When Speed Feels Like Loss
Yesterday’s most honest story about AI and work didn’t come with a layoff chart. It came with a feeling. In Fortune’s Eye on AI, Sharon Goldman captured the thing many teams won’t say out loud in standups: the day your tools make you faster can be the day your craft starts to feel less like yours.
Developers are living it first. “Vibe coding” with Claude Code, Cursor, and their peers has turned multi‑day builds into lunch‑break errands. Erik Meijer, who helped steer engineering at Meta, argued that Claude Code shoved the practice of software engineering forward in months more than decades of research did. If you’ve sat with these tools, you understand that’s not hyperbole—it’s a sudden reallocation of agency. The keystrokes that once signaled mastery turn into choices about what to accept, where to constrain, how to test. The work moves from writing to directing, from crafting to critiquing.
Fortune didn’t stop at velocity. It focused on what that velocity feels like when your identity was built on the friction. Gergely Orosz, long a voice for pragmatic engineering, has been publicly acknowledging the likelihood that most of his future code will be machine‑authored. Founder Andrew Duca described shipping faster and feeling worse—a kind of professional whiplash where tens of thousands of hours invested in a scarce skill suddenly read as replaceable.
The Pivot From Doing to Directing
This is the near‑term truth inside companies: productivity is climbing while pride and flow are sagging. Not because the work is gone, but because the center of gravity has shifted. Many white‑collar jobs are quietly being redesigned from creation into curation—prompting, orchestrating, and auditing AI output. The first draft is no longer the rite of passage; it’s a config file. The skill ceiling has migrated from depth in routine execution to judgment under uncertainty: framing problems, setting constraints, anticipating failure modes, stitching tools together. That’s not a demotion in importance. It is, however, a demotion in visibility, because judgment is harder to watch than typing.
The macro numbers haven’t caught up to this interior rearrangement. The Washington Post’s testing recently found today’s leading models could fully complete only a small fraction of real freelance tasks without human help—about 2.5% across hundreds of jobs. If you’re waiting for a clean line between humans and machines, you’ll wait a while. Augmentation is the mode; autonomy is sporadic. That’s exactly why the first employment effect isn’t a wave of pink slips. It’s a redesign of roles and a redistribution of status.
What Breaks Inside Companies
Organizations have long rewarded output that looks like making: lines of code, decks, drafts. AI blurs that signal. If the bot does the drafting and you do the deciding, what does “exceeds expectations” look like? Legacy career ladders still equate seniority with having written the hard parts by hand. Now the senior work is often making sure the hard parts never become hard—anticipating edge cases, constructing guardrails, formalizing acceptance tests, negotiating tradeoffs. Without explicit rewrites to roles and metrics, the people doing that quiet, higher‑leverage work get less credit while everyone’s dashboards report record throughput. That is a recipe for attrition, not celebration.
There’s also the matter of motivation. Flow comes from immersion and feedback; supervising a model is a staccato experience. You dip into the problem, you dip out to prompt, you evaluate, you iterate. The unit of satisfaction shrinks. Companies that treat this as a mere tooling change will misread the mood. You can raise the output line and still lose the room.
The New Premium
If scarcity is moving, the premium follows. Deep craft in repeatable tasks still matters, but it no longer monopolizes value. The differentiators are reframing messy goals into machine‑tractable plans, composing systems rather than artifacts, and exercising taste about when not to automate. Interface literacy joins domain expertise: knowing which model to ask, how to bound it, what data to distrust. Reliability engineering becomes the culture, not a department.
That shift doesn’t just touch engineering. Marketers are already storyboarding with generators, lawyers are triaging with summarizers, analysts are querying with natural language. The nearer future is not bots replacing teams; it’s teams whose most valuable people are those who can orchestrate machine labor without eroding trust, legal posture, or brand voice. In other words, managers of ambiguity rather than producers of first drafts.
Managing the Aftermath
Leaders should treat this as an identity transition, not a feature rollout. Rewrite performance definitions around outcomes, risk management, and system health, not keystrokes or slide counts. Make the invisible visible: credit the design of prompts, test harnesses, and review protocols the way you used to credit elegant functions. Preserve room for hands‑on practice—a few hours a week of no‑AI work on core skills—so mastery doesn’t atrophy and people keep a tactile connection to their craft. Build dual tracks that honor both the conductor and the instrumentalist: some roles will spend most of their day shaping machine work; others will go deep on the hard human bits the models still fumble.
Most importantly, narrate the change. People can tolerate new tools; they resist status collapse. Explain how judgment becomes the scarce resource, how careers progress in a world of cheap drafts, and how the organization will measure excellence when velocity is the baseline, not the badge.
The Real Headline
Fortune’s piece matters because it names the thing happening right now, inside the firewall: the first shock of AI at work is cultural. The jobs aren’t vanishing in bulk; the meanings are melting and re‑forming. Productivity is up. Flow is down. If you want the gains to stick, plan for the human part with the same rigor you apply to the model evaluations. The transition is not from employed to unemployed, but from maker to manager of machines. Done well, that’s not a loss—it’s a new definition of mastery. Done poorly, it’s a silent drain of pride that no dashboard will catch until the resignation letters arrive.
