Artificial intelligence was supposed to save software engineers from tedious work. One engineer says the reality at his company looks very different. Instead of gaining more time to solve difficult problems, developers are spending long days feeding tasks into AI and approving whatever comes back.
An anonymous engineer using the X handle @v0xium sparked a huge discussion after describing his new job at a large, unnamed company. He said Anthropic’s Claude Code handles specs, code, tests, tickets, reports, and other work across his team. His description was simple and harsh: “Nobody is thinking anymore.” The post drew millions of views and caught the attention of some of tech’s biggest names.
In the X post, @v0xium said he had been in his new role for about half a month. What he found left him frustrated. Employees across engineering levels were using Claude for much of their daily work, while management kept pushing teams to ship more features at a faster pace.
I am done with this shit. It is over. The state of engineering right now is horrible. It has been half a month since I started a new role at a big company. Nobody knows anything here. The specs, code, tests, PRDs, tickets, resolution of those tickets, reports, etc., everything is…
— voxium (@v0xium) September 20, 2026
The engineer says that people were working “12 to 13 hours a day just to press enter.” He said employees had little time to read the generated material or understand the code before moving forward. Engineers were still busy, but the intellectual part of the job appeared to be disappearing from their schedules.
“Writing code was never the bottleneck; it was the systems around it.” He argued that speeding up code generation does not automatically fix slow decisions, weak product planning, unclear goals, or poor coordination.
His frustration quickly reached prominent figures in the tech industry. Elon Musk responded to the post with one word, “Yikes.” Venture capitalist and “All-In Podcast” co-host Chamath Palihapitiya offered a much longer warning about what constant AI use could do to technical talent.
Palihapitiya compared the situation to people “pushing a button in front of casino slot machines.” His larger point focused on quality rather than raw speed. He argued that automated software should understand what people want and help them make better software, rather than simply producing more code at a faster rate.
GitLab CEO Bill Staples also joined the discussion. He wrote that “the human has become the orchestration layer for the machines.”
Faster AI Code Can Create More Work for Humans
Among people using AI agents, 69% said those agents had increased their productivity.
The same survey found serious friction beneath those gains. About 66% of developers said AI solutions that were “almost right, but not quite” caused frustration. Another 45% said debugging AI-generated code could take more time. Trust in AI accuracy also fell, even as adoption continued to rise.
That problem becomes harder when AI can generate code much faster than humans can properly inspect it. A developer might receive a working feature in minutes, but someone still needs to check security, logic, performance, maintainability, and compatibility with the rest of the product. Fast generation does not make those responsibilities disappear.
The Reality of AI Code
The result can be a growing backlog of code that looks complete but has not received enough human testing or scrutiny.
Menlo Ventures partner Deedy Das has raised a related concern. In June, Das said many software engineers were experiencing an “identity crisis bordering on depression.” He described a split between people who heavily rely on AI to produce work and experienced engineers left reviewing, correcting, and understanding the flood of generated code.
“The craftsmen are tired,” Das wrote in his own viral post. His argument suggested that AI does not always remove work from a team. In some cases, it shifts difficult work toward the people responsible for checking everything before faulty code reaches users.
The pressure can also extend beyond normal working hours. A July 2026 Coddy Tech survey of 305 developers found that 80% had experienced AI use as something closer to dependence than an advantage at some point. About 43% said they had continued coding with AI after hours despite planning to stop, while 32% reported delaying sleep.