Artificial intelligence is rapidly changing software engineering, forcing developers to reconsider the skills that once provided career security. Work that centered on writing code, solving problems, and designing software is increasingly shifting toward reviewing and validating code produced by AI. Some engineers worry that relying too heavily on these tools could weaken their core programming abilities, says The Guardian.
The disruption comes after years in which software engineering ranked among the largest and highest-paying professions in the United States. Since ChatGPT arrived in 2022, however, more than 600,000 U.S. technology workers have lost their jobs, according to Layoffs.fyi. Computer science graduates have also faced rising unemployment and underemployment, while U.S. technology job postings declined 36% between 2020 and 2025.
The changing market does not necessarily mean software engineers will become obsolete. Instead, their responsibilities are evolving. As AI handles more code generation, engineers may become increasingly valuable for defining problems, designing systems, identifying vulnerabilities, checking security, and finding errors in machine-generated code. Strong technical fundamentals remain important because effective AI use requires engineers who can recognize when its output is wrong.
Software engineer George Dover illustrates one path forward. After losing his job in 2024, he experimented extensively with AI-generated code, studying its errors, redundancies, and limitations. After about 400 applications, he eventually secured an AI-oriented software engineering position.
Other developers are responding differently. Some are deliberately practicing coding without AI to preserve their skills, while others are considering leaving the profession. Concerns about layoffs and AI adoption are also encouraging collective action. Organizations such as What We Will and the Tech Workers Coalition are helping technology workers navigate layoffs, negotiate severance, develop new skills, and organize.
The emerging software engineer may therefore write less code directly but carry greater responsibility for judging, directing, and correcting what AI produces.