A human metaphor for evaluating AI capability
To use the Mastodon web application, please enable JavaScript. Alternatively, try one of the native apps for Mastodon for your platform.
Stay updated with the latest in technology, AI, cybersecurity, and more
To use the Mastodon web application, please enable JavaScript. Alternatively, try one of the native apps for Mastodon for your platform.
It has been seven years since the original GPT architecture was developed. At first glance, looking back at GPT-2 (2019) and forward to DeepSeek-V3 and Llama 4 (2024-2025), one might be surprised at how structurally similar these models still are. Sure, positional embeddings have evolved from absolute to rotational (RoPE), Multi-Head Attention has largely given way to Grouped-Query Attention, and the more efficient SwiGLU has replaced activation functions like GELU. But beneath these minor refi
It’s been over a year since Star Wars: The Bad Batch ended, but it was a given Clone Force 99 would eventually live on beyond the screen. Next month, we’ll check back in with the team in their first novel, Sanctuary. Written by Lamar Giles (Ruin Road), the tale takes place towards the end of season two after the episode “Pabu.” Now that they’re working with pirate hunter Phee, the Clones—Hunter, Omega, Wrecker, and Tech—are on resource gathering duty after the planet’s recent tsunami. That lead
The first thing to consider when buying a cooler is how you're going to use it. If you aren't heading out for days at a time, you probably don't need an expensive high-end cooler. All the coolers we've recommend above are capable of holding things at a safe temperature for a day, provided you keep them in the shade. Similarly, if you're navigating rugged terrain, you might want beefy wheels—and if you're just going to the beach, you might not need them. Hard-sided coolers: These range from the
Authoritarian regimes thrive by systematically dividing their opposition and silencing the populace, creating a crisis where citizens feel isolated and powerless. This paper proposes a technological solution: a new, secure feature within Telegram to bridge these divisions and forge a genuine national consensus. Executive Summary "The protocol is remarkably efficient, capable of distilling the views of a population of 100 million participants down to a core group of 100 in as few as six weeks.
"Bypassing" specialization in Rust or How I Learned to Stop Worrying and Love Function Pointers I've spent nearly a year developing and refining my own FAT driver in Rust. For much of the last six months, I had to put the project on hold due to school commitments. However, I'm back now, especially since this project has become my most-starred repository on GitHub. During that journey, I (almost) learned how FAT and filesystems in general work behind-the-scenes and in my attempts to navigate the
Choose an area of interest and we will send you notifications of new preprints at your preferred frequency.
With the rise of LLM systems (or “AI” as they are annoyingly called), the term “vibe coding” is all the rage recently. Vibe coding is when you rely almost entirely on these “AI” system to write your code for you via a series of drawn-out conversations. Putting on my cranky old man hat, I find most of it rather frustrating. Now I’m not some retro-sycophant who refuses to use modern tools. I’ve been a software professional for over 30 years at this point and I’ve used lots of tools throughout the
It has been seven years since the original GPT architecture was developed. At first glance, looking back at GPT-2 (2019) and forward to DeepSeek-V3 and Llama 4 (2024-2025), one might be surprised at how structurally similar these models still are. Sure, positional embeddings have evolved from absolute to rotational (RoPE), Multi-Head Attention has largely given way to Grouped-Query Attention, and the more efficient SwiGLU has replaced activation functions like GELU. But beneath these minor refi
I've been working on a complex multi-model database for a few weeks now, and recently I took time to simplify and test out an idea I had on a simple key-value database. I started with the basics: A hash table in memory, a simple append-only log for persistence and durability, and the classic fsync() call after every write to the log for durability. It worked, but wasn't as fast as it could be. In Kevo, that's the approach I use, but in Klay (not public yet, but will be open sourced when ready)
How portable do you really need? Pretty much every projector is "portable" to some degree. Many of the projectors on the best projector list, for example, are small enough to fit in a backpack. They might fill that backpack, but that's their size. If you just want something for the occasional movie night, one of those would be significantly brighter and create a better image. Generally speaking, the smaller the projector, the dimmer it is. How much battery do you need? Most portable projectors
"Bypassing" specialization in Rust or How I Learned to Stop Worrying and Love Function Pointers I've spent nearly a year developing and refining my own FAT driver in Rust. For much of the last six months, I had to put the project on hold due to school commitments. However, I'm back now, especially since this project has become my most-starred repository on GitHub. During that journey, I (almost) learned how FAT and filesystems in general work behind-the-scenes and in my attempts to navigate the
With the rise of LLM systems (or “AI” as they are annoyingly called), the term “vibe coding” is all the rage recently. Vibe coding is when you rely almost entirely on these “AI” system to write your code for you via a series of drawn-out conversations. Putting on my cranky old man hat, I find most of it rather frustrating. Now I’m not some retro-sycophant who refuses to use modern tools. I’ve been a software professional for over 30 years at this point and I’ve used lots of tools throughout the
I'm still chipping away at my summer reading backlog over here, and this week finally made it to Alex Foster's Circular Motion, which came out in May. And, wow, I wasn't quite ready for the emotional journey this one took me on. It's set in a near future — people ride OneWheels and going viral on social media is still a thing some strive for — where the megacompany CWC has created an extreme form of high-speed travel that allows people to zip across the world in no time flat. But, it soon become
A PoisonSeed phishing campaign is bypassing FIDO2 security key protections by abusing the cross-device sign-in feature in WebAuthn to trick users into approving login authentication requests from fake company portals. The PoisonSeed threat actors are known to employ large-volume phishing attacks for financial fraud. In the past, distributing emails containing crypto seed phrases used to drain cryptocurrency wallets. In the recent phishing attack observed by Expel, the PoisonSeed threat actors
We need to augment our command line tools and design APIs so they can be better used by LLM Agents. The designs are inadequate for LLMs as they are now – especially if you're constrained by the tiny context windows available with local models. Agent APIs Like many developers, I’ve been dipping my toes into LLM agents. I’ve done my fair share of vibe coding, but also I’ve been playing around with using LLMs to automate reverse engineering tasks mostly using mrexodia’s IDA Pro MCP , including ex
Looking for the most recent Connections answers? Click here for today's Connections hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle, Connections: Sports Edition and Strands puzzles. James Bond fans, one of the spy's favorite things is the subject of the blue category. Have fun with that. Today's NYT Connections puzzle could be tough. Read on for clues and today's Connections answers. The Times now has a Connections Bot, like the one for Wordle. Go th
Looking for the most recent regular Connections answers? Click here for today's Connections hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle and Strands puzzles. Football fans should cruise through today's Connections: Sports Edition. The purple category isn't too challenging. I figured that one out first for a change. Check out our hints and the answers for today's game. Connections: Sports Edition is out of beta now, making its debut on Super Bowl S
When checking out Uranus, scientists discovered something exciting: that the planet appears to have a long-term... well, call it a situationship. In a new, yet-to-be-peer-reviewed paper flagged by Universe Today, an international group of researchers detailed finding a so-called "Centaur" — a small, icy and rocky body that sits between Uranus and Neptune and is designated as a minor planet — that appears to have been keeping up with Uranus in a delicate cosmic dance for thousands or even millio
Image by Getty / Futurism Rx/Medicines Drugs like Ozempic and Wegovy already had a lengthening laundry list of physical health benefits, but now they're increasingly showing promise as treatments for neuropsychological conditions as well. Specifically, one of the most intriguing potential uses for these GLP-1 drugs is in addiction treatment, for vices ranging from alcohol to opioids to stimulants to gambling. And it's gaining momentum: as STAT News reports, one posh Pennsylvania rehabilitation
Apple @ Work is exclusively brought to you by Mosyle, the only Apple Unified Platform. Mosyle is the only solution that integrates in a single professional-grade platform all the solutions necessary to seamlessly and automatically deploy, manage, and protect Apple devices at work. Over 45,000 organizations trust Mosyle to make millions of Apple devices work-ready with no effort and at an affordable cost. Request your EXTENDED TRIAL today and understand why Mosyle is everything you need to work w
A Smithsonian-led team of researchers have discovered North America’s oldest known pterosaur, the winged reptiles that lived alongside dinosaurs and were the first vertebrates to evolve powered flight. In a paper published today, July 7, in Proceedings of the National Academy of Sciences, researchers led by paleontologist Ben Kligman, a Peter Buck Postdoctoral Fellow at the Smithsonian’s National Museum of Natural History, present the fossilized jawbone of the new species and describe the sea gu
We don’t often think about the internet as being good for our health. In fact, “internet” and “health” are primarily related in a negative context in online spaces -- we’re getting too much screen time, we’re “chronically” online, etc. It may come as a surprise, but access to the internet is commonly recognized as a "super" determinant of health. Limited internet access has been linked to high mortality rates during the COVID-19 pandemic, and having the internet in easy reach means we can stay
Table of Contents I Tested the Inmotion V9 EUC, and It's Easily One of the Best Choices for Beginners The Inmotion V9 is among the best grab-and-go electric unicycles (EUCs) available that I've tested. I say that as someone who's been riding wheels for years now and prefers them to e-bikes. They're not for everyone, but thanks to the V9's size, security features, focus on rider safety and reliable battery risk reduction, it's a fantastic option if you're ready to leave handlebars behind. Plus,
Here’s an evergreen take: There has never been a better time to get off social media. Social services have evolved even further into becoming sticky traps for doomscrolling and AI-generated slop, and are hitherto unprecedented frontiers for rage bait. Bummed out about all the misinformation and being part of a profit machine that funds one increasingly unhinged billionaire or another? Well, there’s a way out. Unfortunately, social media companies don’t always make it very easy to rescind their
In the past few years, no other gadget—headphones, electric bicycle, literally anything at all—has improved as dramatically as the humble robot vacuum. The hardware and software have improved beyond all imagining. There are self-emptying bins, multilevel mapping abilities, and all sorts of bells and whistles, like an automated robot arm! Prices on the entry-level models have dropped significantly, to the point where robot vacuums are now an accessible household tool instead of a statement of bou
Enter the number of editors you will have in this subscription: Editors: Viewers: Based on your editor count ( 2 ), you will have an Ilograph Team subscription. Add 6 or more editors to upgrade to an Ilograph Team+ subscription.
The Environmental Protection Agency said on Friday that it would eliminate its scientific research arm and begin firing hundreds of chemists, biologists, toxicologists and other scientists, after denying for months that it intended to do so. The move underscores how the Trump administration is forging ahead with efforts to slash the federal work force and dismantle federal agencies after the Supreme Court allowed these plans to proceed while legal challenges unfold. Government scientists have b
Paper 2025/1237 Replication of Quantum Factorisation Records with an 8-bit Home Computer, an Abacus, and a Dog Peter Gutmann , University of Auckland Stephan Neuhaus , Zurich University of Applied Sciences Abstract This paper presents implementations that match and, where possible, exceed current quantum factorisation records using a VIC-20 8-bit home computer from 1981, an abacus, and a dog. We hope that this work will inspire future efforts to match any further quantum factorisation recor
The Environmental Protection Agency said on Friday that it would eliminate its scientific research arm and begin firing hundreds of chemists, biologists, toxicologists and other scientists, after denying for months that it intended to do so. The move underscores how the Trump administration is forging ahead with efforts to slash the federal work force and dismantle federal agencies after the Supreme Court allowed these plans to proceed while legal challenges unfold. Government scientists have b