on system design, integration, validation, security, governance, and communication. College-level IT graduates face greater exposure because many entry-level tasks are automatable, but they can remain competitive by developing practical skills in AI-assisted software development, cloud services, data management, cybersecurity, system support, and domain-specific implementation.
For developing countries, the benefits of AI will depend on investments in connectivity, computing resources, local data, education, and institutional capacity. IT education should combine enduring technical foundations with responsible AI use, practical projects, work-integrated learning, and lifelong reskilling.
Ultimately, the key professional competence will not be the ability to use AI in isolation. It will be the ability to combine human judgment, technical foundations, domain knowledge, and AI tools to build systems that are useful, reliable, secure, and socially responsible.
