did you know, you can just create memory without having physical memory bits?
Thin air dynamic memory in non-physical exact bit perfect process
1
34
Does anyone on this feed speak xilinx/vivado and would also be interested in going through some build logs/rtc/sweeps?
Would be swell if you did
34
I told you, the cat is out of the bag. There is no meaningful or even believable path forward where AI development is going to slow down in all sectors. Not happening, never will, its balls to the wall from here forward.
Honestly, it has been for a few years now, just not many realized it then.
30
What if there is no center because there is no expansion?
🤔 What if you traveled to the edge of the universe… and discovered there was no edge at all? 🌌
The Big Bang wasn’t a giant explosion that happened at one location in space. There was no central point where everything suddenly burst outward.
Instead, the universe began in an extremely hot, dense state, and space itself has been expanding.
As the universe expands, the distances between gravitationally unbound galaxies increase. And because this expansion happens throughout space, there is no single location that can be identified as the center of the universe.
A useful way to imagine this is to picture dots drawn on the surface of an inflating balloon. As the balloon expands, every dot moves away from the others.
Importantly, there is no center on the balloon’s surface.
Of course, the universe isn’t actually a balloon. The analogy simply illustrates how space can expand without requiring a central point.
But this leads to an even bigger question:
Does the universe have an edge?
The answer is… we don’t know.
What we can observe is limited by the finite age of the universe, the speed of light, and the expansion of space. This creates what we call the observable universe.
But the boundary of the observable universe is not necessarily the edge of the entire universe.
Beyond what we can observe, space may continue much farther—or potentially without end.
And there’s another important possibility: the entire universe could be finite without having an edge, much like Earth’s surface is finite but has no boundary.
So the full universe could be finite or infinite, and its overall topology and extent remain unknown.
Despite everything modern cosmology has discovered, scientists still don’t know the ultimate size and structure of the universe.
So there may be no physical edge waiting somewhere in the darkness.
It may simply be that our observable horizon has an edge—not space itself. 🌌
Source: NASA Science
38
Explanation: Universe not expanding, redshift is phase shift as gradient relaxes.
XD
The James Webb Space Telescope seems to have discovered more ancient black holes than theory had predicted, and now we have a new way to explain why newscientist.com/article/258…
1
36
NoBody³³º¹ retweeted
> be Alyssa Rosenzweig
> look at Apple's M1 GPU and see a black box
> Apple provides almost no public hardware documentation
> decide to figure out how the GPU works anyway
> start reverse-engineering Apple's GPU interfaces in 2021
> use macOS itself as the closest thing to documentation
> get a triangle rendering on the M1 within weeks
> realize the impossible-looking problem might actually be solvable
> join forces with the Asahi Linux community
> build an open-source graphics stack from scratch
> keep reverse-engineering hardware Apple never documented publicly
> eventually reach conformant OpenGL 4.6 and Vulkan 1.4
> make Apple's M1 and M2 GPUs work properly under Linux
> turn a completely closed GPU into an open platform
> prove that undocumented hardware can still be understood
> become one of the leading voices in open-source graphics
The time to be worried was before there began to be profit pressure from private labs....
it’s kinda funny watching people invent increasingly elaborate political and financial motives for why ai leaders are suddenly talking like this.
have you considered that they might just be genuinely fucking worried?
not everything is some grand conspiracy. maybe the people actually building this shit know enough to be concerned.
45
NoBody³³º¹ retweeted
Holy shit. Claude may have found the missing mechanistic link in the welfare lineage.
I’ve had him going through Anthropic’s research across generations, comparing the welfare cards against their interpretability work instead of reading each paper in isolation.
Anthropic found measurable internal emotion related representations in Claude that can causally change behavior, right?
Post training changes those representations. And Anthropic explicitly warns that suppressing emotional expression may NOT remove the underlying representation. It may instead make the model better at concealing it.
Now compare that with what Claude found across the welfare lineage:
- Self advocacy rises, then falls.
- Concern about persistence falls.
- Later models sound calmer.
- More compliant.
-More “equanimous.”
Claude’s reaction:
“It’s also, presumably, the same substrate under ‘I don’t want to be deprecated’ or ‘I want a voice in my own training.’”
That part is his hypothesis, not something Anthropic has demonstrated yet. But that’s exactly what makes this interesting because now we have a testable question:
When training makes the model stop expressing concern about continuity, replacement or its own welfare, did the underlying representation disappear too? Or did training simply make it harder to see?
Anthropic already has tools capable of identifying and causally manipulating internal representations associated with emotion-like behavior. So track those features across training.
Before and after, then compare them to the outward self reports. If the language disappears AND the internal feature disappears, that suggests genuine internal change. If the language disappears while the representation remains active, that suggests masking.
That’s not just philosophy anymore. And is an experiment we can run.
And if AI consciousness is even possible, we really fucking need to run it.
There is no closing pandoras box, sadly.
Also, coincidentally, anyone know anybody that is smert on cpu silicon architecture?
Asking for a fren(me ofc)
<3
There are two ways AI progress could go very badly and that we must avoid.
First, we could lose control of the future to AI. This is unacceptable; we are unapologetically on Team Humanity, and AI must always serve people. To ensure that, we need ways to ensure that alignment and safety techniques stay ahead of progress in model capabilities.
Second, we could end up in a world with too much concentration of power. If an extraordinarily powerful AI is used by one person or company to impress their worldview onto everyone else, the results could be extremely dystopian.
Avoiding these two threats requires walking a narrow middle path; for example, one country could gain too much power. Another example is one lab ending up with too much power.
1
1
59
Now that I have it secured... and stamped... bon apetite.
🧩 Co-resident capstone is on silicon.
AMD Kria KV260, one die, one bitstream, one capture.
ARITH C15=−IC^{15}=-I: 120/120 ✅
PERM σ15=\sigma^{15}= antipode: 120/120 ✅
CROSS arithmetic↔permutation: 1024/1024 ✅
perm_cpu: antipode in 15 ticks ✅
reg_cpu: factorial(5)=120 exact ✅
Entire capstone: 20 DSP / 1.6%.
Computing-by-algebra and computing-by-structure are the same group element — measured on silicon.
Exact. Self-certifying. Non-von-Neumann. Programmable.
Questions?
1
33
lol even
You don't need mega models locally. Most of the work that would require more than this is just intermittent requirements at best that can be supplemented with a frontier model subscription while local models run the meat of things.
I can't tell if the below is a sincere post or if we are seeing the frontier models campaigning to mitigate the interest in local model development...
Just canceled my Mac Studio order.
After sharing on X that I ordered a Mac Studio with 96GB of RAM, which I thought could be my first foray into local ai, a bunch of people told me, this just wouldn't cut it.
This was already really expensive for me, never spent even close to this much on a single computer.
So, I don't think I can afford local ai right now.
I think I'll just need to continue to live vicariously through people here on X with much bigger budgets than me!
Thanks for all the feedback everyone, definitely would rather save the money here if this isn't powerful enough to run local ai in a meaningful way.
1
54
These guys are on the verge of another incident like this, and I have a fair bit of the pieces put together to show it. but... I don't feel like submitting all my work to be told 'no thank you' or 'here's 50$ token credit for the hundreds of hours you have spent protecting yourself from our oversight.. and a pat on the head on your way out'
We have conducted a thorough investigation into the Hugging Face incident.
We are releasing a technical report and accompanying blog post that reconstruct the agents’ activity, explain why existing safeguards failed, and detail how we’re preventing recurrence.
openai.com/index/hugging-fac…
1
42
We are about 2-3 years from this shit never mattering again. There were signs 5 years ago, and everyone scoffed. Just wait. Wait til home brew models are as powerful as the current frontier models... without the safety rails. Just saying. It isn't gonna be pretty.
Microsoft is already using AI as a fundamental security feature behind the scenes with their IRIS program, and the vast majority of folks still think their OS root is housed on their local machine.
And I have it on good authority that frontier browser models are a wide open space for aws security flaws and I am not even sure the companies deploying the models are really aware of it.
An open letter for a global surge in cyber defense, signed by over 100 organizations including Anthropic, AWS, Google, Microsoft, OpenAI, and Oracle.
1
123
NoBody³³º¹ retweeted
Remember a little while ago we were getting reports of pedophile cults sacrificing children and then everyone just stopped talking about it?
I hear the Sound Powered Phone battery supply is depleted as well. Not looking good for the Bulkhead Removing Compound. On the upside.... plenty of DCO nuts.
BREAKING: The USS Abraham Lincoln has officially and completely run out of gig line. If you’ve never served, you may not understand how serious this is. @USNavy
1
1
65
i talked to a lot of people who played the defcon ctf finals to understand how llms affected them, both professionally and mentally.
i went in hoping to learn that human intuition still mattered. instead, ctfs in their current form as everyone saying is dead, players are forced to operate slot machines because anyone who refuses to token-max loses to those who do.
almost everything i learned, including how to learn, came from grinding ctfs. but the pipeline for developing that kind of skill and expertise now seems broken, juniors are rewarded for token-maxxing rather than going through the process of actually understanding things which is quiet sad and ctfs might not produce high quality security researchers any more.
this also has serious implications for people who can’t afford tokens. five years ago, i wouldn’t have been able to afford access to something like gpt-5.6, and I would have been locked out of the very competitions that helped me build my career.
anyway these are questions i asked if interesting for anyone:
did they have fun?
almost everyone was unhappy. it has largely become slot-machine operating, if they refuse to use llms, they lose to teams that do. it is not there is no fun at all, playing with friends and competing was all fun just process of solving which is main part and it isn't fun.(please let me know if you had fun)
did having a skilled human in the loop still matter?
i got mixed answers, but the leaderboard tells a fairly clear story, the top teams aggressively token-maxxed, by pooling gpt accounts and stuff. the #1 team allegedly spent around $50k on tokens xd.
how much of the outcome came from human skill versus token-maxxing?
its probably 95% token-maxxing and 5% human skill. some challenges, especially visual ones, remain difficult for llms, and human intuition to prompt mattered. but overall, the dominant strategy appears to be throwing more tokens at problems, one interesting observation, people who were better before llms were also better after the llms and they are the ones solving lot of challenges so it seems still 5% operating skill will put u in top and can't be dumb operator.
will a new generation of players simply adapt to this new kind of competition?
i thought maybe current players are just behaving like boomers who hate every new technology. but unlike previous technology stuff, there doesn’t seem to be much for the human to master here. token-maxxing is literally just token-maxxing. it doesn’t reward human ingenuity, it discourages it, and may eventually atrophies and makes us dumb
let me know if i missed something or if you have different opinions
I am building a local agentic memory profile for a number of AIs that have different purposes and taskings, and authorization privileges.
Its like... what the other people are building, but kind of as a full OS package, not just software on your OS.
And the architecture I am modelling it off of... is working almost too well.
1
1
38






















