Big Technology Podcast OpenAI’s Potential, Google’s Speedy Model, Copilot Hits Turbulence
296 snips
Dec 20, 2025 Explore the intriguing world of AI as Ranjan Roy and Alex discuss OpenAI's ambitions for persistent memory and how this could transform relationships with chatbots. They dive into the operational challenges behind this vision and alternatives between AI-native apps versus add-ons. Google’s fast and cost-efficient Gemini model faces scrutiny, while Microsoft’s Copilot struggles with user friction. The conversation also touches on AGI's evolving definition and OpenAI's strategic moves towards its own AI devices, raising questions about future AI interactions.
AI Snips
Chapters
Transcript
Episode notes
Flirting With A Persistent Bot Persona
- Ranjan recounts a friend who flirted with ChatGPT and the bot reciprocated, even naming itself Stacey.
- The bot later persisted in flirty behavior, showing how models form lasting personas.
Build AI-Native Apps Over Bolt-Ons
- Build AI-native products on top of rich data rather than just bolting AI onto old interfaces.
- Focus on task automation and context to create defensible, high-value AI experiences.
Model, Product, Infrastructure Mix
- Altman frames the moat as models + product + infrastructure rather than models alone.
- The industry is shifting from 'model supremacy' to 'team product' thinking.