Everyone thinks they're a developer now because they can prompt an AI to generate code. A viral essay captures what's being lost in the hype.
The core argument: "building an app" and "engineering a system" are completely different activities, but AI has collapsed the distinction in people's minds. AI made rapid prototyping trivial. It didn't change anything about what makes production software actually work.
The author uses Google Search as an example. The interface looks simple - a search box and results. But Google employs tens of thousands of engineers working on relevance, latency, scale, reliability, abuse prevention, security, privacy, cost optimization, and continuous adaptation. The interface is the least complex part of the system.
You can use AI to build something that looks like Google Search in a weekend. You can't build something that handles billions of queries, fights spam, prevents manipulation, scales across datacenters, and continues working when components fail. That's the engineering that AI hasn't touched.
The essay compares AI-generated apps to a clay Bugatti - something that looks like a Bugatti externally but lacks all the engineering that enables 260 mph performance safely. The appearance is easy. The function is hard.
This matters because the hype around AI coding creates unrealistic expectations. People assume that if you can generate a clone of an app in hours, building the real thing must not be that hard. They underestimate the accumulated knowledge, operational experience, and sustained effort required to make software actually work at scale.
We've seen this pattern before. Desktop publishing made it easy to create layouts, which some people thought meant graphic designers were obsolete. Stock photography made images cheap, which didn't eliminate photographers. Website builders commoditized basic sites, which increased demand for developers who could build complex applications.
Each wave of automation lowers the cost of surface-level production while increasing the premium for deeper expertise. AI coding is following the same trajectory.
The technology is impressive. The question is whether we can maintain clarity about what it changes and what it doesn't.
