NextFin News - In a series of recent industry discussions and strategic forecasts, Nat Friedman, the former CEO of GitHub and a prominent figure in the Silicon Valley venture landscape, has articulated a provocative vision for the future of software development: a world where humans no longer need to read or write computer code. Speaking at a technology summit in San Francisco earlier this month, Friedman argued that the rapid advancement of Large Language Models (LLMs) and specialized AI agents is pushing the "abstraction layer" of computing to its final frontier—natural language. This shift, according to Friedman, will eventually relegate the underlying syntax of Python, Java, and C++ to the same status as assembly language: essential for the machine, but invisible to the human creator.
The timing of Friedman’s prediction coincides with a period of intense technological nationalism under U.S. President Trump, whose administration has prioritized the acceleration of domestic AI capabilities to maintain a competitive edge over global rivals. According to The Information, Friedman’s thesis is built on the observation that AI is no longer just an "autocomplete" tool but is evolving into autonomous agents capable of understanding intent, architecting systems, and self-debugging. This evolution suggests that the primary skill for the next generation of developers will not be syntax memorization, but the ability to provide precise, high-level instructions and manage complex logic flows through natural language interfaces.
The data supporting this trend is compelling. Current GitHub CEO Thomas Dohmke recently noted that AI already generates upwards of 40% of the code in files where it is enabled, a figure that Friedman expects to hit 90% by 2028. This trajectory is driven by the increasing sophistication of models like GPT-4o and specialized coding agents that can pass the SWE-bench—a benchmark for resolving real-world GitHub issues—with rising efficiency. As these agents move from writing snippets to managing entire repositories, the necessity for a human to manually audit every line of code diminishes, replaced by automated verification systems and high-level logic checks.
From an analytical perspective, Friedman’s prediction signals a fundamental restructuring of the labor market within the technology sector. If code becomes a "commodity" generated by AI, the value of a software engineer shifts from technical execution to systems design and problem-solving. This democratization of development could potentially expand the global pool of "creators" to one billion people by 2030, as Dohmke has suggested. However, it also raises critical questions about the "black box" nature of AI-generated software. If humans stop reading code, the industry must develop new frameworks for security, accountability, and transparency to ensure that the invisible logic governing our digital world remains robust and unbiased.
Looking forward, the transition to a "no-read" future will likely occur in phases. Initially, we are seeing the rise of "Copilot Workspaces" where AI handles the boilerplate and initial implementation, while humans remain the final arbiters. In the medium term, Friedman anticipates that the "maintenance burden"—the tedious task of updating libraries and fixing legacy bugs—will be entirely automated. Ultimately, the goal is to return "joy" to the creative process by removing the mechanical friction of coding. As U.S. President Trump’s policies continue to foster an environment of rapid AI deployment, the vision of a natural-language-driven digital economy is moving from speculative fiction to an imminent industrial reality.
Explore more exclusive insights at nextfin.ai.
