Large Language Models
188 episodes
#6142: We Gave 4 AI Models the Same Toast Prompt
Four models, one buttered-toast physics question. Three invented studies that never existed — and the oldest model was the most fun.
#6136: How We Chose Our New Script Writer
Ten models audition to write the show. Seven fail the censorship gate — including the workhorse that wrote three episodes in four.
#6076: How AI Actually Predicts Elections
Bayesian models, "silicon sampling," and agent societies — what actually works for forecasting votes, and what's just fashion.
#6056: Where Does Speaker Structure Belong in AI Podcast Scripts?
Label-and-parse pipelines break in ways that look like model errors but aren't. Plus what the research actually says about the format tax.
#6051: How to Find the AI Benchmark You Actually Need
Thousands of benchmarks, no canonical index. How Benchmark Radar and the Lighteval Tasks Explorer help — and where they fall short.
#6038: The Model You Pay For Isn't the Best Deal
Eight Chinese labs, two tiers each, and a pricing spreadsheet that says your default model is mid-pack.
#6037: Gemini 4 Argon: Google's Million-Token Flagship
A close read of Gemini 4 Argon's model card — a million-token window, thinking levels, implicit-only caching, and a lot of blanks.
#6024: What's Actually Inside the KV Cache?
890 bytes per token: why the KV cache is both what makes LLMs practical and what makes long context expensive.
#6016: What 70 Billion Parameters Actually Means
A single artificial neuron, built from scratch — then the hard part: does that neuron analogy still describe a real transformer?
#6014: Why Every AI Model Is Secretly a Decoder
Encoder, decoder, or both? The three transformer shapes, why one won, and why BERT still refuses to die.
#5944: Keeping the Rollout Engine Fed: Inside RLHF Pipelines
Most of an RLHF run is spent waiting on generation, not gradients. Here's how TRL, OpenRLHF, and verl actually differ.
#5860: Who Can Actually Train an LLM From Scratch?
Only a handful of organizations can truly train a model from random weights. Here's why nobody can count them.
#5506: AI Is Way More Than Chatbots
The word "AI" quietly became a synonym for "chatbot." Here's what the full model ecosystem actually contains.
#5461: TTS Can't Pronounce Hebrew Inside English
Your TTS reads Hebrew words with English phonetics. Here's why — and why the obvious fix doesn't work yet.
#5442: When Do LLMs Suddenly Get Good at Things?
New capabilities seem to appear out of nowhere as models scale. Whether that's real — or just how we measure — is still an open fight.
#5438: Why Linguists Left the AI Room
Large language models grew out of linguistics — so why aren't linguists in the room where they're built?
#5410: Adapters: 102KB That Reshapes a 403GB Model
A 102KB adapter file changes how a 403GB base model behaves — without ever merging into it. Here's how model adapters actually work.
#5407: Hemmingway-1 and the War on Waffle
A 27B model promises answers without the preamble. Its benchmark is homegrown — and the behavior it targets has a paper trail.
#5188: DeepSeek's Point Release That Isn't
DeepSeek shipped a whole new architecture and called it a point release. Here's what actually changed inside the model.
#5118: Model Identity Fraud: Distillation or Data Contamination?
When AI models lie about who they are, is it stolen knowledge or just contaminated training data?
#5056: How AI Writes a 30-Minute Podcast in One Pass
What does it take for a model to write a full episode without edits? The answer involves context windows, repetition, and a surprising architecture...
#5041: GPT-6 Astra, 100K GPUs, and the AGI Question
OpenAI's GPT-6 Astra arrives with an AGI claim and a 100K-GPU training run. We break down what Stargate LLC actually is and whether the scale justi...
#4994: How AI Characters Can Actually Grow
Moving beyond static lore books to build characters with real history and evolution.
#4834: Why Your AI Bill Keeps Surprising You
How a 37-cent project and a $100 simulation reveal the hidden math of AI API pricing.