Making AI Reliable by Representing What It Does Not Know
Explore how explicit uncertainty, calibration, and Bayesian updating could make AI decisions more reliable, adaptive, and accountable.
Read the analysisResearch area
Google DeepMind: The Podcast
Transforming Learning with NotebookLM: AI Meets Human-Centric Design
The November 26, 2024 episode of the Google DeepMind podcast explores NotebookLM, a personalized AI assistant from Google Labs. Developers Raiza Martin and Steven Johnson share their insights into the product's development and features.
Read the analysis
AI: Supercharging Scientific Exploration
The October 9, 2024 episode of the Google DeepMind Podcast features Pushmeet Kohli discussing the revolutionary impact of artificial intelligence on scientific exploration.
Read the analysis
Unreasonably Effective AI: Insights from Google DeepMind's Latest Advances
On August 14, 2024, Hannah Fry talked to Demis Hassabis, the CEO of Google DeepMind, discussed the rapid advancements in AI, particularly through the Gemini project, and the unexpected effectiveness of AI models in generalizing concepts from language alone.
Read the analysis