According to Fortune, AI coding tools like Anthropic’s Claude Code, OpenAI’s Codex, and Cursor are enabling “vibe coding,” where developers describe features in natural language to generate software in minutes instead of days. Former Meta engineering leader Erik Meijer said Claude Code alone has pushed software engineering further than 75 years of academic research. However, this productivity surge is causing a wave of disorientation and grief among developers, with influential voices like Gergely Orosz of The Pragmatic Engineer and Andrew Duca of Awaken Tax writing publicly about their sense of loss. Meanwhile, the AI funding frenzy continues, with Anthropic planning a $10 billion raise at a $350 billion valuation and Elon Musk’s xAI closing a $20 billion round backed by Nvidia.
The Unexpected Grief of Ease
Here’s the thing about a productivity miracle: it can feel pretty awful. The Fortune piece hits on something raw that’s been bubbling up in tech circles. Developers aren’t just worried about job loss. They’re mourning the craft itself. That “flow” state, the deep pride of untangling a complex bug or architecting an elegant solution—it’s being replaced by the role of an AI supervisor. You’re not in the zone writing logic; you’re editing and prompting. And for folks who’ve spent a decade or more honing this specific craft, that’s a massive identity shift that happened almost overnight.
Gergely Orosz’s blog post on “The Grief When AI Writes All the Code” and Andrew Duca’s viral tweet about feeling “kinda depressed” aren’t outliers. They’re the canaries in the coal mine for professional knowledge work. When the “10,000 hours” you poured into a skill gets commoditized by a tool, what’s left? Your value has to shift upstream to problem definition, strategy, and taste—things that are harder to measure and, frankly, harder to initially prove you’re good at. It’s disorienting.
A Preview For Every Desk Job
Software development is just the first wave because it’s uniquely suited for AI. The rules are clear, the feedback is instant (the code runs or it doesn’t), and there’s a universe of training data on sites like GitHub. But the philosophical reckoning is coming for everyone in a white-collar role. Think about it. If AI handles the hard *execution* part of your job—the drafting, the analysis, the building—what are you for?
The article asks a brilliant, uncomfortable question: what if the struggle *is* the point? What if wrestling with a difficult problem is precisely what gives work its meaning and allows us to feel pride? We all hate mindless grunt work, sure. But there’s a vast middle ground between data entry and pure strategy. That’s the ground AI is currently colonizing. And it’s forcing a weird choice: do you become an AI manager, or do you deliberately carve out “human-only” work to preserve your own sense of capability? It’s early, but this feels like the real conversation we should be having, way beyond the “will it take my job?” headlines.
The Sobering Reality Check
Now, for all the angst and the astronomical funding rounds—$350 billion for Anthropic is a almost incomprehensible number—it’s worth a reality check. The piece mentions a Washington Post interactive feature based on a study from Scale AI and the Center for AI Safety. They tested top models on hundreds of real freelance tasks. The result? AI successfully completed just 2.5% of tasks on its own. 2.5%!
That’s the giant gap between “vibe coding” a specific function and autonomously doing a real job from start to finish. The output often *looks* right but falls apart on closer inspection. So, are we grieving a bit prematurely? Maybe. But the feeling is real because the *direction* is clear. The tools are getting better, and the economic pressure to use them is immense. Even if AI only gets you 80% of the way there, that still fundamentally changes the nature of the remaining 20% you have to do. You’re not a builder anymore; you’re a quality assurance inspector and a creative director. That’s a huge mental shift.
Finding Flow Elsewhere
So where does this leave us? The piece ends on a poignant note with Erik Meijer, the former Meta exec who marveled at AI’s progress. His solution? He’s gone back to making music with real instruments. Just to “experience that flow” again. I think that’s a powerful metaphor for the near future. As AI takes over the execution of professional skills, we might have to seek the satisfaction of mastery elsewhere—in hobbies, in sports, in arts. Or we have to radically redefine what mastery means at work. It’s no longer about how well you code, but how well you guide the system that codes. That’s a colder, more managerial skill. And not everyone signed up for that.
Basically, we’re in a messy, emotional transition. The money is flooding in, the tools are getting scarily good at specific tasks, and the humans using them are feeling a mix of awe, excitement, and profound loss. Denial, anger, bargaining, depression… acceptance. Sounds familiar, right? The grief cycle isn’t just for personal loss. It might be the perfect framework for the professional upheaval that’s just getting started.
