USP is the first Latin American institution to become part of the AI Alliance, which brings together more than 100 organizations promoting the collaborative development of AI.

Institutional https://jornal.usp.br/?p=800024
Published: 09/03/2024 at 11:02 AM

USP is the first Latin American institution to join the AI Alliance, a collaborative network of companies, startups, universities, research institutions, governmental organizations, and nonprofit foundations working at the forefront of artificial intelligence (AI) technology, applications, and governance. The AI Alliance focuses on fostering an open community and enabling developers and researchers to accelerate responsible AI innovation by promoting scientific rigor, trust, and safety.

As the first Latin American organization to join the AI Alliance, USP will focus on specific needs related to the development of foundational models, such as the unique blend of Spanish, Portuguese, and indigenous languages. The University will contribute to the development of tools for processing Portuguese and indigenous languages of Brazil.

Paulo Nussenzveig – Photo: Marcos Santos/USP Images

“USP closely monitors the impact of artificial intelligence technologies on everyday life and their effects on the social fabric. Being part of an international alliance with partnerships among governments, universities, companies, and civil society organizations to address technological advancements and governance challenges is an important opportunity. AI research directly affects people’s lives. USP seeks to take a leading role in this debate,” says Paulo Nussenzveig, USP’s Deputy Provost for Research and Innovation.

Through member-driven working groups, the AI Alliance pools resources and knowledge to address AI security concerns, while providing a platform to share and develop solutions that meet the needs of developers, researchers, and users worldwide. The alliance works on projects focused on six areas: capacity building and education; safety and trust; applications and tools; hardware enablement; foundational models; and advocacy.

USP will participate in two working groups: AI Safety and Trust Tools, and Foundational Models. The institution’s efforts in the AI Alliance will be coordinated by the Center for Studies in Artificial Intelligence and Machine Learning, led by Professor Fábio Cozman from the Polytechnic School (Poli). “USP’s participation will certainly allow the studies developed by the University to contribute to the progress of artificial intelligence research globally in a two-way exchange. On one hand, we gain more visibility, and on the other, the community in this field gains tools in which we have particular expertise,” Cozman highlights.

USP to have new study centers on greenhouse gases and artificial intelligence. USP’s participation in the Foundational Models working group will be coordinated by Professor Marcelo Finger from the Institute of Mathematics and Statistics (IME) and will cover multilingual and multimodal models, expanding the existing set of tools for foundational models. The group will promote collaboration among developers interested in training, fine-tuning, and deploying these models.

The University’s involvement in the AI Safety and Trust Tools working group will be coordinated by Professor Sarajane Marques Peres from the School of Arts, Sciences, and Humanities (EACH) and will provide insights and best practice guidelines on AI safety, trust, ethics, and cybersecurity. It will also work to improve evaluation methods and establish enhanced resources for testing AI models and applications.

This work will build on the four-year AI partnership between USP and IBM, a founding and key member of the AI Alliance, and will also be supported by their joint research lab, the Center for Artificial Intelligence (C4AI).