From June 23 to 27, the 20th annual Internet Governance Forum (IGF) was held in Norway. This United Nations initiative gathers actors from around the world to debate and shape public policies regarding the present and future of the Internet.
Fundación Vía Libre participated both in person and virtually: Luciana Benotti was a speaker at the session “AI Agents: Ensuring Responsible Development,” and Beatriz Busaniche attended the Forum in person in Lillestrøm, Norway.
On Wednesday, June 25, Luciana Benotti, leader of Vía Libre’s AI team and a researcher specializing in natural language processing and artificial intelligence, participated in a one-hour roundtable. The session focused on the challenges posed by systems capable of researching complex topics, making decisions, and executing tasks autonomously while aligning with user goals.
Luciana introduced the ethical dimension to the discussion, specifically regarding bias. She shared a real case that emerged during an AI teacher training for 500 high school teachers conducted in 2024, using the HESEIA methodology for the critical evaluation of language models.
In this particular instance, teachers developed a vocational guidance agent based on a well-known LLM API, which was provided with a simplified interface. The agent, designed to help students choose a university major, tended to recommend options with lower tuition fees to low-income students. However, given that most universities in Argentina are tuition-free, the agent’s bias toward prioritizing majors based on low cost resulted in an invalid and decontextualized criterion.
Through its participation in the IGF, Fundación Vía Libre reaffirms its commitment to influencing the international agenda, ensuring that public policies on artificial intelligence respect and adapt to the specific dynamics of Latin America.
To learn more about the “HESEIA” teacher training course, please visit the following link: https://ia.vialibre.org.ar/curso-de-formacion/
