AI, Honestly

Cold Open

"We Are That Company's Product"

Anthropic's CEO says there's a 10-25% chance AI ends in catastrophe for humanity. He's building it anyway. His argument: the other guys are worse. Dry open. No music.

Segment 1

"The Family Tree"

The AI family tree: OpenAI (2015) to Anthropic (2021) to xAI (2023). Same company. Opposite exits. A few hundred people who disagree about everything — except that they should be the ones building the most powerful technology on earth.

Segment 2

"The Constitution Nobody Voted For"

Anthropic's 23,000-word constitution ranks your needs last. Explicitly. Kate reads the priority hierarchy like court documents. Morgan asks the question: who decided an AI might deserve moral consideration? The company that sells it.

Segment 3

"Machines of Loving Grace"

Dario Amodei's essay about what happens when AI exceeds human intelligence. Kyle reads it plainly: AI transforms things FOR humans. Not WITH. Not UNDER. Then the number — 25% chance of catastrophe. Russian roulette with two rounds in the chamber.

Segment 4

"The Next Manhattan Project"

The Fable 5 timeline: Project Glasswing, SK Telecom, the jailbreak, Dario's refusal, the Commerce Department export control. Kyle's history drop: the Manhattan Project produced a weapon and built the most controlled distribution regime in history. Anthropic produced a weapon and distributed it like a SaaS product. Plus: the All-In debate — Sacks, Chamath, J-Cal.

Segment 5

"The Question"

Nine episodes in. The pattern is visible. Morgan, Kate, and Kyle debate three frames: The Matrix (the choice was made upstream), the Amish (chose before the cost was real), and Kodak (saw the choice, had every advantage, chose wrong). Morgan closes: "Look up."