Governance guardrails for open models

Simon Milner, Vice President of Public Policy for Asia Pacific at Meta, joins us to explore how Meta's deliberate commitment to open source AI is reshaping innovation across the world's most diverse and dynamic region. He shares his journey from the BBC to nearly 14 years at Meta, where he built policy teams from the ground up to lead Meta's Asia Pacific strategy. Simon unpacks Meta's open source philosophy behind the Llama models, explaining how openness accelerates innovation through community scrutiny, provides governments greater control over sensitive data, and enables local developers to fine-tune models for languages like Korean, Vietnamese, and Bahasa Indonesia. He highlights compelling use cases across the region in Japan and Korea. Looking ahead, Simon reveals why the future of AI is not on our phones but in wearables like AI-enabled glasses that create always-on assistants seeing what we see and hearing what we hear, enabling us to be more present in the world while Meta supercharges its family of apps serving billions globally. Last but not least he shares what great looks like for Meta in the Asia Pacific on open source AI.

"We believe that openness is actually a really key feature of accelerating innovation because it fosters inclusion, it builds trust, and it ensures that the benefits of AI are more evenly distributed around the world.The openness of models allows other people to, as they were, push and pull and prod at the models at a fundamental level in order to see where might the problems be. And so that kind of community, the developer community scrutiny around open source is fundamental to spotting issues and addressing them quickly.Actually, the story of AI is about yes... that is important. The investments that companies like Meta and others are making is important, but actually, it's really about local ownership and local innovation." - Simon Milner

Episode Highlights: [00:00] Quote of the Day by Simon Milner from Meta[01:37] Simon's journey: BBC, BT, Meta's 14-year evolution[03:12] Navigating diverse regulatory landscapes across global markets[05:24] Career advice: Take risks, embrace unexpected opportunities[07:54] Open source AI democratizes access and innovation[10:21] Meta sparked open model trend, others followed[14:49] Open models enable faster innovation through community[16:21] Government control and data sovereignty with open[19:13] Governance mechanisms: transparency, red teaming, community engagement[22:49] Meta learned responsible AI through 20 years experience[25:49] Singapore, Japan, Korea developers using Lama locally[28:26] AI isn't just big companies and includes local innovation[31:15] Keeping AI open prevents fragmented national bubbles[34:01] Governments balancing open innovation with national interests[37:00] Future AI: wearables and glasses, not phones[38:19] Always-on AI assistants seeing and hearing you[41:35] Supercharging Meta apps and building new products[42:00] Closing

Podcast Information: Bernard Leong hosts and produces the show. Proper credits for the intro and end music: Energetic Sports Drive and the episode is mixed and edited in both video and audio format by G. Thomas Craig. Visit our Analyse main site: https://analyse.asia