Why validity beats scale when building multi‑step AI systems

About this episode

The episode discusses the importance of validity over scale in building effective multi-step AI systems with insights from a research paper.

In this episode, Dr. Sebastian (Seb) Benthall joins us to discuss research from his and Andrew's paper entitled “Validity Is What You Need” for agentic AI that actually works in the real world. Our discussion connects systems engineering, mechanism design, and requirements to multi‑step AI that creates enterprise impact to achieve measurable outcomes. Defining agentic AI beyond LLM hypeLimits of scale and the need for multi‑step controlTool use, compounding errors, and guardrailsSystems...

People in this episode

Hosts: Dr. Andrew Clark, Dr. Sid Mangalik

Guest: Dr. Sebastian (Seb) Benthall

Topics covered

• agentic AI
• multi-step AI systems
• systems engineering
• mechanism design
• enterprise impact

Keywords

• agentic AI
• multi-step control
• enterprise impact
• systems engineering
• mechanism design

Mentioned in this episode

Books & works: Validity Is What You Need

More episodes of The AI Fundamentalists

• Metaphysics and modern AI: What is Reasoning and Thinking? · May 5, 2026 · 30 min
• Beyond Boosted Trees: Christoph Molnar on the Rise of Tabular Foundation Models · April 21, 2026 · 32 min
• AI and the lost art of reading · March 3, 2026 · 46 min
• Metaphysics and modern AI: What is causality? · January 27, 2026 · 36 min
• 2025 AI review: Why LLMs stalled and the outlook for 2026 · December 22, 2025 · 42 min
• Big data, small data, and AI oversight with David Sandberg · December 9, 2025 · 50 min