Laura Rodríguez Salamanca reports in Rest of World on AI and Venezuelan migrants’ digital work

Laura Rodríguez Salamanca reports in Rest of World on AI and Venezuelan migrants’ digital work

Originally published April 16, 2025, Laura Rodríguez Salamanca’s report describes how Venezuelan migrants’ digital work is being reshaped by generative AI. Workers who once found income tagging data, moderating content, and writing search-optimized articles now face fewer tasks and lower pay, workers and researchers told Rest of World. Oskarina Fuentes, who moved from Venezuela to Colombia, says tasks have become irregular; she now earns around the minimum wage and borrows money to keep afloat. The report says demand for training data has fallen as models can generate synthetic training data, while clients seek more specialized data and model evaluations. Workers may struggle to reach those tasks: the report cites poor internet and power cuts in Venezuela, aging computers among some migrants, and missing documents that limit access to formal employment. Meanwhile, writer Maryfel Alvarado says she uses ChatGPT to produce more articles for some clients, but not for others who pay decently. Martín Lossada left SEO writing for oil-field work, where he earns more and works fewer hours. The report’s closing concern is that data tasks may disappear altogether, a possibility Fuentes says makes her anxious. How far the shift will reduce these workers’ opportunities remains unknown.

Publication: 18bbee24-2c4b-1bb7-1e17-1e53b005352c

attribution — pass: The reading correctly credits Laura Rodríguez Salamanca and Rest of World. It attributes worker accounts to Fuentes, Alvarado and Lossada, research claims to Maldonado of Fairwork, and company statements to Appen. It keeps claims attributed rather than stating them as fact.

By Laura Rodríguez Salamanca

summary — pass: The reading leads with the development: generative AI is reducing tasks and pay for Venezuelan digital workers. It explains the jobs in plain terms, such as tagging data and writing search-optimized articles. Synthetic data and model evaluations are defined beside the reading. It keeps the barriers, the mixed outcomes for writers and Fuentes's anxiety. The closing question follows from the evidence. One small looseness: the synthetic-data explanation comes from Maldonado but is attributed to 'the report'. This does not distort the meaning.

There are fewer AI-data jobs, too, as models can now generate their own synthetic training data, Óscar Maldonado, who studies the platform economy for Fairwork at the Oxford Internet Institute, told Rest of World.

date — pass: The stated date, April 16, 2025, matches the dateline and the originalPublished UTC day. The page was edited later (dateModified 2025-08-13). However, all cited passages are plainly part of the article narrative. There is no hindsight framing, and the possible disappearance of tasks is presented as a worry, not an outcome. Because this is a publisher page, no archive disclosure is needed.

16 April 2025 • Bogota, Colombia

support — pass: Each claim matches the source text: the scarcer and lower-paid jobs, Fuentes's earnings falling from about $500 to $320, synthetic data and client demands, and the infrastructure, computer and document barriers. Alvarado's output rising from three to five articles, Lossada's higher pay and fewer hours, and Fuentes's anxiety are also supported. The page was edited later, but the cited text is plainly part of the original piece. The limitations correctly note there is no independent measure of how much generative AI caused the decline.

He earns about $200 more and works fewer hours.

media — pass: No media is selected and no verified candidates were supplied, so nothing is misidentified.