Artificial Intelligence and the IT Sector: Automation of Routine Tasks and the Rise of New Engineering Skills
As traditional entry-level software jobs rapidly disappear, AI-based skill sets are becoming indispensable for IT professionals.
The methodology that has been in practice for decades in the field of Information Technology (IT) is currently undergoing a historic transformation. According to a recent Business Standard report, Gururaj Deshpande, Chief Delivery Officer of leading IT services firm LTM, stated that following the advent of Artificial Intelligence (AI), traditional entry-level engineering roles have collapsed. Companies are now moving away from traditional entry-level jobs as AI reshapes their service delivery models.
This transition is extremely significant for the average software developer or IT engineer. Previously, new engineers fresh out of college were assigned tasks such as writing routine code, checking for bugs, performing basic database management, or conducting daily testing. Now, all these routine and repetitive tasks are being completed in moments by advanced AI systems, which has changed the very definition of employment in the tech industry.
Historical Background: The Journey from Computing to Modern AI
The term Information Technology was first used in a modern context in a 1958 Harvard Business Review article by Harold J. Leavitt and Thomas L. Whisler. They defined IT as the techniques for processing, the application of mathematical methods for decision-making, and the simulation of higher-order thinking through computer programs. The foundation of computer science was laid in the 1940s by machines such as the Zuse Z3, Colossus, and the 1948 Manchester Baby. Artificial Intelligence was established as an independent academic discipline with the 1956 Dartmouth Conference.
The history of AI has seen many ups and downs, known as 'AI winters,' which were periods of stagnation in research and funding. However, after 2012, when Graphics Processing Units (GPUs) were used to accelerate neural networks, deep learning methods surpassed all previous technologies. Subsequently, the arrival of the transformer architecture in 2017 led to explosive progress in machine learning capabilities. By the 2020s, Generative AI, and by 2024, the development of reasoning models, began to demonstrate human-equivalent performance in coding and solving mathematical problems.
- Problem analysisHuman evaluation of client business requirements and necessary software features.
- AI prompting and reasoningImmediate generation of primary architecture and code using chain-of-thought methods by specialized models.
- Automated testing and debuggingFinding errors and fixing code by AI agents instead of traditional manual testing.
- Human oversight and deploymentVerification of AI output by senior architects and integration into cloud infrastructure.
How Does the Automation of Routine Tasks Work?
Computing systems previously worked only on pre-determined rules (rule-based programming). In traditional software, an engineer had to write explicit code for every possible scenario. However, modern Large Language Models (LLMs) are trained on data from billions of websites and books. They are trained in 'chain of thought,' or step-by-step reasoning, which enables them to solve complex mathematical and programming logic.
For example, tasks in financial services such as calculating monthly installments (EMI) for loans, estimating returns on investments through SIPs, checking maturity interest for Public Provident Funds (PPF) or Fixed Deposits (FD), or obtaining estimates for pension funds under the National Pension System (NPS) previously required human engineers to develop separate algorithms and interfaces. Today, AI automatically and instantly creates these types of routine calculations and their associated software modules.
The Explosion in Data and Computational Capacity
Looking at the history of information technology, the capacity for data storage and processing has been continuously increasing. According to a study by Martin Hilbert and Priscilla Lopez, the world's general-purpose computing capacity doubled every 18 months between 1986 and 2007. During the same period, application-specific capacity doubled every 14 months, while global storage capacity doubled every 40 months.
Until the year 2002, most of the world's information was stored on analog devices, but that year, digital storage capacity exceeded analog for the first time. By 2007, approximately 94% of the world's data was stored in digital format, with hard drives and optical media being the primary mediums. Data storage capacity grew from less than 3 exabytes in 1986 to 295 exabytes in 2007. This massive database and relentless computing capacity have paved the way for training today's modern AI models.
| Criteria | Traditional entry-level role | AI-native engineering role |
|---|---|---|
| Main task | Manual coding, routine bug fixing, basic testing | AI model integration, prompt architecture, system design |
| Required skills | Knowledge of basic syntax and programming language | Machine learning, management of reasoning models, hallucination control |
| Market demand status | Reduction in hiring by companies (collapse status) | Obtaining premium wages with specialized skills |
| Productivity metric | Lines of Code | Quality of problem solving and system efficiency |
AI-Native Engineering Skills: Changing Demand and Premium Wages
According to a Times of India report, while routine IT tasks are being automated, professionals with specific specialized skills can command higher wages in the market. Knowing only general Java or Python coding is no longer sufficient. Companies now need engineers who possess an AI-native mindset, meaning those who can manage AI agents and build system architectures.
In AI research, agents are systems that perceive their environment and take actions to achieve specific goals. In modern engineering, a deep understanding of various types of machine learning—such as supervised learning (data classification), unsupervised learning (pattern discovery), and reinforcement learning (reward and penalty-based learning)—has become essential. Professionals who can control the flaws of AI models, such as 'hallucination' (providing false or imaginary output), are in the highest demand.
Implications for the Indian IT Sector and General Professionals
India's IT sector is the largest organized sector providing employment to millions of young people. Previously, millions of engineering graduates were trained by service companies every year and placed in back-office coding or maintenance roles. Gururaj Deshpande's statements clearly indicate that this traditional hiring model has now changed forever.
This is a major challenge for new graduates, but at the same time, it is also a new opportunity. Indian engineers must now keep pace with the global ecosystem working in the directions of system optimization, Natural Language Processing (NLP), computer vision, and Artificial General Intelligence (AGI). Companies are now focusing on increasing productivity rather than increasing headcount, which means that one skilled AI-native engineer can deliver results equivalent to a ten-member team from the past.
