Michael Burry, who famously predicted the 2008 financial crisis, expresses skepticism about the massive investments flowing into AI. He joins Jack Clark from Anthropic and podcaster Dwarkesh Patel in a discussion moderated by Patrick McKenzie. They debate whether AI is a legitimate breakthrough or a misallocation of resources. Burry reflects on how perceptions of AI have shifted since 2017, noting that while many once focused on artificial general intelligence (AGI), the emergence of large language models (LLMs) like ChatGPT has changed the conversation and driven huge financial commitments. Jack Clark recounts the evolution of AI development, highlighting the shift from training agents from scratch to leveraging large datasets for training models. The introduction of the Transformer model and insights from "Scaling Laws" led to significant advancements in AI capabilities. He emphasizes that today's AI is already better than anything that existed before, a point that many policymakers misunderstand. Burry adds that, contrary to his expectations, Google hasn’t maintained its lead in AI, and the spending surge began unexpectedly with ChatGPT, despite its limited use cases. Dwarkesh Patel points out that no single AI lab has maintained a lasting advantage, with major players like OpenAI and Google continually swapping positions of leadership. The conversation also touches on whether AI tools can significantly boost productivity. Patel questions the findings of a study showing a decrease in efficiency for developers using coding tools. Jack acknowledges the challenge of measuring productivity accurately, noting conflicting data and the necessity of a robust feedback loop for evaluating AI’s impact on development speed.