// podcast · season 1 · episode 12026-06-03~ 23 min

S1E1 — The Five-Year Bet

Two informed people, opposing reads of where AI lands on January 1, 2030. Concrete predictions with dates and probabilities. We come back in 2030 and grade ourselves.

About this episode

This is our first episode. Season 1 launches today with the question that everybody asks but nobody answers concretely: where is AI on January 1, 2030?

Host Elias Mercer and guest Orion Vale draw a line on the calendar and put real claims on either side of it. Elias is the curve-bender — he thinks the second-order effects of AI accelerating AI research kick in inside this window. Orion is the deployment realist — he watches Fortune 500 companies try to actually put these systems into production and he has receipts on what's working and what isn't.

They disagree about almost everything. They commit to predictions with dates and probabilities. And at the end they shake on the actual bet — including stakes neither of them wants to lose.

What we cover

The show

The Five-Year Bet is a biweekly podcast. Each episode is one guest, one question, one date on the calendar — and concrete predictions that we come back and grade ourselves on. Season 1 runs for 24 episodes ≈ one calendar year of conversations. Then Season 2 begins.

References cited in this episode

Karpathy on Dwarkesh — AGI is still a decade away (Oct 2025) → · Artificial Analysis — GPT-5.5 leads Intelligence Index (2026) → · 80,000 Hours — Will we have AGI by 2030 → · Arcade.dev — State of AI Agents 2026 → · agentmarketcap.ai — Enterprise Agent Deployment Maturity Model 2026 → · Sara Hooker — On the Slow Death of Scaling →

Subscribe

Drop the RSS feed into any podcast app: /podcast/feed.xml. New episode every two weeks.