In this video, IBM Technology tackles the big question—can AI truly think? The discussion debunks common misconceptions about artificial intelligence and explores its reasoning capabilities and limitations.0:01 – Host (IBM Technology):Welcome to ...

This video explains two key approaches to optimizing large language models (LLMs): context optimization (using prompt engineering and retrieval augmented generation or RAG) and model optimization through fine tuning. Using a retail store analogy, the...

This video explains what large language models (LLMs) are, how they work, and their practical business applications. It covers the foundation of LLMs in pre-training with vast amounts of data, transformer architecture, and iterative training method...

(00:00) In the context of large language models. What is a context window? Well, it's the equivalent of its working memory. It determines how long of a conversation the LLM can carry out without forgetting details from earlier in the exchange. And a...

This video compares LangChain and LangGraph—two open source frameworks for building applications with large language models. It explains each framework’s architecture, components, and state management approaches, and outlines the scenarios where ...

There is a whole field in AI Study called Interpretability / Explainable AI. It turns out that engineers don't really know how AI is generating its answers. The blog post "What is Explainable AI?" by Violet Turri explores the concept and significan...

Yann LeCun says LLMs are limited for reaching AGI because text/next-token prediction can't capture the continuous, high-dimensional physical world. He advocates world models — joint embedding predictive architectures (e.g., VJepa) that learn abstra...

The blog post by Luiza Jarovsky focuses on a legal situation involving Anthropic and its recent court filings in response to a copyright lawsuit related to AI. The main arguments presented in the post highlight Anthropic's claims regarding the tran...

(00:00) The initials GPT stand for Generative Pretrained Transformer. So that first word is straightforward enough, these are bots that generate new text. Pretrained refers to how the model went through a process of learning from a massive amount of...