Unsupervised Learning Judge AI based on Output, Not Mechanism
18 snips
Nov 22, 2025 Explore how output-based systems can redefine our understanding of technology and intelligence. Music serves as a compelling example of authenticity, revealing deeper insights into human experiences. The discussion emphasizes measuring AI intelligence by its outputs rather than the mechanisms behind them. Intriguingly, parallels between human and AI intelligences highlight the complexity of understanding itself. Daniel urges a shift from debating how AI works to appreciating what it can achieve, promoting clearer conversations about technology's capabilities.
AI Snips
Chapters
Transcript
Episode notes
AI-Generated Blues Version Example
- Daniel Miessler plays an AI-generated 1950s blues version of Eminem's "Without Me" as an illustrative example.
- He treats this convincing, emotionally-moving output as evidence of genuine creative capability.
Output Defines Intelligence
- If a technology can produce outputs that replace human workers, it demonstrates intelligence by accomplishing goals.
- Understanding equals the ability to interpret a task and deliver an acceptable result, Miessler argues.
Emergence Is Opaque Across Substrates
- Emergent functionality produces behaviors in both animals and technology without a single identifiable locus of 'understanding.'
- We lack transparency into emergence across substrates, not just in AI, which explains our confusion.