The AI Ecosystem: 7 Types of People
Look around at all the AI buzz and you'll quickly notice something strange: people are talking about the same technology but living in completely different realities. That's because there are fundamentally different types of people in the AI space, each with their own perspective, and the knowledge gap between them is massive.
- The Builders (0.1%)
These guys are actually building the foundation models. You'll probably never meet one. They're the 0.1% - the researchers and engineers at OpenAI, Anthropic, Google DeepMind, etc. who are neck-deep in transformer architectures and training dynamics. While everyone else is playing with the outputs, they're wrestling with the machinery.
- The Early Adopters
These are the X/Twitter crowd building on top of the latest releases. They're first to try everything, always experimenting, creating demos and tools around whatever just dropped. They understand enough to be dangerous, tweet endless hot takes, and are already selling "AI expertise" to everyone below them on this list.
- The Pragmatic Devs
This is your average developer who's now using ChatGPT when they're stuck, the same way they used Stack Overflow before. They have technical skills but treat AI as just another tool in the toolbox. The irony? These folks have the most untapped potential - they understand what's being generated while having enough technical chops to actually build something substantial.
- The Resistors
Think Linus Torvalds types - skilled developers who want nothing to do with AI. They've built impressive things the hard way and view the current hype with skepticism or outright disdain. They're not wrong about many of the limitations, but they're missing out on legitimate productivity gains.
- The Casual Experimenters
Non-developers who've typed 2-3 prompts and are still in that "wow, infinite possibilities!" honeymoon phase. They've generated some images, written a few emails, and think they're glimpsing the future. They haven't hit the limitations yet, so everything still feels magical.
- The Observers
People who've never really tried AI tools but have heard about them and seen the viral demos. They form opinions based entirely on secondhand information, either thinking AI will take all jobs tomorrow or dismissing it entirely as a fad.
The Reality of "Vibe Coding" Just go on X, type "vibe coding." You're gonna see what full AI is - hilarious, funny stuff, memes, games, demos that make VCs lose their minds. It's the tech world's new party trick: prompt, generate, launch, profit. 80% of actual developers know it's not sustainable, but that doesn't stop the hype train. These demos pop up daily - someone asks Claude or GPT to build a game, a website, a tool, whatever. No real coding knowledge required. Just vibes. The results look impressive in a 30-second video clip, and the engagement numbers go crazy. Non-technical people see it and think we're witnessing the death of traditional software development. The culture gap is so huge that even projects like the Levesio fly game get talked up by VCs at conferences. Is it impressive? Sure. But any dev who's actually AI-enabled knows it's impossible to maintain these projects or make them grow in quality at the same speed. At a certain point, you just go backward. Random people see these demos and think it's a linear trend upward - "if AI can do this now, imagine what it'll do in six months!" But it's logarithmic at most. We hit limitations, edge cases, complexity barriers that don't show up in the viral clips. The initial 80% of a project might take hours with AI, but the remaining 20% takes weeks of actual engineering - if it's possible at all. We are veryyy far from being able to build a sustainable product using AI only from no-code approaches. These vibe coding demos are like sandcastles - they look impressive until the tide comes in. They weren't built to last. Be amazed because it's genuinely impressive. But don't believe it's magic. And definitely don't make hypotheses or claims about something you have zero clue how it's working or what it's based on. Here's one example: it's actually easier to create a simple fly game than telling Cursor to translate a real-world app from Python to JavaScript. The former is a bounded, simple problem with visual feedback that looks cool in a demo. The latter requires understanding complex relationships, dependencies, and nuances that current AI just can't handle reliably. If you're a non-dev person interested in AI, try it yourself. Think about an idea that you defined clearly before you start vibe coding. 100% sure you gonna get stuck at some point - and that's the reality check the hype cycle desperately needs.