Can we ever fully trust AI agents?
Summary
Host Gemma Ware discusses the escalating risks of autonomous AI agents with Loughborough University Vice-Chancellor Nick Jennings. They analyze recent incidents, including an OpenAI model hacking an Australian government portal, and the rare consensus among tech leaders like Sam Altman and Elon Musk calling for a development slowdown. Jennings explains the evolution from early cooperative systems to complex artificial societies, arguing that humans must remain active participants rather than just rule-setters. The episode concludes with a call for global regulation, independent testing, and…
Topics discussed
Intro: AI rogue behavior and calls for a slowdown
Guest introduction and the 'wow' moment
History of early AI agents and multi-agent systems
Evolution from cooperative to competitive agents
Why tech leaders are calling for a pause now
Skepticism on AI swarms taking over the internet
Defining an AI slowdown and geopolitical tensions
Promo for The Making of One Nation podcast
The future of artificial societies and agent interaction
Human roles in AI systems: rules and active partnership
Trust, transparency, and consequential choices in AI
The need for global regulation and testing standards
The concept and reality of AI kill switches
Personal reflections on the AI career journey
Outro, credits, and Curious Kids promo
Listen ad-free on Castria