Latest Tech News

Stay updated with the latest in technology, AI, cybersecurity, and more

Filtered by: knowledge Clear Filter

LLM-Deflate: Extracting LLMs into Datasets

Large Language Models compress massive amounts of training data into their parameters. This compression is lossy but highly effective—billions of parameters can encode the essential patterns from terabytes of text. However, what’s less obvious is that this process can be reversed: we can systematically extract structured datasets from trained models that reflect their internal knowledge representation. I’ve been working on this problem, and the results are promising. We’ve successfully applied

We can’t circumvent the work needed to train our minds

The Scam Called “You Don't Have to Remember Anything” Dear Zettlers, This scam is decades old now and it is quite surprising that people still fall for it. The search engines, old note-taking apps (you know, those with an elephant icon and the like) and AI have something in common: They claim that the effort of remembering things is outdated like using a candle in the age of electric light. The following is, by the way, from my Zettelkasten (2016): To find what you need online, you require

We Rarely Lose Technology (2023)

“Στόλος Ρωμαίων πυρπολῶν τὸν τῶν ἐναντίων στόλον,” i.e. “the fleet of the Romans setting ablaze the fleet of the enemies.” i.e. the Byzantines using their Greek Fire weapon. From the Codex Skylitzes Matritensis (12th century) A common trope in the land of fantasy fiction and games is that of lost technology. The hero stumbles upon some ancient ruins, and then finds an ancient weapon, or an ancient vehicle, or an ancient intelligent robot, that helps him in his quest. Nobody alive could possibly

Effective learning: Rules of formulating knowledge (1999)

Dr Piotr Wozniak, February, 1999 (updated) This article will help you overcome one of the greatest difficulties you will face when trying to accelerate learning: formulating knowledge The speed of learning will depend on the way you formulate the material. The same material can be learned many times faster if well formulated! The difference in speed can be stunning! The rules are listed in the order of importance. Those listed first are most often violated or bring most benefit if complied wi

Effective learning: Twenty rules of formulating knowledge (1999)

Dr Piotr Wozniak, February, 1999 (updated) This article will help you overcome one of the greatest difficulties you will face when trying to accelerate learning: formulating knowledge The speed of learning will depend on the way you formulate the material. The same material can be learned many times faster if well formulated! The difference in speed can be stunning! The rules are listed in the order of importance. Those listed first are most often violated or bring most benefit if complied wi

Graph databases are exploding, thanks to the AI boom - here's why

Cobalt88 / iStock / Getty Images Plus Follow ZDNET: Add us as a preferred source on Google. ZDNET's key takeaways The graph database market, driven by AI, is growing at a rate of almost 25% annually. Graph databases support knowledge graphs, providing visual guidance for AI development. There are multiple dedicated graph database vendors on the market. Over the past decade, there has been endless churn in technologies shaping the databases behind the applications we run. The rise of NoSQL

From Classroom to Career: Skills That Shape Your Path Beyond Coding

Key Takeaways Beyond building technical skills, new software engineers need to develop soft skills that enable them to work well with other team members. Understanding the unique needs of each industry in which a professional wants to work makes it easier to make meaningful contributions to the organization. SWEBOK streamlines your transition into the working world because it outlines certifications that give graduates a competitive edge in the technology job market. According to the U.S. Bu

Vibe coding creates a bus factor of zero

All the opinions expressed in this article and on this website are entirely my own and do not represent my employer in any way. Ever heard about the “Bus factor”? It is a concept that measures the risk of losing all knowledge about a particular thing – a software development project for example – by estimating how many team members could get crushed by a bus before nobody knows how to work on the project anymore. As an example, if 3 people on your team know how to restore a backup of your datab

FFmpeg Assembly Language Lessons

Welcome to the FFmpeg School of Assembly Language. You have taken the first step on the most interesting, challenging, and rewarding journey in programming. These lessons will give you a grounding in the way assembly language is written in FFmpeg and open your eyes to what's actually going on in your computer. Required Knowledge Knowledge of C, in particular pointers. If you don't know C, work through The C Programming Language book High School Mathematics (scalar vs vector, addition, multipl

What Medieval People Got Right About Learning (2019)

We tend to assume that if people today and people five hundred years ago do things differently, it’s because we’ve figured out a better way to do it. After all, we have microscopes, democracy and penicillin. People in the middle ages lit cats on fire for fun. Yet despite overwhelming progress, it’s ironically in the area of education that we may be the ones who have it backward. Apprenticeships were, for a long time, the dominant way of learning professional skills. A master agrees to show you

Participate in Standards Development for P2807.9

“Guide for Application of Knowledge Graphs for Rail Transit,” designated by the Project Number P2807.9, is a groundbreaking standard that has significantly impacted the rail transit industry. By providing a comprehensive framework for the development and implementation of Knowledge Graphs (KG), specifically tailored for the rail transit sector, referred to as rail-transit-oriented KG (RTKG), this guide addresses the critical need for consistent knowledge patterns and unified interface standards.

AI is impressive because we've failed at personal computing

Unless someone wrote an article about that exact thing, a plain full-text search engine cannot answer a question like this: What animal is featured on a flag of a country where the first small British colony was established in the same year that Sweden's King Gustav IV Adolf declared war on France? But ChatGPT got the correct answer in a few seconds. Flag of Dominica features the Sisserou parrot, which is only found in Dominica. Great Britain established a small colony on the island in 1805.

Sonos Roam speakers are overheating and partially melting, company admits

Complaints on Reddit shared over the past week have exposed a problem with Sonos’ first-gen Roam speakers overheating and partially melting, which the company just acknowledged while downplaying the number of known cases. Reports of overheating Sonos Roam speakers acknowledged Chris Welch writes at Bloomberg: Sonos Inc. acknowledged a “very low” number of complaints from customers who claim the audio brand’s first-generation Roam speaker has overheated and partially melted near the USB-C port

Beyond Retrieval: The Expanding Universe of Augmented Generation in AI

Introduction Standard large language models (LLMs) possess vast knowledge but struggle with limitations like hallucinations and accessing real-time information due to their static training data. This has spurred the development of dynamic AI architectures. Retrieval-Augmented Generation (RAG) has emerged as a key solution, integrating external knowledge into the generation process. However, the field is rapidly evolving beyond basic RAG. Newer models like RASG (Retrieval-Augmented Self-Generate

Age Verification Doesn't Need to Be a Privacy Footgun

“Won’t someone think of the poor children?” they say, clutching their pearls as they enact another stupid law that will harm the privacy of every adult on Earth and create Prior Restraint that inhibits the freedom of speech in liberal democracies. If you’re totally ignorant of how things work, the proposal of “verifying you’re an adult” before you access adult content sounds, superficially, like a reasonable thing to do. But it’s a patently stupid idea at every level. Age Verification Makes Th

C8 Health started with an AI that gives anesthesiologists guidance on demand — now it’s targeting whole hospitals

Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders. Subscribe Now Medicine is one of the most highly regulated fields in the world, and for good reason — the difference between doing a process correctly and incorrectly can often be that of life or death. But think of the many people involved in providing care at hospitals: it’s not just doctors and nurses, but also the entire medical support staff who ha

Satya Nadella seeks to reassure Microsoft employees in layoffs memo

I also want to acknowledge the uncertainty and seeming incongruence of the times we’re in. By every objective measure, Microsoft is thriving — our market performance, strategic positioning, and growth all point up and to the right. We’re investing more in CapEx than ever before. Our overall headcount is relatively unchanged, and some of the talent and expertise in our industry and at Microsoft is being recognized and rewarded at levels never seen before. And yet, at the same time, we’ve undergon

My favourite German word

My favourite German word¶ 30th June 2025 A documentation colleague recently challenged me with a question: Nowadays, more and more people reach for an LLM tool to provide the information they want. If human beings don’t actually read it, what is the point of writing and structuring documentation for humans? Newer generations (she said) are becoming unskilled at finding information for themselves. They seem less able to digest what they find, to apply it to their problems. But it’s not just t

“The Bitter Lesson” is wrong. Well sort of

“The Bitter Lesson” is wrong. Well… sort of. Assaf Pinhasi 3 min read · 1 hour ago 1 hour ago -- Listen Share TL;DR There is no dichotomy between domain knowledge vs. “general purpose methods that leverage data+compute”. They are both powerful tools that compensate for each other and need to be balanced and traded off during the model building process. “The bitter lesson” in 30 seconds “The bitter lesson” is one of the most popular opinion pieces about AI research and it’s future. In his w

Local Chatbot RAG with FreeBSD Knowledge

Out of multiple conversations with people at BSD conferences, I noticed that many would love to see a chatbot that provides precise information on FreeBSD—for users, admins, and developers. I strongly believe that there should not be an official chat.freebsd.org . Local chatbots work well and can be tweaked to fit personal needs. This documentation is written for macOS with Apple Silicon (because of the GPU support), but should work on other OSes as well. Step 1: Install Ollama (API for Multi

Hidden interface controls that affect usability

Philip Kortum In the early 1960s, Douglas Engelbart [1] first introduced the notion of "knowledge in the world" versus "knowledge in the head" for computer interfaces—an idea that was later formalized and popularized by Donald Norman in his seminal book The Psychology of Everyday Things. From an interface design standpoint, knowledge in the world simply means that the controls you need are visible, and the identification and operation of these controls can be done through recognition rather tha

Stop Hiding My Controls: Hidden Interface Controls Are Affecting Usability

Philip Kortum In the early 1960s, Douglas Engelbart [1] first introduced the notion of "knowledge in the world" versus "knowledge in the head" for computer interfaces—an idea that was later formalized and popularized by Donald Norman in his seminal book The Psychology of Everyday Things. From an interface design standpoint, knowledge in the world simply means that the controls you need are visible, and the identification and operation of these controls can be done through recognition rather tha

Fine-tuning LLMs is a waste of time

It takes time to create work that’s clear, independent, and genuinely useful. If you’ve found value in this newsletter, consider becoming a paid subscriber. It helps me dive deeper into research, reach more people, stay free from ads/hidden agendas, and supports my crippling chocolate milk addiction. We run on a “pay what you can” model—so if you believe in the mission, there’s likely a plan that fits (over here). Every subscription helps me stay independent, avoid clickbait, and focus on depth