Paradigm Shock

This episode confronts the frontier model vs. open-weight debate and the economics driving model choice. Kamdar argues that models are increasingly commoditized for routine enterprise uses; open-weight options now offer competitive capability at far lower cost, shifting procurement toward model routing and hybrid architectures. He also challenges narratives about AI-triggered layoffs, citing evidence that AI-native firms often grow headcount while vastly increasing throughput. The episode closes with a forward-looking warning and playbook: companies that pursue an offensive strategy of adopting the right platforms, observability and context engineering will outpace those that cling to protectionism or paralysis.Show LinksThe Practical Work Benchmark Report - https://www.elvex.com/resources/practical-work-benchmark-reportStudy of 3.8 million messages to understand how Prompt Engineering is changing - https://www.elvex.com/resources/context-engineering-replacing-prompt-engineeringelvex website: https://www.elvex.com/Sachin Kamdar on LinkedIn - https://www.linkedin.com/in/kamdar/elvex on LinkedIn - https://www.linkedin.com/company/elvexai/Sponsor PageCyberEDX.com Contact [email protected]

Alvin Wang Graylin dismantles the dominant AGI race narrative, arguing much of it is a market‑manufactured scramble driven by capital, narrative hype and short‑term valuations. He explains the enormous CapEx overhang with trillions sunk into data centers and GPUs while open source models and regional labs close capability gaps at a fraction of the cost. The episode shows why frontier labs’ growth projections are unrealistic. Alvin calls for a reorientation of public policy away from protecting a few winners toward broader economic and infrastructure priorities.The conversation also covers the upcoming U.S.–China AI risk and safety talks where Alvin serves as an advisor. Then the conversation moves to Alvin's proposed reframing of AI risk and governance itself: smaller, specialized models running on consumer hardware can be more dangerous in widely deployed contexts than large frontier models. Alvin and Anjon map how the orchestration layer enables durable advantage and multiplies threat vectors, especially in cyber, chemical, and bio domains, arguing that regulating only large training runs misses the dispersed reality of risk.Bio: Alvin Wang Graylin is a Digital Fellow at Stanford’s Human-Centered AI Institute, Senior Fellow at the Asia Society Policy Institute Center for China Analysis, Professor of AI and Technology Policy at the University of Washington, and author of Our Next Reality. With more than 35 years of experience across AI, semiconductors, immersive computing, and cybersecurity, he brings a rare cross-cultural perspective from extensive work in both the United States and China.Show LinksBeyond Rivalry: A US-China Policy Framework for the Age of Transformative AI The Enterprise AI Playbook: Lessons from 51 Successful DevelopmentsThe Biggest AI Models Are Not The Biggest Threats in Cypher BriefThe Abundanist SubstackAlvin’s LinkedIn PageSponsor PageCyberEDX.comContact [email protected]

This conversation blends philosophy, product, and enterprise realities. Karl frames AI as an extension of human creativity and flourishing, urging leaders to ground governance in purpose and the humanities. He unpacks concrete deliverables from governance engagements such as configured workflows, knowledge graphs, control libraries, and integrations with cloud developer platforms, and highlights unsolved problems like reliably measuring agent ROI and telemetry. The episode closes with a humane reminder: translate values into code to govern in real time, then use the gains from AI to spend time on what matters most.Bio: Karl Herbert Grabbi is the Director of Global Advisory Services at Credo AI, where he leads a team of forward-deployed AI governance engineers working inside enterprises to operationalize trustworthy AI. His work sits at the intersection of innovation, risk management, and human judgment, built on the belief that durable AI governance has to be as much about values and craftsmanship as it is about frameworks and code.Karl’s LinkedIn Page https://www.linkedin.com/in/karl-herbert-grabbi-cfa-7a531058/Credo AI Website https://www.credo.ai/Sponsor PageCyberEDX.comContact [email protected]

The first in person video recording of a Paradigm Shock episode, now available on the new YouTube Channel.In a thought-provoking discussion on Paradigm Shock, University of Notre Dame Professor Dr. Joanna Santos explores the significant intersection of AI and cybersecurity. The episode sheds light on the capabilities of Anthropic's Mythos and other advanced models, an AI that can identify and patch vulnerabilities autonomously but is kept under wraps to prevent misuse. Santos emphasizes the shifting dynamics in cyber threats, as AI speeds up the detection and exploitation race. She also touches on the potential role of quantum computing in cybersecurity and the remarkable journey that has brought her to the forefront of this rapidly evolving field. They finish with thoughts on the World CupNote:Recorded on June 4th 2026YouTube Recording: https://www.youtube.com/watch?v=d10n0oFnlE8&t=378sConnect Joanna Santos GitHub page: https://joannacss.github.io/ Notre Dame Page: https://engineering.nd.edu/faculty/joanna-cecilia-da-silva-santos/ Google Scholar Page: https://scholar.google.ca/citations?user=mkGmYyAAAAAJ&hl=en&oi=sra Notre Dame S2E Lab: https://s2e-lab.github.io/members/Subscribe to Paradigm Shock on YouTube: / @paradigmshock2000 Show Contact Show [email protected]

Charlie Bullock from The Institute for Law and AI joins Anjon Roy on Paradigm Shock to delve into the challenges of AI governance amidst high uncertainty. Highlighting the radical optionality framework, Bullock advocates for developing government capabilities to respond flexibly to AI advancements. They explore policy suggestions like transparency laws and evaluation mechanisms as crucial tools for managing AI's rapid progression. By prioritizing institutional readiness over fixed regulations, the dialogue points to a necessary shift in political focus to adequately prepare for AI's potentially revolutionary impacts on society, even when the exact outcomes remain unpredictable.Show LinksRadical Optionality Paper: https://law-ai.org/radical-optionality/Charlie Bullock LinkedIn: https://www.linkedin.com/in/charlie-bullock-79169b168/Ad LinkRegistration Link for Pause AI - Oxford Style Debate in Chicago on June 10thhttps://luma.com/ais-chicagoContact [email protected]