[IT Cafe] IT Industry Boosts AI Usage... Are Entry-Level Developers Getting Nervous?

Major IT Companies Including LG CNS, Naver, and Kakao
Accelerate Adoption of AI Assistant Tools for Development
Concerns Rise Over Shrinking Roles for Entry-Level Developers

As domestic IT companies actively utilize artificial intelligence (AI) assistant tools in their development processes, a seismic shift is occurring in the developer ecosystem. The rapidly advancing capabilities of AI tools that assist in writing program code are significantly boosting productivity. However, there are concerns in some quarters that the position of entry-level developers will shrink, accelerating the polarization of engineers.

According to the IT industry on August 4, developers at major domestic IT companies are currently using AI development assistant tools in their daily work. AI assistant tools for developers currently available on the market include GitHub Copilot (Microsoft), Cursor, and Gemini Code Assist (Google), among others.

Provided by Pixabay.

AI assistant tools help verify (debug) code errors or identify areas for code improvement during the development process. They can also assist in structuring programs and can generate code from natural language input provided by developers. These tools can handle simple repetitive tasks required during programming or document code content. As the performance of these AI tools that support developers' work has reached a high level, domestic IT companies are either fully adopting AI tools in their operations or encouraging their use.

LG CNS has introduced an AI coding platform into every stage of system development. Since May, the company has been using the AI coding platform 'DevOn AI Driven Development (DevOn AIDD)' throughout the entire development process. LG CNS had already been using an AI coding platform at the coding stage, but has now advanced it into an 'AI Programmer' to assist developers at every stage of system development. In particular, a 'Coding Agent' feature has been added, which automatically performs everything from source code generation to testing and verification simply by entering commands.

NAVER is encouraging its employees to actively use the AI-based integrated development environment (IDE) 'Cursor' in their work. Cursor is an IDE developed by the U.S. startup Anysphere, and it supports features such as AI-powered code auto-completion, modification, and code writing within the code editor.

Kakao also fully adopted the AI-powered 'Vibe Coding' approach at its internal coding competition, the hackathon, held in June. A hackathon is a development competition in which participants solve a specific topic or implement their own ideas as prototypes within a limited time. Since 2013, Kakao has held an annual internal hackathon called '24K,' meaning '24 hours for crew (employees).' This year, the competition time was reduced to 10 hours and the event was renamed '10K.'

Participating teams had to complete everything from planning the program to implementing a minimum viable product (MVP) and presenting it, all within 10 hours. Instead, the AI-based development technique 'Vibe Coding' was fully introduced during the development process. Vibe Coding is a development method in which code is generated by making conversational requests to AI. Participants were able to delegate coding tasks to AI using AI development tools. Because AI could be utilized throughout the entire process, there were even teams composed solely of non-developer positions with no prior development experience.

As the performance of generative AI-based coding assistant tools rapidly improves, some predict that demand for entry-level developers will decrease. The Software Policy and Research Institute, in its report 'Outlook on Changes in SW Developer Recruitment and Generative AI' published in April, analyzed that "while demand for advanced developers is expected to increase due to roles such as AI utilization, system design, and solving complex problems, demand for entry-level developers will plummet as simple coding tasks are replaced by AI."

However, developers believe that it will still be difficult for AI to actually replace developers. While AI excels at writing code according to instructions, its decision-making ability is still lacking. One current developer pointed out, "AI's ability to understand the complex context of program code is far inferior to that of actual developers," and added, "No matter how well it writes code, there is a limit as it cannot make the decisions necessary during the development process."