Microsoft launches a new initiative to accelerate foundational AI model research.

Microsoft announces the launch of Accelerate Foundation Models Research (AFMR), a global initiative aimed at fostering interdisciplinary communities and advancing responsible AI progress. "Having a diverse research ecosystem is crucial to fulfilling the promise of AI. AFMR aims to broaden the use of powerful models by involving academics from outside the field of computer science, in pursuit of wide-ranging opportunities," said a spokesperson. AFMR will provide global academic researchers with access to advanced foundational models through Microsoft Azure AI services. The initiative seeks to expand access beyond computer scientists in technology companies by leveraging diverse expertise from different disciplines, cultures, and languages. This breadth of perspective is expected to enrich AI research and align models with shared human values. It may also lead to applications of AI in fields such as education, healthcare, law, and scientific discovery. "A vibrant and diverse research ecosystem is crucial to realizing the promise of AI," said Peter Lee, Corporate Vice President of Microsoft Research and Incubation. AFMR will pursue three main objectives: 1. Enhancing the alignment of AI with shared human goals, values, and preferences through model research, while improving security, stability, sustainability, responsibility, and transparency. This includes the ability to quickly measure progress through new evaluation methods. 2. Improving human-to-human interactions, increasing trust, wisdom, creativity, and productivity, reducing the digital divide, and mitigating the risks of developing AI that is not beneficial to individuals and society through socio-technical research. 3. Accelerating scientific discoveries in natural sciences through active knowledge discovery, hypothesis generation, and multi-scale, multimodal data generation. This initiative builds upon a successful pilot project conducted earlier this summer, involving researchers from various fields, including English language learning and astronomy. Multiple research projects have already benefited from this initiative. For example, the Korea Advanced Institute of Science and Technology (KAIST) is integrating ChatGPT into English as a foreign language writing education. Stanford University is utilizing AI for medical applications, while researchers from North Carolina Agricultural and Technical State University and Harvard University are applying AI to traffic monitoring and astronomy research, respectively. AFMR represents the evolution of industrial research strategies. Microsoft is not only providing access but actively building an inclusive community to explore the potential of AI. This reflects the notion that diverse voices are crucial for responsible AI development. The response to the initial proposal of AFMR has been overwhelmingly positive, with hundreds of proposals pouring in from universities around the world. This enthusiasm highlights the global academic community's eagerness to fully harness the potential of AI in various fields. Through funding, conferences, and other resources, AFMR aims to break down barriers between academia, industry, and the public. This collaborative spirit may pave the way for new best practices in AI research and development. "We're not just talking about models and algorithms; we're crossing boundaries to address some very significant problems," said Lee. "Our goal is to ensure that AI is not just serving a few, but becomes a change agent for all of us."