Machine Learning Street Talk (MLST)

chevron_right

How Do AI Models Actually Think? - Laura Ruis

whatshot 218 snips

Jan 20, 2025

Guest

Laura Ruis

Laura Ruis, a PhD student at University College London and researcher at Cohere, discusses her groundbreaking work on reasoning capabilities of large language models. She delves into whether these models rely on fact retrieval or procedural knowledge. The conversation highlights the influence of pre-training data on AI behavior and examines the complexities in defining intelligence. Ruis also explores the philosophical implications of AI agency and creativity, raising questions about how AI models mimic human reasoning and the potential risks they pose.

01:18:01

forum

Ask episode

web_stories

AI Snips

view_agenda

Chapters

menu_book

Books

auto_awesome

Transcript

info_circle

Episode notes

insights

INSIGHT

Reasoning vs. Retrieval

LLMs seem to perform reasoning by synthesizing knowledge from many documents.
This differs from fact retrieval, which relies on specific documents.

insights

INSIGHT

Influence Functions

Influence functions reveal how pre-training data affects LLM reasoning steps.
They estimate how model parameters change if specific data points are removed during training.

insights

INSIGHT

Code's Influence

Code heavily influences LLM reasoning processes, both positively and negatively.
This is surprising, as code differs from natural language data.

Get the Snipd Podcast app to discover more snips from this episode

Exploring Reasoning in AI: Insights from Recent Research

02:24 • 3min

chevron_right

Influence of Pre-Training Data on AI Reasoning

05:36 • 16min

chevron_right

Exploring the Intersection of Facts and Reasoning in AI

21:30 • 2min

chevron_right

Refining Definitions of Intelligence in AI Evaluation

23:05 • 2min

chevron_right

Mathematical Reasoning and AI Creativity

25:21 • 20min

chevron_right

Unraveling Agency in AI Systems

45:11 • 20min

chevron_right

Exploring Non-Physical Agency and Collective Behavior

01:05:32 • 2min

chevron_right

Navigating AI's Dual Impact

01:07:04 • 11min

chevron_right

Connectionism and Cognitive Architecture: A Critical Analysis

Jerry A. Fodor

Zenon W. Pylyshyn

In this influential paper, Fodor and Pylyshyn challenge connectionist theorists by arguing that connectionist explanations, at best, can only inform us about details of the neural substrate but fail to explain the systematicity and complexity of adult human cognition. They contend that classical cognitive architectures, which operate on symbols, are necessary to account for the systematic nature of human thought and behavior. The paper contrasts connectionist models with classical models derived from the structure of Turing and Von Neumann machines, emphasizing the importance of symbolic representations in understanding cognition.

Surfaces and Essences

Analogy as the Fuel and Fire of Thinking

Douglas Hofstadter

Emmanuel Sander

This book, written by Douglas Hofstadter and Emmanuel Sander, delves into the cognitive mechanisms that underpin human thought. It posits that analogy-making is the fundamental process by which our brains make sense of the world, constantly seeking strong analogical links to past experiences. The authors use a variety of colorful situations involving language, thought, and memory to illustrate how analogy is essential for thinking, from everyday experiences to the highest achievements of the human mind.

Laura Ruis, a PhD student at University College London and researcher at Cohere, explains her groundbreaking research into how large language models (LLMs) perform reasoning tasks, the fundamental mechanisms underlying LLM reasoning capabilities, and whether these models primarily rely on retrieval or develop procedural knowledge.

SPONSOR MESSAGES:

***

CentML offers competitive pricing for GenAI model deployment, with flexible options to suit a wide range of models, from small to large-scale deployments.

https://centml.ai/pricing/

Tufa AI Labs is a brand new research lab in Zurich started by Benjamin Crouzier focussed on o-series style reasoning and AGI. Are you interested in working on reasoning, or getting involved in their events?

Goto https://tufalabs.ai/

***

TOC

1. LLM Foundations and Learning

1.1 Scale and Learning in Language Models [00:00:00]

1.2 Procedural Knowledge vs Fact Retrieval [00:03:40]

1.3 Influence Functions and Model Analysis [00:07:40]

1.4 Role of Code in LLM Reasoning [00:11:10]

1.5 Semantic Understanding and Physical Grounding [00:19:30]

2. Reasoning Architectures and Measurement

2.1 Measuring Understanding and Reasoning in Language Models [00:23:10]

2.2 Formal vs Approximate Reasoning and Model Creativity [00:26:40]

2.3 Symbolic vs Subsymbolic Computation Debate [00:34:10]

2.4 Neural Network Architectures and Tensor Product Representations [00:40:50]

3. AI Agency and Risk Assessment

3.1 Agency and Goal-Directed Behavior in Language Models [00:45:10]

3.2 Defining and Measuring Agency in AI Systems [00:49:50]

3.3 Core Knowledge Systems and Agency Detection [00:54:40]

3.4 Language Models as Agent Models and Simulator Theory [01:03:20]

3.5 AI Safety and Societal Control Mechanisms [01:07:10]

3.6 Evolution of AI Capabilities and Emergent Risks [01:14:20]

REFS:

[00:01:10] Procedural Knowledge in Pretraining & LLM Reasoning

Ruis et al., 2024

https://arxiv.org/abs/2411.12580

[00:03:50] EK-FAC Influence Functions in Large LMs

Grosse et al., 2023

https://arxiv.org/abs/2308.03296

[00:13:05] Surfaces and Essences: Analogy as the Core of Cognition

Hofstadter & Sander

https://www.amazon.com/Surfaces-Essences-Analogy-Fuel-Thinking/dp/0465018475

[00:13:45] Wittgenstein on Language Games

https://plato.stanford.edu/entries/wittgenstein/

[00:14:30] Montague Semantics for Natural Language

https://plato.stanford.edu/entries/montague-semantics/

[00:19:35] The Chinese Room Argument

David Cole

https://plato.stanford.edu/entries/chinese-room/

[00:19:55] ARC: Abstraction and Reasoning Corpus

François Chollet

https://arxiv.org/abs/1911.01547

[00:24:20] Systematic Generalization in Neural Nets

Lake & Baroni, 2023

https://www.nature.com/articles/s41586-023-06668-3

[00:27:40] Open-Endedness & Creativity in AI

Tim Rocktäschel

https://arxiv.org/html/2406.04268v1

[00:30:50] Fodor & Pylyshyn on Connectionism

https://www.sciencedirect.com/science/article/abs/pii/0010027788900315

[00:31:30] Tensor Product Representations

Smolensky, 1990

https://www.sciencedirect.com/science/article/abs/pii/000437029090007M

[00:35:50] DreamCoder: Wake-Sleep Program Synthesis

Kevin Ellis et al.

https://courses.cs.washington.edu/courses/cse599j1/22sp/papers/dreamcoder.pdf

[00:36:30] Compositional Generalization Benchmarks

Ruis, Lake et al., 2022

https://arxiv.org/pdf/2202.10745

[00:40:30] RNNs & Tensor Products

McCoy et al., 2018

https://arxiv.org/abs/1812.08718

[00:46:10] Formal Causal Definition of Agency

Kenton et al.

https://arxiv.org/pdf/2208.08345v2

[00:48:40] Agency in Language Models

Sumers et al.

https://arxiv.org/abs/2309.02427

[00:55:20] Heider & Simmel’s Moving Shapes Experiment

https://www.nature.com/articles/s41598-024-65532-0

[01:00:40] Language Models as Agent Models

Jacob Andreas, 2022

https://arxiv.org/abs/2212.01681

[01:13:35] Pragmatic Understanding in LLMs

Ruis et al.

https://arxiv.org/abs/2210.14986

Home Top podcasts Popular guests