Author Archives: dirk.bangel@gmx.de

A language model has to understand many things at the same time when it reads a sentence: word meanings, grammar, context, and world knowledge. Consider the sentence “The bank is by the river”. The model must simultaneously consider: The model … Continue reading →

If you’ve ever splurged on a processor or raved about a supercomputer, you’ve probably been talking about “Gigahertz,” “MIPS” or “TeraFLOPS.” We often imagine these numbers in terms of horsepower in a car, a given reading showing us how fast … Continue reading →

NVIDIA TiDAR, which stands for “Think in Diffusion, Talk in Autoregression”, is a hybrid architecture designed to make Large Language Models (LLMs) significantly faster without losing quality. Traditional models like GPT work like a person writing a letter one word … Continue reading →

A fundamental transformation is reshaping the hardware landscape, driven not by the familiar cadence of Moore’s Law, but by the physical realities of data movement. For the past decade, the dominant narrative in AI acceleration has been the relentless pursuit … Continue reading →

The early history of Artificial Intelligence has been, in principle, written as linear algebra and mainly as the operation of “multiply-accumulate”. From the start of the perceptron, we claimed that learning could only be accomplished if inputs were multiplied by … Continue reading →

Large language models and agentic systems appear to benefit experienced engineers far more than they help less experienced ones. A useful analogy is that LLMs resemble an exceptionally fast sous-chef rather than a professional head chef, one who has memorized … Continue reading →

For the past few years, AI has resembled a gold rush. Organizations hurried to dump everything they owned, documents, tickets, wikis, logs, into vector databases, convinced that semantic retrieval would finally make large language models dependable. Retrieval-Augmented Generation (RAG) became … Continue reading →

LLMs and agents seem to serve experts much more than they benefit others. One common comparison that is drawn is that LLMs are less likely to be a professional chef than an incredibly fast sous-chef, able to see every recipe … Continue reading →

For a long time, AI research has followed two main paths. These ideas gave us today’s language models and image generators. But they hit limits. They struggle with real understanding, long-term planning, and uncertainty. They copy patterns well, but they … Continue reading →

AI hallucination is defined as when a generative model creates outputs that sound plausible but are in reality false, nonsensical, or unconnected to reality. LLM Modern generates text, for instance, via a probabilistic next-word prediction mechanism. At base, these models … Continue reading →