How About Tomorrow? How Dangerous Can AI Get, Dax is Down on DeepSeek, and AI First App Development
Feb 3, 2025
Dax explores the surprisingly warm weather in Florida while pondering the perceived dangers of AI. The conversation takes a critical turn as they question the reliability and commercialization of the DeepSeek AI model. Thoughts on techno signatures lead to a broader discussion about the ethos of AI development, including the concept of AI-first applications. Their humorous take on coding tools adds a light touch to the complexities of tech innovation, ensuring an engaging mix of skepticism and excitement for the future.
AI Snips
Chapters
Transcript
Episode notes
Motivation over Knowledge
- AI safety discussions often focus on knowledge as power, but real-world harm depends more on motivation.
- People will find ways to cause harm regardless of AI, making knowledge acquisition less relevant.
Digital AI Harm
- Digital threats from AI can be significant, especially to systems like banking.
- Non-physical harm caused by AI manipulation of digital systems is a realistic concern.
DeepSeek's Dubious Claims
- DeepSeek's claims about training costs are likely untrue, potentially due to export controls on GPUs.
- Dax explains that the reported costs are impossibly low and there are incentives to misrepresent them.