[https://docs.google.com/presentation/d/1kDH9ODSEzCnb2qYhz1fT10Vt1AqQQF0z/edit?usp=sharing&ouid=100090609942785485941&rtpof=true&sd=true](https://docs.google.com/presentation/d/1kDH9ODSEzCnb2qYhz1fT10Vt1AqQQF0z/preview?usp=sharing&ouid=100090609942785485941&rtpof=true&sd=true)
This talk details the philosophy and architecture of Perplexity Computer, a general-purpose orchestrator of frontier intelligence. The frontier AI model landscape is increasingly competitive, with average performance on industry benchmarks becoming nearly indistinguishable. But this belies the fractal complexity of each model's individual capabilities and specializations. Unlocking frontier-level performance for users increasingly demands seamless orchestration of work across multiple models, execution environments, and usage contexts. It will also require reimagining the nature of compute and I/O in personal computing’s next generation.
Jerry Ma serves as VP Global Affairs and Deputy CTO at Perplexity, partnering with government, civil society, and industry to advance democratic and universal AI. He concurrently serves on Perplexity’s engineering leadership team, shaping technical strategy and driving core R&D initiatives to better serve millions of users worldwide. Previously, Jerry was appointed as the first Chief AI Officer and Director of Emerging Technology at the U.S. Patent and Trademark Office, and served as an executive on detail to the U.S. Department of Justice's Antitrust Division.
Jerry has served on the National Science and Technology Council's ML/AI Subcommittee and the National IT R&D Program's steering committee. Jerry's scientific research has been published at numerous leading AI & computer security venues, and he is the recipient of a U.S. Presidential Rank Award and a Samuel J. Heyman Service to America Medal.