# The reality is that coal miners didn't just wake up one day and > Source: > Published: 2026-08-25 11:08:31+00:00 # The reality is that coal miners didn't just wake up one day and The core of his argument is that the US is fundamentally terrible at retraining workers. When heavy industry collapsed, the transition wasn't a smooth slide into the service or tech economy. It was a massive, painful dislocation. If we apply that logic to the current wave of generative AI, we aren't just looking at "upskilling"—we're looking at a massive segment of the population being left behind because the leap from manual or repetitive cognitive labor to managing an AI workflow is too wide for most existing social safety nets to bridge. ## The retraining myth vs. historical reality We often hear the optimistic take that AI will create more jobs than it destroys, which might be true in the long run for the GDP, but it ignores the human friction in the middle. **Skill Gap Disparity:** Moving from administrative work to prompt engineering isn't a linear progression; it's a complete paradigm shift in how one interacts with productivity.**Economic Friction:** Retraining programs are often expensive, slow, and frequently fail to align with actual market needs, leaving workers in a "limbo" state of underemployment.**Historical Precedent:** As Yang pointed out, the decline of coal mining didn't result in a sudden surge of software engineers in Appalachia. It resulted in regional economic decay. ## Why this matters for AI deployment If you're building an AI workflow or developing new LLM agents, you're participating in this shift whether you want to or not. From a technical perspective, we are seeing a massive push toward "agentic" workflows where the human is no longer the operator, but the supervisor. This changes the barrier to entry for work, but it also raises the floor of what is required to be "useful" in an AI-driven economy. Instead of just learning how to use a specific tool, the real skill is going to be high-level reasoning and system oversight. But how do you teach that to a workforce that has spent decades performing specialized, repetitive tasks? The tech community loves to talk about the "democratization of intelligence," but we rarely talk about the "displacement of livelihoods." If we don't figure out a way to handle the social side of this deployment, the backlash against AI might become much more aggressive than just regulatory hurdles—it could become a fundamental rejection of the technology by the very people it's meant to "assist." We need to move past the "everyone can be a coder" delusion and start looking at what real-world, practical training looks like for people whose jobs are being eaten by an API call. [Next Insta360 X6 is trying to make you forget it is a 360 camera →](/en/news/7627/) ## All Replies （0） No replies yet — be the first!