Global AI Initiative  

ICTP has joined an elite group of international research institutes and high-tech companies aiming to shape the evolution of Artificial Intelligence (AI) by working together for the advancement of open, safe and responsible AI.

Launched on 5 December 2023, the group, called the AI Alliance, is spearheaded by IBM and Meta. The Alliance is focused on fostering an open community and enabling developers and researchers to accelerate responsible innovation in AI while ensuring scientific rigor, trust, safety, security, diversity and economic competitiveness. Members of the Alliance --  leading developers, scientists, academic institutions, companies, and other innovators -- will pool resources and knowledge to address safety concerns while providing a platform for sharing and developing solutions that fit the needs of researchers, developers, and adopters around the world.

In addition to ICTP, the AI Alliance founding members include CERN, ETH Zurich, Imperial College London, UC Berkeley, and the Simons Foundation.

"ICTP is pleased to be part of the high-level, international research community underpinning the AI Alliance that includes major players in the field," said ICTP Director Atish Dabholkar, adding, "We share a commitment to exploring new scientific horizons and to sharing new knowledge openly. AI has the potential to radically change the way science is done, and by joining this Alliance we will ensure that our large international network of scientists have open access to the latest in AI innovation, training and governance, regardless of geographical borders."

One of ICTP's key contributions to the Alliance will be sharing knowledge and resources about AI with developing countries. "For AI to benefit people and societies worldwide, the growing divide between the global South and North on the ability to develop open AI technologies must be reduced," said Sandro Scandolo, ICTP scientist and Senior Coordinator of the Centre's Research and Partnerships. "ICTP will expand its capacity-building efforts to ensure that scientists in the developing world stay up-to-date on the latest development of this rapidly evolving field." 

To that end, ICTP plans to host several international conferences, workshops, and schools dedicated to AI and related issues.  For 2024, these include an Advanced School on Applied Machine Learning, a Workshop on Machine Learning on Low-Power Devices, and a School and Conference on Frontiers at the Intersection of Quantum Simulation and Machine Learning.

ICTP will also continue to explore the use of AI in its physics research. The Centre's Quantitative Life Science section has used AI and Machine Learning in their research to teach mechanical gliders to soar like birds or to explain an unusual behavioural trait of vampire bats that could be transferred to cooperative robots. Researchers in ICTP's Science, Technology and Innovation unit have developed expertise in machine learning models that, instead of running on very powerful computers, run on very small and cheap microcontrollers, making the technology accessible to scientists in developing regions.