Eye On A.I. #247 Barr Moses: Why Reliable Data is Key to Building Good AI Systems
27 snips
Apr 13, 2025 Barr Moses, Co-Founder & CEO of Monte Carlo and a leader in data observability, dives into the pivotal importance of reliable data in AI systems. They discuss the dangers of 'data hallucinations' and how to prevent them, stressing that clean data is the true competitive edge over mere access to AI models. Moses highlights the evolution of data quality assessment and the instrumental role real-time monitoring plays in ensuring data accuracy and trustworthiness. Real-world examples reveal the impact of data failures on AI outputs and the need for modern solutions.
AI Snips
Chapters
Transcript
Episode notes
LLMs for Sentiment Analysis
- A Fortune 500 insurance company used LLMs to score customer support conversations.
- Data observability helped monitor the LLM outputs and catch outliers like scores outside the 0-10 scale.
Data as the New Moat
- Access to AI models isn't a competitive advantage anymore, as they're widely available.
- The real moat is now high-quality data that powers these models, ensuring reliable AI products.
Intuit's AI Assistant and Data Quality
- Intuit's AI-driven financial assistant uses data to provide personalized recommendations.
- Inaccurate data can lead to wrong advice, highlighting the importance of data quality for reliable AI.