A.I. Isn't Replacing Every White-Collar Job, at Least Not Yet.

A.I. code generators can speed up some software projects but typically require close human oversight, and experts warn they can introduce code quality and security problems.

A new wave of A.I. code generators is changing how software is built. Programmers report much faster progress on some projects while noting the need for close review and testing. Researchers and security experts say these systems can produce flawed or insecure code that creates long-term maintenance costs. Debate continues over how these tools will affect different levels of software work. What we know: - Tools such as OpenAI's Codex and Anthropic's Claude Code can produce working code quickly and have allowed experienced programmers to complete some projects far faster than before. - People using these systems report they must supervise, test and revise generated code to catch mistakes and security issues. - Studies from Carnegie Mellon found short-term speed gains but also declines in code quality that can increase "technical debt" and slow projects over time. - Public examples, including a social network for A.I. agents called Moltbook, have shown how vulnerabilities can expose private data when generated systems are not carefully secured. - Experts are divided: many expect routine junior programming tasks to be most affected, while some argue skilled programmers will continue to be needed for planning, review and complex systems work. Summary: A.I. code generators are accelerating certain development tasks while also introducing quality and security risks that require human oversight. The immediate effects may include faster completion of some projects and shifts in the types of tasks done by different programmers. Whether these tools will substantially reduce demand for experienced coders remains undetermined at this time.