Unsupervised Learning cover image

Unsupervised Learning

Ep 58: Google Researchers Noam Shazeer and Jack Rae on Scaling Test-time Compute, Reactions to Ilya & AGI

Mar 17, 2025
Noam Shazeer, co-inventor of the Transformer, and Jack Rae, Research Director at DeepMind, dive into the future of AI. They discuss the groundbreaking capabilities of Gemini 2.0 for reasoning and creativity. The duo explores the complexities of AI evaluation metrics and the evolving role of test-time compute, emphasizing efficiency over traditional methods. They also reflect on the philosophical challenges of AGI, the rise of vision-based models, and AI's transformative impact on healthcare and education, highlighting the balance between innovation and safety.
01:09:28

Podcast summary created with Snipd AI

Quick takeaways

  • Test time compute is pivotal for AI development, impacting model performance and improving reasoning and creative writing capabilities simultaneously.
  • The need for meaningful and evolving evaluations in AI modeling is critical, as traditional benchmarks become less relevant as technology advances.

Deep dives

The Role of Test Time Compute in AI Advancement

Test time compute is identified as a crucial aspect of AI development with significant implications for how models operate and improve. The discussion highlights a focus on reasoning tasks, such as math and coding, which initially guided the development of Gemini models. A key surprise was discovering that enhancing thinking processes during training not only benefited reasoning tasks but also improved performance in creative writing, allowing the model to produce entertaining and meaningful content. This multidimensional capability suggests that the integration of reasoning and creativity can lead to more versatile AI applications.

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
App store bannerPlay store banner