The future of programming and the new role of the programmer in the age of AI

The role of programming is undergoing a profound change. We are entering what Andrej Karpathy calls “Software 3.0,” where traditional coding and machine learning coexist with natural language prompts. In “Software 1.0,” you wrote it; it was a program created by humans. In “Software 2.0,” you trained it; it was a model that learned from data. In “Software 3.0,” you simply ask it. 

Thus, today’s developers use AI tools as virtual colleagues, improving code and designs through conversation. Importantly, modern AI platforms are becoming operating systems in themselves, so engineers will coordinate AI workflows instead of manually writing each line. We are moving away from the notion of developers as mere manual programmers and toward becoming coordinators of AI-driven development ecosystems.

Human supervision and the ladder of abstraction

Even though AI handles execution, human ingenuity and strategy remain essential. We must design the right concepts and oversight layers for AI agents, preserving our agency and values. In other words, humans establish the purpose and objectives — the what and the why — while AI systems determine how to achieve them.

As AI Policy Perspectives points out, the human role is analogous to setting a navigation destination: “steering the what and the why, even as AI optimizes the how.” Executives and architects won’t debug every line of code, but rather define desired outcomes and review high-level summaries. With each step up the ladder of abstraction, new complexities emerge, so our teams must incorporate governance and oversight into every AI workflow.

In this way, AI platforms are becoming a kind of high-level operating system or framework, and the true value of the engineer will not lie in their ability to memorize syntax, but in their ability to orchestrate complex AI workflows, integrate services, and, above all, formulate the right questions.

A new playing field where human responsibility increases

This new scenario represents a crucial shift. Technology leaders will focus on defining desired outcomes at a macro level and verifying the high-level results that AI delivers. The price of this dynamic is a new and complex layer of responsibility: governance.

As we move up each step in abstraction, the complexities multiply. When we delegate tasks to AI, we must incorporate robust oversight and governance systems into every workflow. We must design ethical safeguards and ensure that human values ​​and agency are not diluted.

In practice, this means that non-technical skills, such as critical thinking, business strategy, systems architecture, and ethics, will become core to development. An engineer’s ability to translate a complex business problem into clear instructions for AI will be infinitely more valuable than their mastery of any specific programming language.

The future of programming isn’t total automation, but rather human amplification. By freeing us from the manual and repetitive tasks of coding, AI pushes us toward higher-value tasks: strategy, purpose, and oversight.

Programmers who don’t embrace this role shift and remain anchored to the line of code will be left behind. It’s crucial to embrace this new horizon and the future of programming, which will undoubtedly boost the role and importance of programmers within organizations.

The author of this article is Gastón Milano. With more than 25 years of experience in technology, software development and innovation, Milano is currently Chief Technology Officer (CTO) of Globant Enterprise AI, Globant’s enterprise artificial intelligence platform. A Systems Engineer from the University of the Republic (Uruguay), he specialized in Model Driven Development, a field in which he has been a reference since joining GeneXus, a company acquired by Globant in 2022, in 1997. Milano’s career combines a strong technical vocation with teaching and outreach. In 2019, he was appointed member of the National Academy of Engineering of Uruguay.