Carnegie Mellon University (CMU) is launching the Open Forum for AI (OFAI) to provide holistic, objective expert advice regarding AI, particularly as it relates to research, technical prototypes, policy recommendations, community engagement, and inclusion. The overarching vision is that AI should augment human capability rather than replace it; that AI development needs to be human centered. Leveraging CMU’s AI expertise and capacity, OFAI will include partners that offer a range of complementary expertise necessary for collective action for AI strategy and associated policy development.

Founding Reason

OFAI’s launch signals a powerful call for deeper technical understanding and more rigorous scrutiny in AI, including but also extending beyond open source software. OFAI will advocate in an objective manner for wider involvement in the AI ecosystem to shape policy-making and address issues such as AI safety, privacy, and equity, as underscored in memoranda like the White House AI Executive Order. There is growing consensus that an open AI ecosystem will mitigate risks while fostering innovation and inclusion, but little clarity about the characteristics of such an ecosystem. OFAI aims to address this gap in a comprehensive and inclusive manner.

One current narrative is that AI can be executed only with large-language models and massive amounts of data and computational power; that is, with technology infrastructure available only to a certain few. A corollary of this narrative is that AI is too complex to be examined transparently. Within this scenario, the default may become large technology companies self-regulating, with small and medium enterprises and underrepresented groups finding it difficult to participate in the AI ecosystem.

OFAI will augment this narrative by conducting research, making policy recommendations, developing outputs including AI technical prototypes, and convening diverse groups such that AI becomes a multi-stakeholder effort with facets at different levels of granularity. At the crux of these efforts will be the technical framework that outlines and defines openness of AI based on both artifacts and orchestration of those artifacts.