March 23, 2026
Turkeys vs Turing
White-Collar AI Apocalypse Narrative Is Just Another Bullshit
Commenters split: 'Slow then sudden' doom vs 'Relax, it's hype'
TLDR: The piece argues AI hasn’t wiped out white‑collar work—customer support is rehiring and a 90% automated system was shelved because the last 10% is hardest. Commenters are split between “a sudden AI wave is coming” and “calm down,” with extra spice from doomscrolls, turkey warnings, and bot-conspiracy jabs.
The article slams the “AI will kill desk jobs” story with a simple vibe check: customer support hiring bounced back in mid-2025, and one team that automated 90% of support tickets still killed the project because the brutal 10% ate all the time. Translation: we’ve built a talking FAQ, but the messy, one-off problems still need humans. The author calls these “semi-decidable” jobs—easy stuff gets solved, weird edge cases chew budgets—and cracks that most coding is the boring bits you could do drunk.
Cue the comments cage match. One camp is yelling “the real wave is coming”, with robotswantdata warning it’ll be “slow then sudden” as AI agents reshape work. Another camp is like bluegatty, shrugging that it’s already bad for some roles, but not the whole office. Then there’s the drama gremlin energy: intellectronica’s Thanksgiving turkey parable (“life’s great… until it isn’t”) spooked the thread, kensai linked a doom scroll of layoffs at x.com/TechLayoffLover, and monegator accused the room of doomposting and even hinted at bots pushing fear. The memes? “FAQpocalypse,” “turkeys vs Turing,” and a running joke that AI nailed the easy 80%, leaving humans to wrestle the cursed 20%. If the apocalypse is coming, it’s apparently on hard mode.
Key Points
- •The article contrasts claims of rapid AI capability gains with customer support employment rebounding toward pre-Covid levels by mid-2025.
- •Despite availability of subsidized models and RAG, companies did not fully automate customer support and began rehiring staff.
- •An internal LLM system reportedly automated 90% of support cases, but the project was dropped because the remaining 10% took most of the team’s time.
- •The author characterizes many white-collar roles as semi-decidable: routine cases are automatable, while rare, ambiguous cases dominate costs (an 80/20 dynamic).
- •A time-boxing and backlog (“dovetailing”) workflow is described to show how unresolved, difficult cases accumulate and drive overall costs, limiting full automation.