Big Technology Podcast The Big GPT-5 Debate, Sam Altman’s AI Bubble, OnlyFans Chatbots
327 snips
Aug 22, 2025 In this engaging discussion, experts tackle the implications of GPT-5 and whether AI hype is cooling off. They dive into the challenges of integrating AI into real-world applications, especially in industries like healthcare and even OnlyFans. The conversation highlights concerns over AI's effectiveness, with studies showing a high failure rate in projects, while also exploring the potential for AI to serve as companions. The notion of distinguishing between AI's thinking versus doing capabilities sparks thought-provoking debates about its future and societal impact.
AI Snips
Chapters
Transcript
Episode notes
Perceived Downgrade From Prior Models
- Many users feel GPT-5 is a downgrade from earlier thinking-focused models like O3 because it reduces the depth of reasoning by default.
- Ethan Mollick warned GPT-5 can unpredictably switch between high- and low-quality modes within a single conversation.
Build Orchestrators, Not Monoliths
- Design AI systems as orchestrators that pick specialized tools rather than one monolithic model for every task.
- Focus on integrating predefined tools and routing intelligence to solve practical workflows effectively.
AI During A Nepal Trek
- Alex used GPT-5 while trekking in Nepal and found it shifted from a thoughtful partner to an overeager helper that attempted tasks he didn't need.
- He preferred deeper conversational reasoning over agentic actions for planning sensitive activities like altitude decisions.