Azeem Azhar's Exponential View

Physicist and entrepreneur Steve Hsu, whose startup Superfocus tackles hallucination problems in large language models, joins Azeem to discuss AI agents, hallucination challenges and what happens when technology meets labor markets. They discuss: (01:31) The deeper shift that Superfocus represents (07:00)  Will models overcome hallucination? (10:15)  AI Agents can replace 80-90% of call center calls(12:27)  What it’s like showing customer support AI to customer support people (22:36)  China's mayors are like mini CEOs (30:05)  What will matter most in the supposed "AI race"? (35:58) DeepSeek was not part of the Chinese Government (38:23)  How open source will change the future of deployment (40:59)  What the public doesn't understand about AI tail risk (48:01) How AI plush toys can teach French to 2-year-olds This was originally recorded for "Friday with Azeem Azhar", a new show that takes place every Friday at 9am PT and 12pm ET. You can tune in through Exponential View on Substack. Produced by supermix.io and EPIIPLUS1 Ltd

Niall Ferguson, a renowned historian and geopolitical analyst, discusses the evolving U.S.-China relationship and its implications for the global order. He decodes Trump's foreign policy, emphasizing its assertiveness and possible voter backlash due to tariffs. Ferguson analyzes China's tech advancements and the significance of Taiwan in U.S. strategy. He also explores the shift from a liberal order to a multi-polar dynamics, highlighting how historical context shapes current geopolitics.

Ryan Petersen, CEO and founder of Flexport, shares his expertise on the fast-evolving landscape of global trade amidst ongoing tariffs and geopolitical tensions. He discusses how U.S. businesses are adapting to shifting manufacturing dynamics, and why shipping remains vital to everyday life. Petersen dives into the fragility of global trade, the impact of technology on logistics, and the urgent need for businesses to embrace agility. He also contemplates the future of globalization and what a potential U.S.-China trade deal might entail.

Packy McCormick, founder and investor at Not Boring, joins to explore the evolving tech landscape. He reflects on Web3's early enthusiasm and its challenges, questioning the balance between idealism and reality. The discussion includes Europe's awakening to technological needs and innovative energy solutions. Packy also delves into the future of nuclear power amid a solar-driven energy world, while examining how AI stands to redefine human value and self-expression in a rapidly digitizing society.

Jared Kaplan, co-founder and chief scientist of Anthropic, discusses the potential arrival of human-level AI in just 2-3 years, much sooner than expected. He highlights how Claude's reasoning capabilities are evolving, allowing AI to tackle complex tasks efficiently. Kaplan emphasizes the importance of constitutional AI and interpretability to ensure safety as models grow more powerful. The conversation also touches on the competitive landscape of AI development between the U.S. and China, and the ethical considerations essential for harnessing AI responsibly.

Kevin Kelly, co-founder of Wired Magazine and visionary futurist, shares his bold insights on the future of AI. He discusses the implications of declining populations on economies and reimagines a bot-to-bot economy where humans might even be paid to read emails. Kelly explores how AI could enrich human purpose and the value shifts that may arise from a scarcity of people. He offers a thought-provoking glimpse into a world where human attention becomes a prized commodity, highlighting the essential balance between technology and human connection.

Kai-Fu Lee, a leading AI researcher and entrepreneur from Taiwan, explores the future of AI in 2025, emphasizing the contrasting trajectories of the Chinese and U.S. markets. He highlights how Chinese companies are crafting unique generative AI solutions without predefined SaaS layers, leading to specialized integrations. Lee predicts a shift towards 'AI-first' applications across industries, with significant reductions in operational costs and an increased focus on inference over training. The discussion also touches on the evolving role of productivity tools and innovative search technologies.

Nathan Benaich, Founder and General Partner of Air Street Capital, discusses the future of AI in 2025. He highlights the explosive growth of frontier models and the rise of natural AI workflows, making interactions with technology more seamless. Despite advancements, the journey toward AI-designed drugs remains complex. Benaich also addresses the geopolitical implications of AI funding and infrastructure, noting the strategic significance of AI labs on a national scale. His insights reveal a landscape balancing incredible potential with nuanced challenges.

Dylan Patel, founder of SemiAnalysis, shares his expertise on semiconductors and AI infrastructure. He discusses the staggering rise in capital expenditures by major tech companies, predicting investments in AI-driven data centers could reach $500 billion by 2025. New specialized GPU cloud providers are entering the market, threatening the dominance of hyperscalers. Patel also highlights that while GPU shortages are notable, the true bottleneck lies in broader infrastructural complexities, crucial for future AI advancements.

In a captivating conversation, Michael Liebreich, founder of New Energy Finance and host of the Cleaning Up podcast, teams up with Azeem Azhar to tackle the future of energy. They discuss the rapid advancements in AI, solar, and battery technologies, highlighting how these innovations could combat climate change. The duo also delves into the challenges of hydrogen versus battery energy, the monopolization of the energy sector, and the rising energy demands of data centers. Their insights spark a dialogue about the societal implications of this energy transition.