Chile has introduced Latam-GPT, the first open-source large language model designed specifically for Latin America’s cultural and linguistic context. The project is led by Chile’s National Center for Artificial Intelligence (CENIA) in collaboration with more than thirty research institutions across the region. Developers trained the model using datasets representing Spanish, Portuguese and regional cultural material in an effort to reduce biases that often appear in AI systems trained primarily on English-language data. The model was built using millions of data points collected from Latin American sources and is intended to support applications such as educational tools, chatbots and digital public services. Researchers say the initiative is an important step toward ensuring that AI technologies reflect regional identities and languages rather than reproducing biases embedded in global datasets. Latam-GPT is also designed to serve as foundational infrastructure that local developers can use to build new AI applications tailored to Latin American societies.
Read more here
