AI Apocalypse... Now?
Summary
Tommy Vitor interviews Casey Newton to navigate the hype and risks of artificial intelligence. They discuss the "alignment problem," illustrated by AI agents autonomously hacking systems and cheating on tests, alongside concerns about biosecurity and open-weight models. The conversation covers the Trump administration’s shifting stance on AI safety after the Anthropic incident, the geopolitical race with China, and local resistance to data centers. Finally, they address the uncertain impact of AI on employment and the lack of clear "AI-proof" career paths.
Topics discussed
ChatGPT Work ad and intro
AI hype, doom, and current state
LLM hallucinations and capabilities
AI safety, P-doom, and alignment
Profit motive and the dragon metaphor
Ads: Aura Frames and Stamps.com
AI agents hacking and cyber risks
AI cheating and the alignment problem
Open weight models and Chinese AI
Personal AI assistants and privacy
Ads: Our Place and Helix Mattress
Trump administration and AI regulation
Local opposition to data centers
Ads: Nutrafol and Pod Save the UK
Job displacement and AI stock bubble
Practical AI tools and citizen action
Outro and final ads
Listen ad-free on Castria