In this episode, Herman and Corn unpack the groundbreaking news of Alithia, Google DeepMind’s new agent capable of autonomous mathematical research. They explore the technical shift from simple pattern matching to "System 2" deliberative reasoning, explaining how "test-time compute" allows models to "think" through complex proofs before they speak. Beyond the tech, the duo discusses the "brain drain" from universities to corporate labs, the rise of independent institutes like Mila and AI2, and why we should be skeptical of vendor-led benchmarks. Is this the end of the human mathematician, or just a powerful new tool for discovery? Tune in to find out how the frontier of AI research is being rewritten.
