@TrueRick312
Account based inSwitzerland!
About this account
- Account based in
- Switzerland
- Connected via
- Web
X says this location may be affected by a proxy or VPN.
Account-level information from X, not a live location or the device used for a specific post.
Building a better DeFi for users at https://nitter.cf/t.co/LHJsBwlwp4 | AI Architect, Senior Data Scientist and Full-Stack dev | Bitcoin + Monero
Joined June 2022
- Tweets3.4K
- Following268
- Followers21.3K
- Likes2.2K
Not your node, not your privacy.
On a Monero desktop wallet, the node you connect to sees which transactions you broadcast. Feather, open source and lightweight, defaults to letting you pick or run your own node instead of a stranger's. A small setting that quietly decides who watches your traffic. #Monero $XMR
ALT A small brass key beside a partly folded paper map on a dark matte surface.
My assessment after testing Opus 5.5 on a simple programming task:
TL;DR: No substantial improvement in the model's critical judgment, i.e., in its ability to reason symbolically. It remains a token-spitting machine, not a truly intelligent mind.
During a simple planning phase in Cursor to extend the functionality of an existing 40-line Python script, it took 5 iterations to arrive at a plan that actually made sense. It keeps overcomplicating everything.
After I asked it to build the solution, it managed to turn 40 lines of good, concise, functional Python into over 1,000 lines of "slop" spread across 5 modules.
The task could clearly have been solved with fewer than 200 lines of code with just the script and a config file.
The problem that "vibe-coders" refuse to see here (because "it just works")? Maintenance costs and bug risks rise exponentially with code size and complexity. Generating 5 to 10 times more code than necessary during each "vibe-coding" iteration causes the cost of operating the resulting codebase to skyrocket—eventually reaching a point where even the machine itself can no longer handle it.
Is it useful for non-programmers (and even average programmers with limited talent) to create personal tools for non-professional use? Yes. Can expert developers and software architects rest easy knowing this won't replace them? Also yes.
Yet further proof that the claim that we can achieve ASI using current methods (data-averaging machines) violates the fundamental laws of physics and mathematics, and is therefore impossible, merely a delusion intended to prop up fantasy business valuations.
As the paper cited below demonstrates, the so-called "recursive self-improvement" of AI models, given their nature as data-averaging machines, translates into the n-dimensional equivalent of applying successive rounds of smoothing to a numerical series: a progressive loss of the original signal, which can never uncover new values within that original signal.
It is the principle of Entropy applied to information in action.
There is no such thing as a free lunch in AI, either.
This is to my point about compression being inevitable as human knowledge is too small for what we are building. As well, humans are too redundant in their wants needs and questions.
AI-generated content will clearly contain propagation errors just like human history of knowledge does. Only LLMs will iterate those propagation errors infinitely faster with less ability to self- correct, for want of understanding.
This gets to Ballard’s test. LLMs cannot attain understanding (AGI) as understanding cannot exist unless reason first exists without language. A likely impossibility for a language model.
Research on this is already focused on getting around this in some way. Though many also have not yet conceded the point.
I wonder what was of Petrov:
"On the night of 25–26 September 1983, Lieutenant Colonel Stanislav Petrov was the duty officer at the secret Serpukhov-15 command bunker south of Moscow. His job was to monitor the new Oko satellite early-warning system, which watched U.S. missile fields. Cold War tensions were already high after the Soviet shoot-down of Korean Air Lines Flight 007 three weeks earlier.
Shortly after midnight, alarms sounded and a screen flashed “LAUNCH.” The system reported one intercontinental ballistic missile launched from the United States, then quickly added four more. The display upgraded to “MISSILE ATTACK.” Protocol called for Petrov to report a confirmed incoming strike up the chain of command so Soviet leaders could decide on retaliation under a launch-on-warning posture. Flight time for such missiles was roughly 25–30 minutes.
Petrov did not immediately pass it on as a real attack. He later said he had “a funny feeling in my gut.” His main reasons were practical: a genuine U.S. first strike was expected to involve hundreds of missiles to overwhelm Soviet forces, not five; the Oko system was new and he considered it unreliable; and ground-based radar, which could only see over the horizon later, showed nothing. He reported the incident as a system malfunction. After about 20–25 minutes with no detonations or further confirmation, it was clear he was right.
Investigators later found the cause was an unusual optical alignment: sunlight reflecting off high-altitude clouds over the northern United States hit the infrared sensors of a satellite in a highly elliptical Molniya orbit and produced signatures the software interpreted as missile exhaust plumes."
We have reached the point where AI-generated content is so clever that the average person thinks it is just nonsense.
That is the singularity threshold for most people: the moment they are no longer able to accurately evaluate what the models are saying.
Let's no forget wrench attacks if you keep it at home.
The first layer of defence is privacy. They can't target you if they can't see you have money. That's why you never hear of Monero wallets being drained.
Monero slowly but constantly adding the most powerful tools of cryptography to keep your privacy.
The third version of beta stressnet for Full-Chain Membership Proofs (FCMP++) and CARROT will go live on October 5!
'Full-Chain Membership Proofs prove the output spent is one of any output on the chain, effectively removing all of these risks. This means every input goes from an immediate anonymity set of 16 to 100,000,000.'
Pace the frontier my balls
Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR.
We don’t yet understand what this system does, but only a handful of known systems share its features, and all of them are able to cut, copy, and paste DNA. Historically, the discovery of such programmable systems has helped revolutionize medicine. CRISPR, for instance, is now the foundation of genetic medicines. But it will take much more work to learn what this system does, and whether it can be put to similar use.
Read more: anthropic.com/news/claude-di…
If you want to self-custody your net worth, you really need to take security very seriously. That essentially involves having a second machine only for the cold-wallets, and use that only to transfer funds to your hot wallets where you trade.
And never connect the cold wallet anywhere. Never.
Coruna / DarkSword is being used in the wild.
Attackers socially engineer a Safari click, then walk the full chain: WebKit/JSC memory corruption → PAC bypass → sandbox escape → kernel/root. From there they pull Keychain + wallet data and drain seeds / private keys.
In plaintext, you visit a website and lose your crypto.
If your seed lives on an iPhone, treat this as a wake-up.
Get a hardware wallet. (and update iOS).
It has been over two weeks since GPT-6 Astra was released, yet:
- I haven't seen anyone talking about it on X. It’s the first time a frontier model has been released without flooding the timeline.
- The model is still unavailable in Cursor. It’s the first time a new frontier model hasn't been available within 24 hours.
Which makes me wonder:
- Is point 1 related to point 2?
- Is the second point SpaceX's fault or OpenAI's? Who is denying the platform to whom?
Palantir says.
BREAKING: Palantir's Alex Karp says OpenAI will never IPO.
His theory: nationalization is the only real exit.
When asked what the S-1 risk factors look like, Karp's answer was simple: there is no S-1.
The liability exposure from frontier AI is so large that no public market can absorb it.
The only entity big enough to backstop it is a government.
If Karp is right, OpenAI doesn't become the next Google.
It becomes a utility. Or a weapon.
The most valuable AI company in the world may have no clean path to public markets.
Is nationalization actually the most likely outcome for frontier AI labs?
