GTC Live Washington, D.C. Pregame - Chapter 2: Agentic AI for Every Industry

The GTC Live Washington, D.C. Pregame podcast explores the transformative potential of agentic AI across various industries, emphasizing America's leadership in AI and its global technology deployment. The discussion anticipates millions of AI agents enhancing productivity and efficiency in diverse sectors.

Eravon Srinivas from Perplexity envisions their browser, Comet, as a personal assistant that boosts user engagement by allowing questions from any platform, with future advancements in AI capabilities like task booking. Scott Wu of Cognition reports rapid growth in coding agents for enterprises, balancing AI hype with tangible productivity gains in software engineering.

Shiv Rao shares insights from healthcare, highlighting the slow pace of operations and clinician burnout, emphasizing AI's potential to alleviate clerical burdens and enhance patient care. George Kurtz discusses the complexities of security in the AI era, stressing the need for integrated solutions and effective data management, while warning that AI also democratizes access to sophisticated tools for adversaries.

The podcast debates the future of healthcare documentation, focusing on clinician-patient interactions and compliant billing documentation. Scott reflects on historical shifts in programming tools, questioning the future role of developers. The evolution of consumer experiences through technology, particularly in booking services via conversational agents, is also discussed, highlighting the need for agents that understand user preferences.

Speakers address the advantages of web access and advanced reasoning models to enhance user experience, envisioning mobile apps that allow task delegation through voice commands. They emphasize the importance of asynchronous operation for browser-based agents, illustrating their potential to multitask effectively.

Security concerns in AI are raised, questioning whether larger companies have an advantage due to data scale and the potential for startups to disrupt the security space. The conversation shifts to how Washington can enhance national security through improved procurement processes and public-private partnerships, advocating for forward-thinking technologies.

Insights on the cost of AI inference suggest potential reductions, with companies exploring strategies to lower these costs through collaborations and fine-tuning open-source models. The need for advanced reasoning models is acknowledged, despite their high operational costs, as they significantly improve user experiences.

Shiv emphasizes the challenges in healthcare, advocating for strategies that include model distillation and fine-tuning while addressing privacy and security complexities. The potential of agentic systems to replace traditional operating systems is highlighted, with Aravind pointing out the limitations of current mobile operating systems and the critical role of browsers in accessing third-party applications.

Aravind speculates on future hardware innovations, such as headsets and goggles, while acknowledging the continued importance of web access and browsing tools. He briefly mentions significant updates from OpenAI and Microsoft in the healthcare sector, underscoring the need to stay informed about industry advancements.