Patrick Ciriello’s Toyota Highlander became home sweet home—four grim months of it, family crammed in back, cat on the dash.
Walmart lots for overnight parking. Libraries and McDonald’s for free WiFi and job apps. A 60-year-old software vet with a master’s, he’d designed systems for banks, unis, pharma giants. Dotcom bust? Bounced back. ‘08 crash? Ditto. COVID? Scraped by. But 2023? Nada. Hundreds of apps—IT support, customer service, even deli counter. Zero bites.
Then that cryptic LinkedIn DM: “content writer.” Scam, he figured. Replied anyway. Boom—AI training gig. Labeling data. Evaluating outputs. Turns out, he’s not alone. Five silver-haired pros told the Guardian the same tale: fields drying up, age bias biting hard, so they’re annotating datasets for the very AIs gunning to eat their lunch.
Desperation Hits the Data Labels
U.S. workers over 60? They hunt jobs 50% longer than 20-somethings, per AARP’s Richard Johnson. Regain prior pay? Rare. Employers peg ‘em as pricey, outdated, tough to upskill. Half of 50-54-year-olds get shoved from long-term gigs pre-retirement, says Urban Institute. Pandemic nuked 5.7 million over-55s; many still adrift, Economic Policy Institute data shows.
Johnson nails it:
“There’s just a lot of desperation out there.”
Bridge jobs used to mean temp gigs, retail, Uber. Now? For engineers, docs, lawyers—it’s AI data annotation. Flexible hours. Remote. use old expertise. Pays $20-180/hour, depending on niche. Companies like Mercor, GlobalLogic, TEKsystems? They feed OpenAI, Google, Meta. Ciriello’s in.
But here’s the data-driven rub—and my sharp take: this ain’t salvation. It’s triage. Market dynamics scream short-term patch on a gaping wound. Older workers flood in because tech’s ageism is brutal; median tenure for 55+ devs? Under two years now, BLS stats whisper. AI training scales their know-how without the corporate politics. Smart pivot? Sure. Sustainable? Laughable.
Why Can’t Older Techies Land Roles Anymore?
Look at the numbers. Tech layoffs since 2022: 500,000+ jobs vaporized, per Layoffs.fyi. AI hype accelerates it—companies slash headcount, tout “efficiency.” But age skews the pain. Texas A&M’s Joanna Lahey tracks it: older applicants ghosted 40% more than juniors in controlled tests. Skills gap? Myth. It’s bias—“too expensive,” “can’t learn new stacks.”
Ciriello’s arc? Textbook. Industrial printer gig gone. Vermont motel aid cut—state-funded ‘cause hardship. Then car-camping. One thousand alerts daily. Desperation.
Lahey sees upside:
“[AI] training work may be better in some ways than those earlier alternatives.”
Flexibility trumps flipping burgers. But zoom out: global data annotation market? $2.4B in 2023, headed to $8.3B by 2028, Grand View Research says. Demand exploding as LLMs guzzle quality labels. Clients? Not just Big Tech—healthcare, finance need domain pros to flag AI hallucinations. A doc annotating med Q&A? Gold. Nurse tweaking diagnostics? Priceless.
Yet my unique angle—historical echo you won’t find in the Guardian piece: this mirrors ’90s coal miners retraining for IT helpdesks. Back then, unions pushed code bootcamps; now, no one’s organizing. Older coders label prompts instead of mining bits. Same trap—low-barrier entry, high burnout, automation lurking. Remember when CAPTCHA farms got AI-solved? Annotation’s next; Scale AI’s already piloting synthetic data to cut humans.
Does AI Training Pay Enough to Matter?
High-end? $180/hour for PhDs in rare domains. Ciriello? Won’t say, but it’s steady—beats zero. Side hustle for some. Lifeline for others. Firms like micro1, Alignerr boast contractor armies; 70% remote, per job boards.
Market truth: supply chasing demand. Platforms like Appen, Clickworker flooded with boomers. Pay eroding—entry-level dipped 15% YOY, Indeed data. Pros hold premium ‘cause nuance matters. An ex-banker spotting fintech edge cases? AI learns faster. But here’s my bold prediction: by 2026, agentic AI self-improves annotations 30%, Gartner-like forecast I’d bet on. These gigs? Two-year bridge, max.
Corporate spin? Companies pitch it as “democratizing AI.” Bull. It’s cheap expertise extraction—your 30 years for pennies on the expertise dollar. Desperation masks exploitation. Older workers aren’t “staying afloat”; they’re subsidizing the machine that sinks ‘em deeper.
And the irony? They’re training replacements. Ciriello tweaks industrial print models; soon, AI designs printers sans him. Docs flag med errors—until Gemini diagnoses solo. Brutal math: McKinsey says 45% knowledge jobs AI-disruptable by 2030. Older cohort hit first—less agile.
Policy fix? AARP pushes tax credits for upskilling. But tech’s velocity? Unforgiving. Johnson’s right—desperation rules.
The Bigger Market Shift
AI data needs explode: OpenAI alone burns billions on RLHF (that’s reinforcement learning from human feedback). Annotation’s 80% of training costs, insiders leak. Older pros fill quality gap—youths lack depth.
But strategy verdict: don’t bet the farm. Diversify—consulting, teaching, niche SaaS. Annotation? Cash now, pivot fast.
🧬 Related Insights
- Read more: Facebook’s Co-Improving AI Plan: Smart Dodge or Desperate Cope?
- Read more: Holotron-12B Delivers 2x Agent Throughput on One H100 – NVIDIA’s Secret Weapon Emerges
Frequently Asked Questions
What is AI data annotation work?
Labeling datasets, rating AI outputs, suggesting fixes. Pros like ex-software engineers or doctors use domain smarts to make models safer, accurate.
How much do older workers earn in AI training?
$20-180/hour. Domain experts top end; averages $40-60 for tech vets. Flexible, remote—but inconsistent volume.
Will AI training jobs last for older workers?
Short-term boom, 2-3 years max. Synthetic data and self-improving AI erode demand. Bridge, not destination.
