Sep 29, 2026 · 1h 31m
AI progress shifts the debate from capability to control
#243: GPT-6 Sol and Luna, Opus 5.5, the New Microsoft Copilot, Jensen Huang vs. AI Doomers & Introducing AI Score
As models become cheaper and more capable, organizations face a harder question than adoption: how to make AI reliable, secure, and governable.
- 1Falling model costs make expanding useful AI applications more important than narrowly managing token expenses.
- 2Jensen Huang’s engineering optimism collides with warnings that advanced AI could exceed human control.
- 3AI agents are moving toward remote task execution, raising practical questions about access, privacy, and security.
Don't miss
The hosts connect remote computer operation with a broader shift from asking AI questions to delegating research and production tasks.
The brief
Paul Roetzer and Mike Kaput open with a skeptical listener mood, then ask whether falling costs and improving capabilities should push organizations toward broader experimentation.
The episode’s central tension emerges in a debate over Jensen Huang’s confidence that AI safety can be solved through engineering, against warnings about losing control of advanced systems.
SmarterX’s AI Score frames readiness as both an organizational and individual challenge, while the hosts show how AI-assisted coding can turn ideas into working prototypes.
Meta’s Muse and remote computer operation point toward agents that can act across websites and devices, but Amazon’s access restrictions expose an emerging control battle.
OpenAI’s agent-security review and proposed international standards underscore the stakes: autonomous systems may need stronger oversight before they can safely improve themselves.
Books & mentions
Some links are affiliate links — PodLume may earn a commission if you buy.
