Chile launches open-source AI model designed for Latin America
SMRTR summary
Eight terabytes of data, equivalent to millions of books, now power Latin America's first homegrown artificial intelligence language model. Chile launched Latam-GPT this week, marking a two-year collaboration between over 30 institutions across eight countries to counter the English-language dominance plaguing global AI systems.
"Latam-GPT is trained with a proportion of Latin American data that previously did not exist online and was not included in existing models," said Rodrigo Durán, executive director at Chile's National Center of Artificial Intelligence. The open-source model captures regional cultural realities often overlooked by mainstream AI platforms like ChatGPT.
Built on a modest $550,000 budget, the project aims to give Latin America its own voice in the global AI race, where the US, China, and Europe control most of the world's powerful data centers. While operating primarily in Spanish and Portuguese for now, developers plan to incorporate Indigenous languages in future versions.
The initiative represents more than technological advancement. As one researcher noted, it ensures "at the very least, everyone is included in the training."
SMRTR provides this summary for quick context. The original article belongs to AP News.
Read the original article