The debate over AI replacing human jobs has moved beyond routine tasks and is now entering areas once thought to remain human-led, such as code review. A recent exchange between Elon Musk and a Google AI engineer has reignited discussions on whether even software engineers responsible for reviewing code could soon be replaced by AI systems.
The discussion began when Logan Kilpatrick highlighted a shift within the development ecosystem, noting that the bottleneck has rapidly moved from code generation to code review. He remarked that existing systems and norms are not yet equipped to handle this transition. His observation suggests that while AI tools have become highly efficient at generating code, human engineers are now spending more time reviewing and validating that output.
Responding to this, Musk shared a strong perspective on the future, stating that even code review could soon become obsolete as AI takes over the task. Kilpatrick replied briefly, saying he hopes that will be the case.
While Musk did not elaborate on his reasoning, recent advancements in AI coding platforms provide context to his remarks. Companies in this space are already working to minimize the need for continuous human oversight in software development.
One example is Cursor, which has introduced a system enabling AI agents to handle repetitive engineering tasks within workflows. Traditionally, developers operate in a loop-issuing instructions to AI, reviewing outputs, and refining prompts. However, this process becomes increasingly difficult to manage as projects grow and multiple AI tools are used simultaneously.
Cursor’s new approach aims to reduce this constant involvement. Instead of relying on manual prompts, AI agents can automatically initiate tasks when specific conditions are met-such as the addition of new code, incoming messages in team communication tools, or scheduled triggers.
Once activated, these systems can independently review code, analyze updates, and identify potential issues. Human engineers are brought in only when their input is necessary.
Explaining the concept, Cursor’s engineering lead for asynchronous agents, Jonas Nelle, said that humans are not entirely removed from the process but are engaged at the appropriate stages rather than initiating every step.
The concept builds on tools already in use within the platform. Automated systems can continuously scan newly added code to detect bugs without requiring manual input. With the latest advancements, this automation can extend to more complex functions, including security checks and comprehensive analysis across entire projects.
The company states that such systems are already operating at scale internally, handling multiple tasks each hour. As Nelle explained, while any automated task could technically be initiated by a human, automating these processes expands the range of tasks AI models can effectively perform within a codebase.
These developments raise a broader question about the future role of human engineers-if AI can both generate and review code, what responsibilities will remain for people?
If AI takes over its own tasks, what will humans do next? Elon Musk explains.
Musk has previously shared similar views on the future of automation. In earlier discussions, he stated that AI and robots could eventually replace all jobs. However, he suggested this may not necessarily be negative, adding that work could become optional-more like a choice, such as growing your own food instead of purchasing it.