Latest Episode
Experimental Selected episodes are also translated into 8 other languages. Other languages →
Meet the Hosts
From the blog · Behind the scenes What One Episode Costs to Make About 27 cents with a premium model writing the script, about 16 cents with a cheaper one. Where the money goes, as of 31 October 2026, and how much a single sponsor could take off. 
#6502: What a Cross-Entropy Loss of 1.8 Actually Means
A cross-entropy loss of 1.8 nats means a language model is about as uncertain as choosing uniformly among six equally likely tokens.
#6501: Splitting Frontier AI Across Thousands of GPUs
A 405-billion-parameter model requires 810 gigabytes just for weights, yet an H100 GPU holds only 80.
#6500: Why LLM Leaderboards Need Psychometrics
Applying item response theory to AI benchmarks reveals up to 31 percent of leaderboard positions shift when raw accuracy is adjusted for difficulty.
#6499: Why Simple Rules Beat Value Models in Reasoning RL
DeepSeek R1-Zero boosted AIME accuracy from 15.6% to 77.9% using only two deterministic checks and no learned critic.
#6498: Why You Must Read AI Traces Before Writing Evals
Reading just thirty real interaction traces by hand reveals failure modes that generic automated evaluation metrics consistently obscure.
#6497: Why Speech Recognition Confidence Scores Lie
Cross-entropy loss rewards saturation, leaving speech recognition models to assign winning tokens over 0.9 probability even when they are guessing.
#6496: How Speech Models See Audio as an Image
Modern speech recognition models like Whisper never hear raw audio; they inspect an eighty-pixel-tall grayscale image of the sound.
#6495: Teaching Speech Models to Choose Their Own Audio
Selecting the most uncertain speech recordings often just surfaces silence, clipping, and abandoned half-sentences.
#6494: Where Speech Models Actually Learn
Just three self-attention heads out of twenty in Whisper's decoder trigger over 75 percent of hallucinations on non-speech audio.
#6493: Why Speech Models Forget Other Voices When Tuned on Yours
Adapting Whisper to one person's voice can trigger catastrophic forgetting, stripping away timestamps, segmentation, and accuracy on other speakers.