Why Real-World AI Is Harder Than the Benchmark | EliseAI’s Mario Martone
The a16z Show
- Published
- October 18, 2026
- Duration
- 24:25
- Summary source
- description
- Last updated
- Oct 21, 2026
Discusses machine-learning.
Summary
What changes when an AI mistake doesn’t just hurt a benchmark score, but affects someone’s housing or healthcare? EliseAI Head of Applied Research Mario Martone argues that deploying AI in essential industries requires a very different standard of reliability.Mario joins MTS to explain how EliseAI handles roughly 5 million phone calls a month, why LLMs ar…
Intelligent Report
Sign in to read teasers, or upgrade to Research Pro to commission intelligent report for this episode. Learn more →
Show notes
What changes when an AI mistake doesn’t just hurt a benchmark score, but affects someone’s housing or healthcare? EliseAI Head of Applied Research Mario Martone argues that deploying AI in essential industries requires a very different standard of reliability.Mario joins MTS to explain how EliseAI handles roughly 5 million phone calls a month, why LLMs aren’t the right solution for every problem, and how the team combines language models, traditional machine learning, neural networks, and intern
Themes
- machine-learning