Chile launches Latam-GPT to combat AI biases

SANTIAGO (Web Desk) – Chile is set to launch Latam-GPT, an open-source artificial intelligence model designed to challenge the US-centric biases prevalent in mainstream AI systems. The model was developed by the Chilean National Center for Artificial Intelligence (CENIA) using millions of data points from Latin America to reflect the continent’s cultural diversity.

The initiative aims to prevent stereotypical depictions of Latin Americans, such as images of Chilean men wearing ponchos in front of the Andes, and to ensure local traditions and languages are properly represented in AI models.

Unlike closed generative models like ChatGPT or Google’s Gemini, Latam-GPT is fully open-source, allowing programmers to customize its functions. It has been trained on over eight terabytes of data, equivalent to millions of books, and primarily supports Spanish and Portuguese, with plans to incorporate Indigenous Latin American languages.

UN Experts Alarmed by Conviction of Imaan Mazari and Hadi Ali Chattha, Cite Threat to Rule of Law

The project was made possible with a modest budget of $550,000, funded largely by the Development Bank of Latin America (CAF) and CENIA resources. While the first version was developed on Amazon Web Services, future training will take place on a supercomputer at the University of Tarapaca in northern Chile.

CENIA director Alvaro Soto emphasized that Latam-GPT will allow institutions and companies to develop AI applications tailored to regional needs, such as hospitals with logistical challenges or customer service programs that recognize local slang, idioms, and speech patterns.

While experts acknowledge that Latam-GPT cannot compete with major global AI models, Chilean entrepreneur Roberto Musso praised its potential, stating his company plans to use it to create more culturally accurate customer service applications for airlines and retailers.

Chilean Science Minister Aldo Valle said, “Latin America cannot simply be a passive user of AI systems. That could result in the loss of a significant part of our traditions.”

Latam-GPT marks a major step toward regional AI sovereignty, highlighting the growing global trend of culturally sensitive, public AI models seen in Singapore’s SEA-LION and Kenya’s UlizaLLama projects.