Generative AI

Generative AI as a term is heard on ToTheMoon mostly in the 2024 episodes, when the hosts and their guests discussed not the models but the market around them: where the money goes, who has any defense against copying, and where the technology already pays for itself. The concept is treated pragmatically — as a wave like earlier technological revolutions (episode 011), to be tested against revenue, unit economics and real workflows.

In the spring of 2024 episode 002 recorded that GPU sales far exceeded the revenue earned with them, yet that generative AI would soon do routine work better than ninety percent of people. By summer the reading had shifted: episode 008 said the premium for the words generative AI in a pitch deck was gone, a thin wrapper over someone else's API had no defensibility, and investors were looking for companies with real revenue. Episode 018 singled out only Roblox among small listed companies. Specific applications got their own treatment: customer support (episode 014) and healthcare (episode 016).

Skepticism about generative search was spelled out in episode 017: the user is better off, but the unit economics break on expensive model calls. In 2025 episode 057 took stock: no new business models or large enterprise stories had yet been built on generative AI. By 2026 the angle had shifted from the market to labor: in episode 151 Anna Volchek, drawing on research from Stanford, NYU and MIT, described the invisible layer of work around generation — checking, rechecking, coordination and follow-up control — that had never existed at this scale before.