Better Offline Monologue: The Agony Of GPT-5
63 snips
Aug 15, 2025 The tumultuous launch of GPT-5 sparks heated discussions around user dissatisfaction and unrealistic expectations. Critics highlight the emotional attachments people have to AI models and the flawed marketing strategies of OpenAI. There’s a deep dive into the misguided trust society places in AI, questioning the clarity of guidance from developers. Yet, amidst the critique, there's a whimsical take on the future of AI, encouraging listeners to embrace financial freedom while navigating this evolving technology.
AI Snips
Chapters
Transcript
Episode notes
GPT-5 Is Largely Incremental
- GPT-5 is a multi-model routing system that chooses submodels for different tasks.
- Ed argues the result is largely incremental rather than transformational.
Routing Can Favor Cheap Serving
- OpenAI can route users to cheaper submodels when load or rate limits kick in.
- Ed suggests this incentivizes cost-saving choices over user outcomes.
Users Felt Betrayed By Model Removal
- Users reacted emotionally when OpenAI removed legacy models like GPT-4, describing deep attachments and grief.
- Ed recounts subreddit posts of users calling restored GPT-4 "lobotomized" or "not the same."