AI Alignment
317 episodes
#6197: Talking to Your Own Voice Clone
What actually happens when you sit down and talk with an AI clone of your own voice — and what Evan Ratliff's Shell Game revealed.
#6145: Do LLMs Nod Along With Conspiracy Theories?
A 98-prompt eval asks one question: when a believer brings a conspiracy theory, does the model push back, or help them dig?
#6143: Propaganda Bait: Testing 7 Models in 5 Languages
Five prompts, seven models, five languages — a first look at a benchmark that doesn't exist yet, testing what LLMs do with government talking points.
#6137: Two Ways a Chinese Open Model Gets Censored
Same weights, different landlord: how to tell a serving-time filter from censorship baked into the model itself.
#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.
#6129: How to Build an AI Preference Dataset
Pairwise comparison, forced choice, or just "this one or that one" — the engineering behind collecting human preferences on AI outputs.
#6126: Can You Train a Machine to Be Funny?
Comedy fights everything next-token prediction is built for. Turns out you can train a model to land a joke — and then measure it.
#6119: Two Keys and a Polite No: Labeling a Claude Code Approve Button
Daniel has two spare laptop keys and wants them to say yes and no. The electronics are easy — it's the labels that get hard.
#6112: The Interpreter That Doesn't Know Who's Talking
Voice-to-voice translation is basically solved. The part where the machine knows who's in the room and who holds the power? Nobody's shipped it.
#6110: The Chatbot OpenAI Almost Shelved
Instruction models existed for months. Then a rumour about Anthropic pushed OpenAI to ship a chat box in two weeks.
#6103: When AI Picks Your Spouse, Stocks, and Eggs
From $31 eggs to a ChatGPT-officiated wedding, people are handing real life decisions to machines — and publishing the results.
#6101: When Chatbots Run for Office
From a Wyoming mayoral bot to Albania's AI minister, six real attempts to put machines on the ballot — and what they reveal about empty platforms.
#6092: 到底是谁在塑造中国的人工智能思想？
不只是实验室的老板：那些定义中国如何打造、又如何争论人工智能的研究者和思想。
#6071: Outside the English Bubble: Where to Find Dubbed Asian Documentaries
There's no Netflix for dubbed world documentaries. But AI dubbing on Bilibili and YouTube is about to unlock the part of a country that talks to itself.
#6068: Chinese AI Models: Refusing vs. Reframing
New studies show Chinese AI models have shifted from refusing sensitive questions to answering them — in the Party's voice.
#6062: Who Actually Shapes Chinese AI Thinking?
Beyond the lab CEOs: the researchers and ideas defining how China builds — and argues about — artificial intelligence.
Also available in:中文
#6059: Ten Documentaries for Daniel: AI, Power, and Who Holds the Levers
A hand-picked slate of 2026 documentaries on AI anxiety, platform power, intelligence, and media dynasties — plus where to actually watch them.
#6036: GPT-6 Astra: A Million Tokens, No Fine-Tuning
A million-token window, ten built-in tools, a 272K pricing cliff — and no fine-tuning. What GPT-6 Astra actually offers builders.
#6034: Claude Opus 5.5: What Anthropic Actually Shipped
Anthropic's new flagship is cheaper than the model it replaces — and won't let you turn thinking off.
#5989: Twenty Tiny Beats That Make a Podcast Stick
Twenty novel recurring segment pitches — and the test each one has to pass: does it add context, or just noise?
#5972: AI Meeting Agents Are Becoming Stand-Ins
Bots now book follow-ups, update CRMs, and attend on your behalf. Here's what's real, what's speculative, and where the etiquette is heading.
#5946: Inside Humanity's Last Exam: 2,500 Questions AI Can't Ace
2,500 questions, ~1,000 experts, and a top score that jumped from 2.7% to 54.8% in 18 months. What is HLE actually measuring?
#5945: How a Reward Model Actually Trains an LLM
One number, no gradients, four models in the room. What a reward model actually does inside the RLHF loop.
#5873: Abliteration: How $5 Undoes a Model's Guardrails
A rank-one weight edit costs under $5 and strips refusal behavior from open models. Then an automated tool beat human experts at it.