Part V: Ideological Defense of AI-Assisted Development
The Instrumentalist Position
AI is a tool. Like all tools, it amplifies capability. A hammer doesn’t “build houses”—it enables humans to build houses faster. A bulldozer doesn’t “move earth”—it enables humans to move more earth than shovels allow. AI doesn’t “write code”—it enables humans to express intent more efficiently.
This is the instrumentalist position: AI assistance changes the scale of what’s achievable, not the nature of the work. The developer still needs to know what they want to build. They still need to verify the output. They still need to debug failures. They still need to make architectural decisions.
What they don’t need to do is type every character. They don’t need to memorize syntax. They don’t need to manually implement well-known patterns from scratch. The tedious parts are delegated; the creative parts remain.
Against “Purity” Arguments
Some developers argue that AI-assisted code is somehow “impure”—that real developers write everything themselves, that understanding requires manual typing, that there’s virtue in suffering through boilerplate.
This is cargo cult programming. It mistakes the method for the goal.
The goal is working software. The goal is maintainable code. The goal is correct implementations. Whether a human typed every character or described intent to an AI is irrelevant to these goals. What matters is: Does it work? Is it readable? Is it tested?
The purity argument also has a troubling implication: it gatekeeps programming based on typing ability, memorization capacity, and tolerance for tedium. These aren’t virtues. They’re accidents of circumstance that historically excluded people from the field.
Who Benefits from Anti-AI Sentiment?
When a new technology threatens existing power structures, those structures push back. The printing press was resisted by scribes. The automobile was resisted by horse breeders. The computer was resisted by human calculators.
AI coding assistance threatens the status hierarchy of software development. Seniority matters less when juniors can access the same knowledge. Memorization matters less when lookup is instant. Tedious expertise matters less when the tedium is automated.
Those who built their careers on syntax mastery, framework memorization, and boilerplate tolerance have rational incentives to delegitimize AI assistance. “Real developers don’t use AI” protects their investment in now-obsolete skills.
This isn’t to say all criticism of AI coding is self-interested. Valid concerns exist about understanding, security, correctness. But the volume and vehemence of anti-AI sentiment far exceeds what valid concerns would predict. Something else is going on.
Vibe Coding as Democratization
Consider who gains from AI-assisted development:
Non-native English speakers who can describe intent in their native language and get idiomatic code
Career changers who have domain expertise but not CS degrees
Domain experts who can implement their ideas without becoming full-time programmers
Junior developers who can learn by seeing their natural language transformed into code
Accessibility-limited developers who struggle with typing but not with thinking
Time-constrained developers who can’t spend weeks on boilerplate
Vibe coding doesn’t eliminate the need for programming skill. It relocates where that skill matters. Instead of syntax, it’s architecture. Instead of typing, it’s verification. Instead of memorization, it’s judgment.
The skills shift, but the expertise doesn’t disappear. It concentrates on what actually matters: making good decisions about what to build and verifying that it was built correctly.