The Digital Press

Jonathan Blow criticizes current AI and large language models (LLMs) for not meeting his high standards in coding, especially for complex, low-level tasks, but many engineers find AI tools quite capable for everyday programming needs.

Why it matters: AI coding tools can significantly speed up routine development tasks and improve productivity for many engineers.

The big picture: While AI struggles with complex low-level coding, it excels at simpler, common code and secondary tasks like documentation and debugging.

The stakes: Overestimating AI’s current capabilities could lead to unrealistic expectations, especially for advanced, performance-critical software projects.

Commenters say: Most appreciate AI’s help with mundane tasks but agree it falls short on complex programming challenges and specialized engineering work.