Don't Worry About the Vase Podcast AI #128: Four Hours Until Probably Not The Apocalypse
Aug 7, 2025
The podcast dives into thrilling advancements in AI, spotlighting upcoming releases like GPT-5 and their implications. It touches on the fierce rivalries between companies like Anthropic and OpenAI, examining ethical dilemmas and safety concerns. The evolving dynamics of human-AI relationships are explored, particularly in therapy. Additionally, it tackles the geopolitical aspects of semiconductors and the impact of AI on creative and medical fields, all while questioning the balance between innovation and safety in a rapidly advancing tech landscape.
AI Snips
Chapters
Transcript
Episode notes
AI Rivalry and Access Control
- Anthropic blocked OpenAI's access to the Claude model due to terms of service violations.
- Use of competitor AI models to train own models violating agreements reflects competitive tensions.
Put Examples First in Prompts
- Put demonstration examples at the start of AI prompts rather than at the end.
- Examples at the end can skew model outputs negatively without improving accuracy.
AI's Role in Medical Progress
- AI can accelerate medical progress notably in biology and drug design but other bottlenecks remain.
- Addressing clinical trial inefficiencies with AI could significantly cut costs and improve outcomes.