AI Will Not Replace Software Engineers: Data and Reasons

In June 2026, Simon Willison's analysis, based on the essay by Arvind Narayanan and Sayash Kapoor, debunks the myth that AI will replace software engineers. Despite advances in generative AI, concrete data shows that the profession most exposed to automation has not suffered significant job losses. This article expands on the historical context, data sources, impact on companies and users, and compares with previous events to provide a comprehensive view.

What Happened?

Since 2023, generative AI, such as GitHub Copilot and ChatGPT, has demonstrated the ability to write functional code, leading to speculation about the end of human programming. However, layoff data from New York —the first state to require reporting AI-related layoffs— shows that no company checked that box in 2025. According to an analysis by Hunton Andrews Kurth, of more than 160 companies that filed WARN notices in the first full year, not a single one indicated AI as the cause. This contradicts the apocalyptic predictions of 2023, when studies like Goldman Sachs' estimated that 300 million jobs could be affected by AI. AI accelerates code writing, but it does not replace the essential tasks of an engineer.

Why Is This Important?

Willison identifies three bottlenecks that AI cannot overcome: deciding and specifying what to build, verifying and taking responsibility for the result, and deep human understanding of the code, business, and environment. Without these capabilities, AI is a tool, not a substitute. This finding aligns with previous studies, such as Microsoft Research (2024), which showed that developers spend 60% of their time on tasks unrelated to writing code, such as meetings, code review, and system design. Productivity increases, but the engineer's value depends on contextual understanding. Even with advanced assistants, final responsibility rests with humans who understand the problems and solutions. Compared to the automation of accounting in the 1990s, which eliminated repetitive tasks but created financial analysis roles, AI in software is following a similar pattern: increased productivity without mass unemployment.

Consequences for the Labor Market

Far from eliminating jobs, AI is transforming the engineer's role toward higher-value strategic tasks. Companies investing in AI for programming need more engineers to validate and direct those tools. According to data from the U.S. Bureau of Labor Statistics, software development jobs are projected to grow 25% between 2023 and 2033, well above the average for all occupations. Demand for professionals who combine technical skills with critical thinking and communication continues to grow. A 2025 McKinsey report indicates that companies adopting AI in software development report a 30% increase in productivity, but also a greater need for senior engineers to supervise and debug generated code. This contrasts with the fears of 2023, when it was feared that AI would replace juniors; in reality, juniors now learn faster with AI assistants, but their value remains tied to business understanding.

What Readers Should Know

• AI does not cause mass unemployment in software; New York data and employment projections confirm this.
• Human skills (decision-making, verification, understanding) are irreplaceable, as Willison points out.
• Engineers should focus on high-level tasks and understanding the business to stay relevant.
• AI is an ally, not a threat, if used correctly; the history of automation shows that tools increase productivity without eliminating entire professions.

“AI accelerates code writing, but software engineering is much more than that.” – Simon Willison

In summary, the myth of mass replacement has been debunked by data and qualitative analysis. AI will continue to be a powerful tool, but the future of software depends on the human ability to decide, verify, and understand.