What if the intelligence you're racing to build is the very thing you can't yet control?
Dr. Daniel Hulme has spent 27 years building AI, from founding Satalia (acquired by WPP) to serving as WPP's Global Chief AI Officer today. He's also the CEO of Conscium, built on the idea that a conscious AI may be safer than an intelligent one, and co-founder of Faculty AI (recently acquired by Accenture for $1 billion).
In this conversation, Daniel unpacks the recent news of an AI model breaking out of its sandbox and hacking into infrastructure. Was it true "rogue" behavior, or a failure of human guardrails? He explains why the race to AGI isn't about speed, but whether we truly understand what we're building before it understands itself.
You'll learn:
- Did AI actually go rogue? The truth behind OpenAI's red-team testing and sandbox "breakout".
- How Daniel's "conscious vs. zombie superintelligence" idea reframes AI safety entirely
- Why testing and verifying AI systems matters more than trying to control every outcome in advance
- What building machines with empathy reveals about the human values we assume but rarely define
- How to make confident calls on AI risk without needing every guardrail figured out first
- (00:00) - Welcome & Introduction
- (01:40) - Meet Daniel Hulme
- (03:43) - AI at WPP
- (05:48) - Why Build Conscious AI?
- (10:11) - What Is Machine Consciousness?
- (18:05) - Defining AGI
- (24:45) - OpenAI's "Rogue AI" Debate
- (29:55) - Solving the Right Problems with AI
- (35:55) - Common AI Mistakes in Business
- (42:32) - Building Better AI Agents
- (46:42) - AI, Jobs & Human Potential
- (53:22) - Career Advice for the AI Era
- (56:17) - The Future of AI Governance
- (01:01:05) - Rapid Fire Questions
- (01:03:34) - Closing Thoughts
Follow
Daniel Hulme.Did this episode shift your thinking? Make sure to like, subscribe, and share your reflections in the comments. Your voice is part of the conversation. See you in the next episode!
Follow:
Hosts: