Connecting Language and (Artificial) Intelligence: Princeton’s Tom Griffiths

2026-01-20 · Source: MIT Sloan Management Review · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Computational Cognitive Science · Depth: Advanced, extended

Summary

Tom Griffiths, professor and director of Princeton's computational cognitive science lab, discusses his forthcoming book, "The Laws of Thought," which explores the mathematical and linguistic foundations of both human cognition and artificial intelligence. The book outlines three primary frameworks for understanding the mind: rules and symbols (mathematical logic), artificial neural networks, and Bayesian probability. Griffiths explains how these frameworks, initially developed to understand the physical world, have been applied to the mental world, addressing challenges like learning and fuzzy concepts. He argues that these three frameworks are largely sufficient, operating at different levels of analysis—computational (logic, probability) and algorithmic/implementation (neural networks)—to explain intelligent systems. The discussion also touches on the unique constraints shaping human intelligence versus AI, and the role of language in these models.

Key takeaway

For AI Engineers and Executives evaluating AI capabilities, recognize that current AI systems, particularly large language models, operate under different constraints than human minds. Your expectations for generalization and learning from limited data should be adjusted accordingly. Focus on developing AI systems that are complementary to human abilities rather than striving for direct replication, and consider neurosymbolic approaches to enhance reliability and address current limitations in areas like systematic generalization and complex arithmetic.

Key insights

Human and artificial intelligence can be understood through three complementary mathematical frameworks: rules/symbols, neural networks, and Bayesian probability.

Principles

• Intelligence is shaped by problems and constraints.
• AI and human minds will diverge due to differing constraints.
• Metacognitive labor will become increasingly vital.

Method

Understanding intelligence involves three levels of analysis: computational (abstract problem), algorithmic (processes), and implementation (physical realization), with different mathematical frameworks applying to each.

In practice

• Consider AI systems as "different" rather than "super-human."
• Focus on complementarity between human and AI abilities.
• Explore neurosymbolic approaches for reliable AI.

Topics

• Laws of Thought
• Human Cognition
• Artificial Intelligence
• Neural Networks
• Bayesian Probability

Best for: AI Scientist, Research Scientist, AI Researcher, AI Engineer, Executive

Related on AIssential

• Connecting Language and (Artificial) Intelligence: Princeton’s Tom Griffiths — Intelligence, human and artificial, is best understood through the complementary interplay of logic, neural networks, and probability.
• Chatbots vs Ambiguity: Why Machines Struggle to Understand Us (A TOC Perspective) — Ambiguity in chatbot queries is a fundamental computational limitation, not merely a technical bug.
• History of Artificial Intelligence — From the Turing Test to Deep Learning and Large Language… — AI history reveals a consistent pattern of methods evolving due to prior approaches hitting structural limits.
• The next paradigm: AI, Neural Networks, the Symbolic Brain — Current neural networks are fundamentally limited; a new symbolic brain paradigm is needed.
• Will human minds still be special in an age of AI? — Human and artificial intelligence are fundamentally different, shaped by distinct constraints and leading to varied problem-solving approaches.
• The Predictive Mind: How Human Brains and AI Speak the Same Language — Human brains and large language models both operate on predictive processing, continuously generating expectations and updating based on new information.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by MIT Sloan Management Review.