Latam-GPT: Chile Launches Latin American AI to Challenge Stereotypes
The launch of Latam-GPT marks a significant milestone for artificial intelligence in Latin America. Announced on February 10, 2023, this initiative aims to create a unique AI model that challenges stereotypes and promotes regional identity. Coordinated by Chile’s National Center for Artificial Intelligence (CENIA), Latam-GPT is a collaboration involving institutions from at least 15 countries including Brazil, Colombia, Mexico, and Argentina.
Overview of Latam-GPT
Unlike mainstream AI platforms like ChatGPT, Latam-GPT is designed as a linguistic model for regional applications. It will be publicly available as open-source software, allowing companies and public institutions to access it free of charge. This model is intended to counteract the prevalence of stereotypes often associated with Latin America, as expressed by Chile’s Minister of Science, Aldo Valle.
Key Objectives
- Break down stereotypes and prevent uniform representation of Latin America.
- Empower local institutions in developing AI technologies.
- Use more representative data from the region in AI training.
One concern highlighted is the limited representation of Latin American data in existing AI systems. Alvaro Soto, the director of CENIA, points out that, despite including some Latin American data, other global models often feature inadequate regional representation. This is crucial for accurately representing the diverse cultures and identities within Latin America.
Funding and Development
Latam-GPT has been developed with funding of approximately $550,000 from the Development Bank of Latin America and other sources. The model is trained on over eight terabytes of data, equivalent to millions of books. Experts note, however, that funding limitations compared to large-scale international projects could hinder its competitive potential.
Potential Applications
Despite these challenges, the project leadership is optimistic about various applications. Potential uses include:
- Healthcare: Addressing logistical issues and resource allocation.
- Public services: Tailored tools for improved communication.
One of the early adopters of Latam-GPT is the Chilean company Digevo. They plan to develop customer service chatbots that can understand regional dialects and idiomatic expressions, enhancing user interaction.
Future Directions
Currently, Latam-GPT primarily supports Spanish and Portuguese. Plans are in motion to gradually incorporate indigenous languages, reflecting the region’s rich linguistic diversity. The first version of the model was developed using Amazon Web Services, with future transitions to a supercomputer housed at the University of Tarapacá in northern Chile, costing around $5 million.
The launch of Latam-GPT represents a vital step toward technological sovereignty in Latin America and aims to reshape the narrative surrounding the region in the global AI landscape.
