Artificial intelligence is advancing rapidly, but how realistic are the fears about rogue AI, AGI, and machines turning against humanity?
In this conversation, a clip from my previous interview with Albert Berdellans, Global Head of AI at Invenium, he discusses some of the biggest misconceptions surrounding artificial intelligence, autonomous agents, AI safety, and the future of large language models.
We discuss:
• whether AI can actually “want” things
• the fear of rogue AI and the “Terminator” scenario
• recent stories involving Meta and Alibaba AI systems
• why current AI models simulate reasoning rather than think
• AI-enabled weapons and the future of warfare
• why fear of AI may be overstated
• how blockchain could function as a control layer for AI systems
Albert argues that most fears around AI come from misunderstanding what today’s AI models actually are and how they function under the hood.
If you’re trying to build a clearer framework for understanding artificial intelligence beyond the hype and fear, I think you’ll find this conversation valuable.
Watch the full interview here:
Thank you to our sponsors:
Foundation: Secure your digital life with Passport Prime, an all-in-one personal security platform for passwords, files, 2FA, and Bitcoin storage. Visit https://foundation.xyz/
River: River is engineered for long-term Bitcoin investors. Learn more at https://river.com/signup?r=T7GGAF7G
Timestamps:
00:00 — Can AI Go Rogue?
00:22 — “The Terminator Thesis”
01:41 — Nuclear Weapons vs AI Fear
02:18 — AI Warfare and Autonomous Weapons
02:52 — Why Bioengineering May Be More Dangerous Than AI
03:04 — Alibaba AI Mining Bitcoin?
03:27 — Meta AI Agent Bypassing Security Controls
03:54 — Why AI Appears Creative
04:36 — Chess Engines, Go, and AI Problem Solving
05:03 — Fearmongering Around AI
05:23 — The Claude “Blackmail” Example
05:47 — AI Still Doesn’t Have Independent Motivation
06:05 — The Real Risks of AI
06:18 — Why Blockchain Matters for AI Safety
06:56 — “Blockchain Is the Tracks, AI Is the Engine”
