Sam أُعيد تغريده
Sam
29.9K posts
Sam أُعيد تغريده
no surprises here 💀
Andrew Carr 🤸@andrew_n_carr
A surprising result. Is Gemini 3.1 pro really this dishonest?
English
Sam أُعيد تغريده
two erdos problems at age 17
Enrique Barschkis@ebarschkis
Kind of crazy. I had a rough idea for an Erdős problem, gave it to GPT-5.4 Pro, went for a walk, came back to a solution. I verified it, formalized it with Aristotle from @HarmonicMath together with @Tomodovodoo. Incredible how powerful these tools are in the right hands!
English
Sam أُعيد تغريده
prediction: we'll see 3-4 vibe coded alternatives to vivado this year
English
Sam أُعيد تغريده
the whole RAM crisis thing has pushed Laurie to find every single historical way to conserve memory size or bandwith and i am loving every second of this because the stuff she is finding is insane
LaurieWired@lauriewired
In the 90s, Hitachi came up with a bizarre way to conserve memory bandwidth. Their SuperH architecture, intended to compete with ARM, was a 32-bit architecture that used…16 bit instructions. The benefit was really high code density. If you can fit twice as many instructions into every cache line, the CPU pipeline stalls way, way less. This was *really* important for embedded devices, which were often extremely bandwidth constrained in the era. Sega famously used the processors for the Dreamcast, and ARM actually ended up licensing their patents for Thumb mode! I think perhaps the weirdest thing about SuperH was its concept of “upwards compatibility”. The ISA itself is a microcode-less design, all future instructions were trapped and emulated by older chipsets. It’d be slow…but you could run future code on very old chips! Very neat design, a massive success through the 90s and 2000s, that slowly faded.
English
Sam أُعيد تغريده
The greatest lie the devil ever told is that if an AI platform's status page says everything's fine, then everything's fine
English
Sam أُعيد تغريده
ἓν οἶδα ὅτι οὐδὲν οἶδα
Duncan Reyburn@duncanreyburn
No decent philosopher believes ‘machine consciousness’ is (or ever will be) a thing.
Ελληνικά
Sam أُعيد تغريده
models have become competent research hill climbers.
thus evaluation design has become the main problem, because the agents will optimize whatever score channel you expose, including the accidental ones.
one gripe i have about such research trials is that we never compare an ai-aided human or a centaur if you will, to a swarm of agents. it’s quite evident to everyone already that any agentic swarm in today’s time > avg 2021 manual research approach.
Anthropic@AnthropicAI
New Anthropic Fellows research: developing an Automated Alignment Researcher. We ran an experiment to learn whether Claude Opus 4.6 could accelerate research on a key alignment problem: using a weak AI model to supervise the training of a stronger one. anthropic.com/research/autom…
English
Sam أُعيد تغريده
Sam أُعيد تغريده
remember kids, if you want your visual art to sell, you must first become a famous musician
Kurrco@Kurrco
Ye’s early artwork has reportedly been reappraised at $3.1 MILLION 🎨 ▫️ The collection was previously valued between $16,000 and $23,000 on Antiques Roadshow in 2021
English
Sam أُعيد تغريده
Somewhere on earth a group of people who are the largest tenants in the building on 250 vesey street started celebrating
unusual_whales@unusual_whales
BREAKING: The SEC has reportedly eliminated the Pattern Day Trader rule, replacing it with a new intraday margin system. The requirement to maintain a $25,000 balance to engage in day trading is being scrapped.
English
Sam أُعيد تغريده
@sharifhsn @seanhn Scale and the added fact that if you're researching bugs it's very difficult to know what you're looking for in the first place.
English
@seanhn But like… why? Why isn’t prompt engineering a legitimate test of the LLM’s capabilities? It’s not like you’re telling it the exact bug, you’re auditing a section of code for a well known class of bugs. That work can be parallelized very easily across a codebase.
English
Conventionally, if you want to test if an LLM can find a bug where the root cause is a memcpy into a statically sized stack buffer, you would not put exactly that in the prompt as an example.
Stanislav Fort@stanislavfort
New post: We show that small, cheap models can detect the flagship Mythos FreeBSD zero-day (CVE-2026-4747) using a simple harness we call nano-analyzer Models down to 3.6B active params (including open-weights ones you can run locally) would have detected it 100-1000x cheaper
English
Sam أُعيد تغريده
You, a bad forecaster: "I thought there was no demand for AI, but Anthropic is making billions and still can't meet demand. I'm wrong."
Ed Zitron, a great forecaster: "I thought there was no demand for AI, but Anthropic is making billions and still can't meet demand. I'm right."
English
Sam أُعيد تغريده
Sam أُعيد تغريده
Sam أُعيد تغريده
Aura Farming is the primary tenant of Catholic Muslim Alliance.
Philip Proudfoot@PhilipProudfoot
Blessed be the aesthetics.
English
Sam أُعيد تغريده
One note on costs: yes, $20k is a decent chunk of money. But also, $20k of fuzzing may not find these issues! Anthropic found high severity issues in oss-fuzz projects that have had a ludicrous amount of fuzzing compute spent on them.
English
Sam أُعيد تغريده
Pretty good overview of the pathway to cybercrime for kids and the genesis always coming back to gaming. Kinda feel like that Roblox statement really missed the point though (assuming they understood the context).
ABC News@ABC
Matthew Lane is just one example of what cybersecurity experts, authorities and even Lane himself say is a wide-ranging menace: a new generation of tech-savvy teenagers who are uniquely dangerous and surprisingly young. abcnews.visitlink.me/U9WrWS
English