Machine Learning Street Talk
Summary
Machine Learning Street Talk covering AI Benchmarking, AI Alignment, AI Evaluation, and Time Horizons Benchmark. Notable guests include Beth Barnes (AI Alignment and Capabilities Researchers), David Rein, Dan Hendrycks (AI Safety Researcher), and Blaise Agüera y Arcas. Episodes span from Aug 2025 to May 2026.
Current AI evaluation methods are critically flawed, with issues like data contamination and shortcut learning creating a disconnect between high benchmark scores and real-world utility. The 'Time ...
The speaker presents an artificial life experiment (BFF) that demonstrates a phase transition from a non-computational state to a complex, self-replicating, 'living' state. The primary driver of no...
Current AI models, including large language models, have fundamental architectural limitations, failing at basic algorithmic tasks like addition with a 'carry' operation, which cannot be solved by ...