Moonshots with Peter Diamandis Why We Need New AI Benchmarks, Which Industries Survive AI, and Recursive Learning Timelines | #218
349 snips
Dec 23, 2025 
Matthew Fitzpatrick, CEO of Invisible Technologies and former quantum AI leader at McKinsey, explores the future of AI. He discusses which industries will thrive with AI and the unique challenges companies face transitioning to AI-native operations. Fitzpatrick emphasizes the need for custom task-specific benchmarks rather than relying on generic standards. He shares insights from AI implementations in healthcare and sports while urging companies to collect focused data and operationalize AI initiatives effectively. A fascinating dive into the intersection of business and technology!
AI Snips
Chapters
Transcript
Episode notes
Uneven Industry Impact
- AI impact will vary widely by industry; knowledge-work documentation roles face the biggest change.
- Sectors like oil, gas, and real estate will see much less structural disruption in the near term.
Start With Two High-Value Pilots
- Pick two to three high-value use cases and run a real pilot rather than letting a thousand flowers bloom.
- Use outcome-based vendors (paid on results) to limit risk and accelerate reliable deployment.
From Broad Benchmarks To Custom Evals
- Public benchmarks (e.g., coding) show broad model improvements but miss enterprise needs.
- Companies must build custom, task-specific evals to judge accuracy and human-equivalence for deployment.