An engineer working at a major tech company described a work environment where AI tools are generating nearly all aspects of software development, leaving engineers with long hours spent mainly validating code instead of reading or understanding it. The engineer, known on X (formerly Twitter) as v0xium, said that he and his colleagues spend 12 to 13 hours a day pressing Enter to approve code generated by Anthropic's Claude Code. He noted that while the company claims pushing code is no longer a bottleneck, the team remains slow. He added that no one reads the code, and that engineers at all levels—from junior to senior—rely on the AI for tasks ranging from writing specifications to generating final reports. He described the state of engineering as "catastrophic," with no one understanding the work they are doing and no one correcting bugs. Chamath Palihapitiya, an investor and co-host of the podcast All-In, echoed these concerns, comparing the situation to turning a generation of technical experts into retirees pushing buttons in front of slot machines. Earlier in the year, Deedy Das, an associate at Menlo Ventures, mentioned that software engineers are facing an identity crisis, even bordering on depression, due to being overwhelmed by code they did not write themselves. Studies on the impact of AI on developer efficiency have shown mixed results. In July 2025, the METR institute found that experienced developers took 19% longer to complete tasks with AI assistance, despite believing they were 20% faster. However, the organization estimated that improvements in AI tools likely led to better results by early 2026. A study by Anthropic in early 2026 found that engineers who used AI assistance scored 17% lower on a comprehension test, particularly in debugging performance. Engineers who used the AI to understand the code performed significantly better. The study suggested that the AI itself does not make people less competent, but those who merely approve of the code without understanding it may be losing skills quickly.