George Hotz-the hacker who first cracked the iPhone at 17 and reverse-engineered the PlayStation 3 before Sony sued him-has published a blog post arguing that widespread adoption of AI coding agents will end in disaster.
"I’m calling it now, the adoption of AI agents into software development will be one of the most costly mistakes in the field’s history," Hotz wrote. "Agents cannot program, and it’s taking longer and longer to realize that they can’t."
The post arrives five days after prominent AI researcher Andrej Karpathy joined Anthropic's pre-training team, taking the opposite view-that AI agents have already transformed development. The split highlights an unsettled debate between two respected engineers.
Hotz spent six months testing agents on real projects, including parts of his open-source framework Tinygrad and a firmware reverse-engineering task. He describes the experience as a tool that "frontloads all the progress" but never completes the finishing work.
His central argument: high performers have the expertise to catch agent-generated errors before they ship, but weaker engineers-who use agents to produce 10 times their previous output-cannot. At scale, that dynamic degrades average code quality, masked by higher volume.
Hotz predicts "a golden era for buckets and buckets of slop, and a dark age for gems of quality." He points to Apple pushing AI coding tools across its engineering organization and asks: "Do you think macOS will get better or worse in the next 2 years?"