Is there a 10% chance AI “could kill all humans”?
Summary
Hosts Danny Forin and Katie Prescott discuss the alarming release of OpenAI’s new model, Astra, which Jensen Huang hailed as AGI, alongside warnings from Anthropic that AI could kill humans within a decade. The episode highlights the disconnect between cozy marketing and serious safety concerns, noting that Anthropic withheld its latest model from UK safety testing. Additionally, the hosts preview upcoming documentaries, including a trailer for a film about disgraced Theranos founder Elizabeth Holmes and a doc about Sam Altman, while teasing an interview with Bill Gates on AI regulation.
Topics discussed
Sponsor: Bombas Sport Socks
Sponsor: KPMG and AI investment trends
Sponsor: ServiceNow AI control tower
Host banter and introduction from Calgary
Intro: AI safety warnings and Anthropic alignment
Preview: Upcoming Bill Gates interview
Elizabeth Holmes documentary trailer release
Resignations and 10% human extinction risk claim
OpenAI Astra launch and security concerns
Marketing critique: Cozy video vs malicious actions
Cal Newport's critique and need for regulation
Anthropic culture and UK AI Security Institute access
Public opinion and focus on practical AI benefits
Transition to Hollywood tech documentaries
Sponsors: KPMG, Vitality, and ServiceNow
Elizabeth Holmes doc and AI era frauds
Upcoming films: Artificial and The Social Reckoning
Outro and Bill Gates episode reminder
Sponsors: KPMG, One Times Square, and Acast
Listen ad-free on Castria