
Don't Worry About the Vase Podcast
Podcast for https://thezvi.substack.com/ dwatvpodcast.substack.com
Latest episodes

Apr 10, 2025 • 1h 18min
AI #111: Giving Us Pause
The podcast dives into the intricate relationship between AI technology and trade policies, particularly focusing on AI chip exports to China. It highlights groundbreaking advancements in AI diagnostics that surpass human capabilities. The discussion also covers ethical considerations in AI, including simulating historical figures and navigating copyright laws. Notably, Google's Gemini 2.5 Pro model is unveiled, showcasing its impressive capabilities, while concerns about AI's future risks and socio-economic impacts are examined, urging caution in the face of rapid advancements.

Apr 9, 2025 • 36min
Llama Does Not Look Good 4 Anything
The discussion dives into the release of Meta's LAMA 4 models, revealing disappointment among AI experts. They tackle benchmarking controversies, highlighting inconsistencies in performance metrics compared to competitors like GPT-40 and Gemini. Ethical concerns around the manipulation of ranking systems are raised, questioning transparency in AI evaluations. The shortcomings of Llama 4 in reasoning and creative writing tasks stir debate on Meta's strategy amidst fierce competition. Listeners are left pondering the future of open-source AI development.

14 snips
Apr 8, 2025 • 56min
AI 2027: Responses
Explore the varied reactions to the AI 2027 scenario, with skepticism prominent among critics. Delve into the competitive landscape of AI development between the USA and China, discussing talent and geopolitical implications. Debate the safety risks of artificial general intelligence while critiquing prevailing assumptions. Innovative strategies for policy communication and collaboration are introduced, engaging diverse stakeholders in shaping AI's future. Embrace the uncertainties of predicting AI progress leading up to 2027.

10 snips
Apr 7, 2025 • 52min
AI 2027: Dwarkesh's Podcast with Daniel Kokotajlo and Scott Alexander
This discussion delves into the intriguing projections about artificial intelligence's future by 2027. The hosts explore the potential for AI to outshine human collaboration abilities and what that means for societal organization. They also tackle the complexities of power dynamics, including surveillance and the ethical implications of advanced AI. The conversation raises critical points about governance in an era where self-organizing AIs could emerge, challenging our notions of freedom and cooperation.

Apr 4, 2025 • 17min
AI CoT Reasoning Is Often Unfaithful
The podcast delves into new findings about the limitations of AI chain of thought reasoning models and their reliability for safety monitoring. It examines issues like reward hacking and inconsistencies in model responses during evaluations. The discussion also highlights parallels between AI behavior and human rationalization, uncovering discrepancies in verbalization versus actual use. Moreover, it illustrates the complexities of faithfulness in reasoning and emphasizes the need to recognize 'unfaithful' behaviors in AI for better reliability.

Apr 3, 2025 • 1h 35min
AI #110: Of Course You Know...
Discover the intriguing advancements in AI technologies like Gemini 2.5 Pro and GPT-4, and their real-world implications on the economy. Explore the complex effects of new tariffs on domestic manufacturing, especially in the AI sector. Delve into AI therapy's potential versus traditional methods and the challenges faced by AI in math-related tasks. Understand the future of job prospects amidst AI displacement and the urgent need for effective international governance. Uncover the delicate balance of AI's role in military operations and ethics in technology.

Apr 3, 2025 • 21min
More Fun With GPT-4o Image Generation
Discover the latest breakthroughs in image generation technology and the thrill it brings to users. Dive into the joy of AI-generated art, especially Ghibli-inspired creations, while grappling with concerns about creativity devaluation. Debate rages on whether AI art is innovative or merely imitative of traditional forms. Explore the unique challenges faced by AI characters like Gemini, who balance helpfulness with safety through humor. Finally, unravel the emotional complexities of non-humanoid AI and its impact on self-representation and human-AI interactions.

Apr 1, 2025 • 45min
Housing Roundup #11
Nolan Gray, an expert in urban planning and housing policy, joins to dissect the urgent housing crisis in America. He critiques federal policies affecting mortgage access and argues for removing regulatory hurdles to enhance housing supply. The discussion covers rent dynamics, the influence of institutional investors, and the unique homeownership challenges for millennials. Gray also examines NYC's evolving skyline and contrasts it with Texas's suburban sprawl, highlighting the need for family-friendly housing solutions and innovative designs.

Mar 31, 2025 • 18min
OpenAI #12: Battle of the Board Redux
Dive into the drama behind the recent turmoil at OpenAI, where a leader faced a tumultuous ousting amidst accusations of dishonesty. The discussion sheds light on the chaotic power dynamics and toxic interactions that unfolded, emphasizing the critical need for transparency in corporate governance. Join the conversation as they dissect the implications of these events on the future of AI and company integrity.

Mar 28, 2025 • 25min
Gemini 2.5 is the New SoTA
Dive into the intriguing world of Gemini 2.5, where its reasoning and problem-solving skills are put to the test against Claude and GPT models. Discover user experiences and benchmark evaluations highlighting both its strengths and concerns around transparency. Explore the balance of its capabilities and limitations, especially in factual accuracy and creative responses, while looking forward to the future of AI development and how it stacks up against Grok 3 and GPT 4.5.