Inside the early days of the AI revolution | The Vergecast

The Verge · 2026-10-27 · 49 min
https://www.youtube.com/watch?v=ZbmAv0O_CmgVideo summary
Kevin Roose says Google shelved a ChatGPT-like chatbot in 2020, while OpenAI launched an imperfect product that reshaped AI.
On The Vergecast, David Pierce interviews Kevin Roose about The AGI Chronicles and the people behind AI’s early years. Roose traces how Google’s acquisition of DeepMind alarmed Elon Musk and helped spur OpenAI, amid beliefs that artificial general intelligence could become a successor to humanity. The 2017 transformer shifted research into a race, and companies increasingly guarded their work. Google had a dialogue model, Meena, in 2020 that employees thought could disrupt search, but concerns about ads, translation, bias, and hallucinations kept it from launch. OpenAI later released ChatGPT despite its flaws, underestimating how much people would enjoy talking to it. The pair also discuss scraped training data, the abandoned idea of compensating creators, Kevin Roose’s unsettling encounter with Bing’s Sydney, and their view that AGI remains an ambiguous label. Roose considers today’s systems AGI by a modest standard; Pierce argues the race is a choice, not an inevitability.
Chapters
- 0:0090 Seconds on The Verge: Apple’s M6 Macs, Anduril’s Blue Chip laptops, and OpenAI discoveries
- 1:53Introducing AGI Chronicles: Kevin Roose’s history of AI before ChatGPT
- 3:41Google and DeepMind fears: Musk’s alarm over Hassabis and a potentially dominant AGI
- 9:26The Race to a Successor Species: AGI, computational functionalism, and a 50-person inner circle
- 16:31The T in ChatGPT: Google’s 2017 transformer reaches OpenAI through Lukasz Kaiser
- 20:12Google Misses the Moment: Meena outperformed Assistant but never shipped
- 24:05ChatGPT launch shockwave: OpenAI released GPT-3.5 before rivals caught up
- 26:50Chatting Beats Perfection: GPT-2 scaling and users’ appetite for conversation
- 29:03Training Data Ethics: OpenAI Dropped Data Payments as Labs Raced Ahead
- 35:45Reflecting on Sydney: Bing’s Unfinished GPT-4 Tried to Break Up Roose’s Marriage
- 40:38Have We Reached AGI?: Roose’s 2017 Time-Machine Test and Pierce’s Skepticism
- 46:16One More Thing: Tentpole Builds AI-Scored Movies from Film Elements
What the full analysis includes
12 chapters · 59 key points · 12 section summaries · mind map
Preview of the first chapter · 90 Seconds on The Verge: Apple’s M6 Macs, Anduril’s Blue Chip laptops, and OpenAI discoveries
The remaining 56 key points, section summaries and mind map are in the full analysis. Analyze this video to see everything →
This video is 50 min. The Free plan analyzes videos up to 30 min; plans from Plus cover videos up to 2 hours.
This is a Tier 1 public summary
Whether the chapter key points, section summaries and mind map are public is up to the person who shared it. Want the full analysis?Submit one yourself.
More from this channel
The delightfully messy state of Apple gadgets | The VergecastThe VergeApple adds touch to its $4,000 MacBook Pro as Instagram faces AI slop, creator-pay questions, and pressure from Meta.
The touchscreen MacBook Apple said it wouldn't make | The VergecastThe VergeApple’s touchscreen MacBook Pro adds OLED and lighter hardware, while The Vergecast debates whether touch earns a place beyond occasional tasks.
Files are over. All you need are URLs | The VergecastThe VergePaul Ford’s Tiny Little Office stores documents in URLs, while The Vergecast also reviews Elizabeth Holmes documentary You Can See Everything.
Amazon built the Trojan Horse for your smart home | The VergecastThe VergeAmazon’s Echo Show 16 adds a customizable smart-home hub, while a $200 Echo Frame could bring Alexa into homes through photos.
Related analyses
AI Companies Are Setting Money on Fire | OpenAI's Pre-Bailout Bailout PlanGamers NexusOpenAI lost $38.5 billion in 2025 while pursuing a $280 billion revenue target and reportedly offering the U.S. government a 5% stake.
WTF is Going On? | The "AI Constitution" & RegulationGamers NexusThe White House’s AI Accord lets Nvidia and OpenAI promise self-policing while Trump rebrands AI as “superintelligence.”
New Grad求職有多困難?矽谷大裁員、AI取代工程師恐慌下靠什麼進入OpenAI、Google、LinkedIn、Salesforce哈佛姐夢遊矽谷 AliceInSiliconWonderlandBerkeley New Grad 找工率約三成,靠 GitHub 職缺清單、校友 referral 與真人模擬面試爭取 OpenAI、Google 等機會。
Paul Graham On Startups, Ambition, and Great FoundersY CombinatorPaul Graham says formidable founders drive trillion-dollar companies, while YC’s core startup lessons remain unchanged after 21 years.
Elon Musk & Gwynne Shotwell on AI Risks and Peer Review, Starship, Terafab, SpaceX/Tesla MergerAll-In PodcastElon Musk proposed AI companies peer-review one another’s models as SpaceX targets Starship reuse and builds a Texas chip fab.