AI tools are revolutionizing software development by automating repetitive tasks, refactoring bloated code, and identifying bugs in real-time. Developers can now generate well-structured code from plain language prompts, saving hours of manual work. These tools learn from vast codebases and provide context-aware recommendations that increase productivity and reduce errors. Engineers can prototype faster, iterate faster, and focus on solving increasingly complex problems instead of starting from scratch.
As code generation tools grow in popularity, questions arise about the future size and structure of engineering teams. Earlier this year, Garry Tan, CEO of startup accelerator Y Combinator, said that about a quarter of its current customers use AI to create more than 95% of their software. In an interview with CNBC“What this means for founders is that they don’t need a team of 50 or 100 engineers and they don’t need to raise that much capital. Funding takes much longer,” Tan said.
Coding using AI While it may provide a quick solution for budget-strapped companies, the long-term impact on the field and workforce cannot be ignored.
Human expertise could decline as AI-powered coding rises
In the age of AI, the traditional path to coding expertise that has long supported senior developers may be at risk. Easy access to large-scale language models (LLMs) allows junior programmers to quickly identify problems in their code. While this speeds up software development, it can take developers away from their work and slow the growth of core problem-solving skills. As a result, you may be able to avoid the intensive and sometimes unpleasant hours required to build your expertise and get on the path to success as a senior developer.
Consider Anthropic’s Claude Code. It’s a terminal-based assistant built on the Claude 3.7 Sonnet model that automates finding and resolving bugs, writing tests, and refactoring your code. Natural language commands reduce repetitive manual tasks and increase productivity.
Microsoft has also released two open source frameworks, AutoGen and Semantic Kernel, to support the development of agent AI systems. AutoGen enables asynchronous messaging, modular components, and distributed agent collaboration to build complex workflows with minimal human input. Semantic Kernel is an SDK that integrates LLM with languages such as C#, Python, and Java, allowing developers to build AI agents to automate tasks and manage enterprise applications.
The availability of these tools from Anthropic, Microsoft, and others may reduce opportunities for programmers to develop and deepen their skills. Rather than “banging their heads against the wall” to debug a few lines or select a library to unlock new features, junior developers might just ask AI for help. This means that advanced programmers with problem-solving skills honed over decades can become an endangered species.
Over-reliance on AI to write code risks undermining developers’ hands-on experience and understanding of key programming concepts. Without regular practice, you can have a hard time debugging, optimizing, or designing systems on your own. Ultimately, a decline in this skill can impair critical thinking, creativity, and adaptability. These qualities are essential not only for coding, but also for evaluating the quality and logic of solutions generated by AI.
AI as a Mentor: Turning Code Automation into Practical Learning
While there are legitimate concerns that AI will diminish the skills of human developers, companies should not ignore AI-enabled coding. Developers must carefully consider when and how to introduce AI tools during development. These tools don’t just increase your productivity; They act as interactive mentors, guiding programmers in real time by providing explanations, alternatives, and best practices.
when you areUsing sed as a training tool allows AI to enhance its learning by showing programmers why the code is broken and how to fix it, rather than simply applying a solution. For example, junior developers using Claude Code may receive instant feedback on inefficient syntax or logic errors, with suggestions linked to detailed explanations. This allows for active learning rather than passive correction. This is a win-win, allowing you to shorten project schedules without having junior programmers do all the work.
Additionally, the coding framework can support experimentation by allowing developers to prototype agent workflows and integrate LLM without requiring prior expert-level knowledge. By observing how AI builds and refines code, junior developers who engage with these tools can internalize patterns, architectural decisions, and debugging strategies, mirroring traditional learning processes such as trial and error, code reviews, and mentoring.
However, AI coding assistants should not replace actual mentorship or pair programming. Pull requests and formal code reviews remain essential for mentoring new, inexperienced team members. We are still far from the stage where AI can improve the skills of junior developers on its own.
Companies and educators can build structured development programs around these tools that focus on understanding code so that AI is used as a training partner rather than a crutch. This causes programmers to question the AI’s output and requires manual refactoring exercises. In this way, AI is not a replacement for human ingenuity, but a catalyst for accelerated experiential learning.
Bridging the gap between automation and education
When used intentionally, AI does more than just write code. We teach coding by combining automation and education to prepare developers for a future where deep understanding and adaptability remain essential.
Embracing AI as a mentor, a programming partner, and a team of developers who can address the problem at hand can bridge the gap between effective automation and education. We help developers grow with the tools they use. As AI evolves, so too will human skill sets, creating a generation of efficient and deeply knowledgeable programmers.
Richard Sonnenblick is the Chief Data Scientist for: floor plan.
