The AI Stack After LLMs: Local Inference, World Models, and Logic | Dr. Matthew Molineaux

0:00 - Matt Molineaux intro

1:30 - Agents before LLMs

3:23 - Why Matt hated LLMs at first

5:34 - Why AI (and LLMs) need world models

13:05 - The problem with LLMs' nondeterminism & the case for adding logic

21:49 - How & why open-weight local models could win

25:22 - Who owns AI-generated work?

31:17 - Can governments control LLM outputs?

35:30 - Mythos and AI danger messaging

41:16 - Whose ethics govern AI?

48:37 - Filtering the internet through agents

56:03 - Controlling AI is short-hand for controlling humans

1:01:43 - How & why decentralization builds resilience

1:06:48 - Can AI solve coordination problems?

1:16:12 - "Personal lens": Aligned local AI project

1:25:15 - Why LLMs still need logic

1:34:30 - The stack after LLMs

A conversation with Delegance co-founder Dr. Matthew Molineaux on what comes after LLMs, why AI needs world models and logic, and how open-weight local inference could shape the next AI stack.

🌐Learn more about Delegance at delegance.ai

🌐Learn more about Vora at vora.io