I strongly disagree. AI is a knowledge gap amplifier—to an almost absurd degree. I’ve watched top-tier professors in their respective fields write prompts, and the results they extract from the models are exponentially better than what average users get.

Umberto Eco once said that the internet amplifies the wealth gap. AI is the absolute pinnacle of that phenomenon.

I'm from South Korea, and recent studies here are already showing a severe 'AI divide' emerging among middle and high school students. Lower-income households struggle to maintain educational engagement, premium AI subscriptions are prohibitively expensive, and crucially, the inputs (prompts) these students provide simply aren't good enough to get valuable outputs.

Let’s be honest: the free tiers of GPT and Gemini are terrible. For context, I spend around $300 a month on premium models and APIs, and the gap between free and paid—from output limits to tool availability—is massive.

More importantly, the current architecture of AI dictates that output is strictly bounded by input. Because LLMs navigate a latent semantic space based on token distribution, the deeper and more specific your terminology, the higher the quality of the response.

Asking an LLM to simply 'add a login feature' versus asking it to 'build a login feature while designing the core logic for the Auth server; keep the access token to a 15-minute lifespan in client memory; issue the refresh token as a Secure cookie, and apply Refresh Token Rotation (RTR) to prevent token hijacking' yields entirely different dimensions of code.

This creates a paradox: to use AI effectively, you ironically need deep, pre-existing domain expertise. Yet, the more you rely on AI, the more you outsource your critical thinking to it, making it harder to cultivate that exact expertise over time.

I do not believe AI is an egalitarian tool