Argentina's AI Social Twin Stumbles on Typos in Policy-Prediction Debut
In brief
- Argentina's Ministry of Human Capital unveiled Gemelo Digital Social, an AI system to predict social policy outcomes.
- The promotional video contained multiple spelling and grammatical errors, including misspelled Spanish words.
- The video featured a fake minister with holograms, unrelated flags, and corporate logos, sparking ridicule.
The System
Argentina's Ministry of Human Capital announced the Gemelo Digital Social, a virtual replica of Argentine society built to ingest data from government and private sources. The stated goal is to move the country from a reactive state—one that responds to social crises after they happen—to a predictive one that can model poverty and track policy effects in real time. Argentina's government claims this would be the first time the digital twin concept is applied to social policy at a national scale.
The ambition is clear. Instead of firefighting poverty and inequality after they emerge, the system would let policymakers simulate scenarios and optimize decisions before implementation.
The Rollout
The execution, however, was less polished. The promotional video released to announce the Gemelo Digital contained multiple spelling and grammatical errors—including a misspelling of the Spanish word "predecir" (to predict) as "predicir" and a missing accent on "múltiples" (multiple). Tech commentator Maximiliano Firtman flagged the broader problems: a fake minister presenting with holograms, Singaporean flags, an Amazon AWS logo, and what he called "a terrible speech."
President Milei closed the announcement with "MAGA. VLLC!"—a reference to Trump's slogan paired with his own political branding.
The irony wasn't lost on observers. A system designed to predict the future couldn't catch a typo in its own announcement. Social media users and tech commentators seized on the contradiction, turning the launch into a case study in how not to market an AI initiative.
The Broader Context
Argentina's bet on predictive AI for social policy isn't unique. In April 2025, the U.K. Ministry of Justice was reported to be secretly building an AI system to predict who might commit murder using mental health records and addiction history. Both efforts raise questions about how governments deploy predictive models and whether the infrastructure and oversight can keep pace with the ambitions.