- Kalshi’s trading volume rocketed from hundreds of millions per month in early 2025 to $18 billion in May 2026 — but at the annual Manifest conference in Berkeley, the rationalist intellectuals who championed prediction markets as a civilizational knowledge tool expressed widespread dismay that sports betting now dominates both Kalshi and Polymarket.
- Robin Hanson (“godfather of prediction markets”), Scott Alexander (Astral Codex Ten), and Eliezer Yudkowsky — thinkers whose ideas influenced Sam Altman, Dario Amodei, and others — fear the platforms’ pivot to sports prop bets and “mention markets” undermines their public benefit potential and is building long-term regulatory risk if Democrats return to power.
- AI forecasting poses an existential challenge to the model: a Preseen engineer claims to have built a bot that turned $35 into $2 million trading prediction markets, and Future Search co-founder Dan Schwarz argues “AI is plausibly beating the entire market now” — raising the question of whether human forecasters retain any meaningful edge.
- Yudkowsky raised the field’s most fundamental concern: there’s “a problem with resolving a prediction market when one of the outcomes is that all paid participants are dead” — pointing to AI existential risk as the ultimate limit case for markets premised on human survival.
What Happened?
The Manifest 2026 conference — the annual gathering of prediction market enthusiasts held at Lighthaven in Berkeley — drew 900 attendees and surfaced a growing schism. Platforms like Kalshi and Polymarket have seen explosive growth, with Kalshi alone hitting $18 billion in monthly trading volume in May. But the majority of that volume is sports betting, not the geopolitical, scientific, or policy forecasting the prediction market purists envisioned. Robin Hanson presented on “futarchy” — using prediction markets to guide government and corporate decisions; Scott Alexander said he was “disappointed that sports gambling has become such a prominent part”; Eliezer Yudkowsky, wearing a gold sequined jacket, dismissed sports results as “sportsball” mid-panel. An AI company, Preseen, claimed its bot turned $35 into $2 million; Future Search argued AI bots are outperforming human traders on some market types already.
Why It Matters?
Prediction markets’ original promise was epistemic: a market mechanism that aggregates dispersed information better than any individual expert. That promise still holds in theory, but financial incentives are pulling platforms toward sports and celebrity prop bets, which generate volume without contributing to societal knowledge. The CFTC is currently defending the platforms against state gambling regulators, but that protection depends on the Trump administration. Veteran traders at the conference warned that sports-betting dominance is building long-term regulatory vulnerability. Meanwhile, if AI bots are already outperforming the crowd, the fundamental premise — that human insight is what makes these markets accurate — comes into question.
What’s Next?
AI forecasting is likely to reshape prediction markets in two ways: by improving accuracy (more bots → more liquidity → better price signals) and by potentially commoditizing the human forecaster’s edge entirely. Scott Alexander envisions a future where AI bots bet on thousands of narrow questions — weather at specific locations, local policy outcomes — that are currently too thinly traded to be accurate. “AI lowers the cost of doing that to cents, and then you have all the questions you ever need.” Whether Kalshi and Polymarket evolve toward that vision of genuine epistemic utility — or remain sports betting platforms with a rationalist origin story — is the defining question for the sector.
Source: Bloomberg Businessweek
