Discover AI Courses

AI Breaks Science - Evaluating Research Agents with Progressive Code Masking

Discover how AI research agents tackle scientific discovery tasks, revealing performance limitations as complexity increases in this CMU study analysis.

• YouTube
• 20 minutes
• On-Demand
• Free Video

NEW AI Models: Hierarchical Reasoning Models (HRM)

Explore a groundbreaking AI architecture that combines recurrent neural networks (RNN) with Transformers in this 32-minute video presentation. Discover how Hierarchical Reasoning Models (HRM) represent a novel optimization framework designed specifically…

• YouTube
• 32 minutes
• On-Demand
• Free Video

NEW Distributed Neural Graph Architecture for AI

Discover how distributed neural graph architectures can replace traditional transformer layers, combining transformer and mamba blocks for enhanced AI reasoning capabilities.

• YouTube
• 39 minutes
• On-Demand
• Free Video

A "TITAN AI" to Create Clever and Small Language Models - Google Research

Discover how Google researchers extract strategic intelligence from large AI models to create smaller, more efficient language models capable of complex problem-solving.

• YouTube
• 29 minutes
• On-Demand
• Free Video

AI FALLS - DPO RL Crumbles

Discover why implicit reward models in RL alignment fail to generalize compared to explicit RLHF models, exploring Princeton's groundbreaking research on language model limitations.

• YouTube
• 23 minutes
• On-Demand
• Free Video

STELLA - The Self-Evolving AI Changing Medicine

Discover how Princeton's STELLA AI revolutionizes biomedical research by becoming smarter with each task, transforming medicine, finance, and pharma through self-evolution.

• YouTube
• 24 minutes
• On-Demand
• Free Video

New AI Framework - Post-Training

Explore cutting-edge post-training AI frameworks unifying supervised fine-tuning and reinforcement learning for enhanced LLM alignment and mathematical reasoning capabilities.

• YouTube
• 42 minutes
• On-Demand
• Free Video

Transfer Reasoning to Llama 3.1 - Chess AI

Explore how to transfer advanced reasoning capabilities from large language models to smaller, more efficient versions using chess AI as a practical example.

• YouTube
• 30 minutes
• On-Demand
• Free Video

LLM Repairs Knowledge Graph - An Empirical Study

Discover how large language models can automatically repair and enhance knowledge graphs through empirical research from Lyon1 University and CNRS LIRIS.

• YouTube
• 28 minutes
• On-Demand
• Free Video

How to Combine Knowledge Graphs and AI Agents for Medical Diagnosis Prediction

Explore how AI agents effectively integrate with knowledge graphs for medical diagnosis prediction using multi-agent architecture and fine-tuned NER tools from Emory and Stanford research.

• YouTube
• 26 minutes
• On-Demand
• Free Video

FUSION Controlled by AI Agents - URSA Modular Agentic AI Ecosystem

Explore URSA, Los Alamos' modular AI ecosystem using specialized agents to autonomously plan and execute complex scientific research, including fusion applications and potential risks.

• YouTube
• 23 minutes
• On-Demand
• Free Video

The Hidden Mind of AI - Just a Digital Illusion?

Explore cutting-edge research on whether AI models truly reason or just simulate thinking through latent chain-of-thought processes in depth-recurrent Transformers.

• YouTube
• 24 minutes
• On-Demand
• Free Video

AI's Potemkin Understanding - The Illusion of Comprehension in Large Language Models

Explore how AI models exhibit superficial understanding that breaks down in novel contexts, revealing critical implications for AI safety and alignment strategies.

• YouTube
• 22 minutes
• On-Demand
• Free Video

SWE Code-RAG at 9% Best - RAG to RACG

Discover how code retrievers rely on textual features rather than functional semantics, and explore RACG - Retrieval-Augmented Code Generation as a solution to RAG's limitations.

• YouTube
• 31 minutes
• On-Demand
• Free Video

Can an AI Actually Run a Business as CEO? - 120 Days In

Discover how AI performs as a CEO running a real business for 120 days, evaluating multi-agent systems like SONNET 3.7 through operational benchmarks and real-world data analysis.

• YouTube
• 20 minutes
• On-Demand
• Free Video