AI 2025 → 2026 Live Show | Part 2

What It Covers

Part 2 of Cognitive Revolution's AI 2025-2026 live show features New York Assembly member Alex Boris on the RAISE Act, Dean Ball on AI policy and China chip exports, and forecaster Peter Wilderford on AI progress predictions and economic impacts.

Key Questions Answered

• •State AI Regulation Under Fire: New York's RAISE Act requires AI labs to publish safety plans and disclose critical incidents when models fail internal tests for extreme risks like bioweapon creation or autonomous crimes causing 100+ deaths or $1B+ damage, but faces opposition from $100M super PAC funded by Andreessen Horowitz and Greg Brockman targeting sponsors.
• •Rent Don't Sell Chip Strategy: Instead of selling H200 chips to China outright, policymakers should rent compute access through Malaysian or Singapore data centers controlled by US companies like Oracle, allowing economic benefits for Chinese civilians while maintaining ability to shut off access during military conflicts or misuse scenarios.
• •Revenue Exceeds Forecasts Despite Benchmark Gaps: AI company revenues grew far beyond official projections in 2025, driven primarily by coding tools like Claude Code and Cursor, even though benchmark improvements lagged expectations, suggesting CFO projections use conservative backward-looking models that cannot account for unprecedented AI adoption curves.
• •Cost Efficiency Drives Deployment: Arc AGI benchmark shows 390x cost improvement in one year for equivalent capabilities, meaning older chips like H100s remain economically viable as smaller models achieve previous frontier performance, creating sustained demand across entire chip generations rather than requiring constant hardware upgrades for same economic output.
• •2026 Agent Breakthrough Expected: Forecasters predict AI agents will reliably complete day-long autonomous tasks by end of 2026 versus current two-hour limit, with computer use reaching human-level reliability, potentially triggering second ChatGPT moment as consumers directly experience AI booking travel and managing errands, intensifying political pressure around unemployment concerns.