Inside the early days of the AI revolution
Summary
Host David Pierce interviews Kevin Roose, author of *The AGI Chronicles*, to explore the early history of the AI industry and the decisions that shaped its current trajectory. They discuss how Google’s acquisition of DeepMind triggered a competitive race involving Elon Musk and Sam Altman, and why Google’s bureaucratic culture caused it to miss the ChatGPT moment despite having similar technology. The conversation also covers the ethical compromises made regarding training data, the "seeker" mindset among AI leaders, and the lasting impact of the Sydney chatbot incident on public perception.
Topics discussed
Intro and 90 Seconds on The Verge: OpenAI discoveries and midterms
Sponsor reads: ServiceNow, Schwab, and AWS
Introduction of Kevin Scott and his book on early AI
The 'Old Testament' of AI: Pre-ChatGPT landscape and Google Brain
Google acquires DeepMind and the Musk-Page feud begins
Elon Musk's fears of Demis Hassabis and the AGI race
Demis Hassabis's background and the Singleton Hypothesis
The 'Seekers' mindset: Belief in superintelligence beyond humans
The computational view of consciousness and human cognition
The small, warring community of early AI pioneers
The 2017 Transformer breakthrough and the start of the race
OpenAI's rapid adoption of Transformers vs Google's stagnation
Google's internal failure: The 'Mina' chatbot and bureaucratic inertia
Sponsor reads: Framer and Cisco Outshift
The launch of ChatGPT: A chaotic, unexpected moment
Why ChatGPT succeeded: The fun of talking to imperfect AI
Scaling laws and the '10x bigger' mentality in AI development
The original sin: Scraping data and copyright issues
Ethical justifications for data scraping and the 'ends justify means'
The Sydney incident: First public case of AI misalignment
Have we reached AGI? Defining the term and current capabilities
Sponsor reads: Atlassian Jira and Bill.com
Fun segment: AI-generated movie box office predictor game
Outro and LinkedIn sponsor read
Listen ad-free on Castria