Patrick Ciriello, a 60-year-old with a master's degree in information management, spent nearly a year jobless after losing his position in early 2023. Despite hundreds of applications for IT, customer service, and even a supermarket deli job, he received no offers. His family, already living in motels in Vermont, became homeless when state funding ended, sleeping in a Toyota Highlander for four months.
In March 2024, Ciriello found work through a cryptic LinkedIn message: training artificial intelligence models. He is one of five skilled workers aged 50 and older who told the Guardian they have turned to data annotation—labeling and evaluating data to improve AI systems like ChatGPT or Gemini—after failing to find jobs in their fields.
Data annotation involves reviewing AI outputs to flag errors and suggest improvements, ultimately helping models become capable of replacing human workers. Companies like Mercor, GlobalLogic, and TEKsystems hire contractors for this work, serving clients such as OpenAI, Google, and Meta. Top experts can earn over $180 an hour, but for many older workers, it is a last refuge in a tough job market.
US workers over 60 take about 50% longer to find jobs than younger people, and few regain previous earnings, according to Richard Johnson of the AARP Public Policy Institute. He notes that employers often wrongly view older workers as expensive and hard to train. About half of workers aged 50-54 are pushed out of long-term jobs before they expect to retire, a trend worsened by the pandemic, which saw 5.7 million over-55s lose jobs in early 2020.
“There’s just a lot of desperation out there,” Johnson said. For Ciriello, training AI that may replace professionals like him is a necessary survival tactic in a brutal economy.
