I Believe That It's Important For All of Us to Understand What 'Decentralization' Truly Means. Please, Let's Talk About That
"We told our suppliers not to deliver raw materials. Our workers were put on leave," Zhang said.
Let's keep in mind this is "business as usual" in China. Contrary to what happens outside, they have huge manufacturing hubs where businesses open and close all the time, all crammed into a relatively "small" area full of suppliers, manufacturers, logistics, etc.
For reference of the scale, Guangzhou has a population of 16.5 million, Shenzhen has 13.5 million, Foshan has 9.6 million. That's metro area, the whole province of Guangdong has over 127 million people.
Never, EVER, do anything security related while sleep deprived, drunk, high, having sex, or all of the above.
After that... no, don't trust. Zero trust.
There are basic hygiene measures to run anything related to any exploit — including "just" PoCs — depending on how risky a total pwn would be:
- container
- VM
- separate machine
- airgapped
- airgapped in a faraday cage (tinfoil works wonders to kill WiFi, a cheap DVB stick turned SDR is great for making sure).
Reading through the code is nice, and should be done anyway from an educational point of view... but even when "sure", basic hygiene still applies.
Keeping tokens in one VM (or a few), while running the exploit in another, is also a good idea. Stuff like ”Windows → WSL2 → Docker", works wonders (but beware of VSCode's pass-through containers). Bonus points if passkeys and a fingerprint reader get involved. Extra bonus points for logging out before testing (if it asks to unlock any passkey... well, don't), then logging out again afterwards.
What I'm not so sure about, is deleting the siphoned data without alerting the potential victims. Everyone kind of failed at security, but still. A heads up to rotate all keys, would be nice.
Not sure if they've edited it, but right now it reads:
the historian George Dyson envisioned the internet as a sentient being that would one day reach artificial general intelligence (AGI)
[...]
Inside China, such a network of large-scale AGI systems could autonomously improve repression
The whole piece looks like written by, or with the use of, some LLM.
Other than that, there are two valid points that could be made:
- Massive application of AI to city-wide surveillance, with zero regards for privacy, could provide an AI agent system with enough compute power to self-train in realtime.
- DeepSeek is plausibly a Trojan horse, trained with a repression based bias, if not directly with hidden malware features.
The near future will see a soft "AI war" in the form of publishing models — to be used as agent cores — with different ideological biases.
Not sure if entrapment, chasing thought-crime, a case of "laws don't apply to aliens", or just a way of diverting some public funds.
There used to be chatbots posing as minors for some time already, expanding it to other areas is disturbing though.
what is gone, exactly?
By adding support for alternate stores, the monopoly argument is gone: everyone can build their own store now. Meaning, everyone with a store can kick out anyone else, and tell them to just build their own.
comply with their own ToS
...which they can change at any moment, but don't really need to; most ToS include clauses about refusing service without having to explain why. If you ever agree to a ToS, better make sure they're even supposed to notify you if they ever decide to cut you off.
LLMs use a tokenizer stage to convert input data into NN inputs, then a de-tokenizer at the output.
Those tokens are not limited to "human language", they can as well be positions, orientations, directions, movements, etc. "Body language", or the flight pattern of a bee, are as tokenizable as any other input data.
The concepts a dolphin language may have, no matter what they are, could then be described in a human language, and/or matched to human words for the same description.
It mentions landfills, but not landfill methane collection? Shouldn't that be step one, to deal with what's already there?
I'd say "no, but"... on Beehaw, you may find whole comment threads removed as "non productive".
There is no financial incentive to promote "engagement", so the policy is to apply a wide brush stroke to anything that looks like disruptive bickering... plus the subjective interpretation of the philosophy "rules". Sometimes nuances get lost, sometimes off-topic discussions that could've been migrated to a different community, or continued via DM, get lost, and you may see mods deleting comments they "don't agree with" in those wide brush strokes.
On average, it keeps a semblance of civilization though, so... 🤷
How to make cheap products in China:
- Use the same production line (same tooling)
- Use less skilled, cheaper workers, producing lower quality
- Replace materials with lower quality ones
- Skip materials where they aren't seen
- Reduce requirements for the quality check
Brands, and the fear of losing the investment put into establishing one, is the only thing keeping people in check. When brands stop caring about quality, they lower it. When manufacturers want to differentiate themselves from scammers, they establish a brand 🤷
Read the case, the whole thing started because Google banned Epic from the Play store, and the only reason for it to become a case, was the monopolistic position. That's gone now, they're free to refuse service to whoever they want, whenever they want, for no reason at all... and if you don't agree, go sue them, they'll show you the precedent followed by the door.
Don't get me wrong, the US is totally going towards a dystopia. I was talking about the remaining 95% of the world and the long term goals. The US has done its part, some US companies can still do some stuff, but the US as a country, or its monopoly-USD "billionaires", are no longer a relevant part of the equation.
Worldwide, NPUs and client-side AIs will lead the next changes. Some people in the US will have a chance to leverage those, learning will become a local query away, only limited by each person's curiosity. Keep in mind a full copy of the Wikipedia is only 100GB, all of Project Gutenberg is only 70GB, I have both and more on the same smartphone I'm writing this from (and it's not even a "flagship" model). There is a lot, and I mean A LOT of knowledge to be extracted from there, which just so happens to be part of the training data for chatbot AIs, meaning they're particularly suited for retrieving it.
You're right, you don't "need" neuromorphic hardware to get those facts right... but at the same time, you don't "need" those facts to use neuromorphic hardware to retrieve them as quickly as you can ask for, then get them with all explanations, related keywords, topics, plus links to sources for it all. With a simple text-to-voice, it will even help you read it!
I know, it may sound like the world upside down. Another way of seeing it, is as the pivot point of a balance, the joining of different ways of approaching knowledge. Interesting times lie ahead 😉
This group has been studying dolphins since 1985 using a non-invasive approach to track a specific community of Atlantic spotted dolphins. The WDP creates video and audio recordings of dolphins, along with correlating notes on their behaviors.
Just the 40 years, for that specific group of dolphins.
Tesla is going to fall hard for a very simple combination of reason:
- It's a meme stock, with 0% dividends and the P/E of a startup despite being 20 years old.
- It's not going to miraculously start making money, when the worldwide sentiment is to let it burn (sometimes literally).
Considering the Elon an oppressor or not, has nothing to do with it. Personally, I'm saddened by the fate of SpaceX engineers once the Elon loses the virtual backing that those fantasy TSLA shares are giving him.
Fighting is not safe, it isn't supposed to. Staying is not safe, not even if you're clapping with both cheeks at every word they say; first they will reduce the pool of "citizens", then infighting will begin. Going to the US right now, whether on a visa, or as a returning "citizen" (for a shifting definition of "citizen"), is not smart.
I do like Balatro and want to play it on my phone, but if I want to do that I have to buy another license, which I can't even do because I don't run Google Play Services
Spoiler: you can use the LÖVE loader to run the "PC version" of Balatro on Android, since it's all written in Lua.
They don't ban the entire company from distributing any software.
They can do whatever, it's their store.
Keep in mind that Epic Games v. Google has made Google add features to allow alternative app stores on Android... which automatically removes the monopoly argument and lets Google ban anyone they want from the Google Play store.
All of us computer geeks in the 90's (and I do mean all) were evangelical about the internet ushering in a Renaissance in dissemination of truth. We were naive.
Hard disagree. The truth has come out, and the Internet has allowed more people than ever to see it... in all its glory and horror. Actually, we're still in the process: only 5.5 billion people have Internet access right now, that's 67% of the world population, we're still missing 1/3rd of the whole picture.
Just wait until you see what 90% of the truth looks like.
As for AI... it's safe (and who's going to read it this deep in the thread anyway?) to tell you a little secret: neuromorphic hardware.
The goal has never been centralized AI like what is being sold right now, not even the dream of an AGI, or some super-AI. The goal is giving every person a self controlled personal assistant capable of sifting through the Internet, or in other words: a smartphone with an NPU running an AI of their choice customized to their personal preferences. The goal is direct democracy where the interests of everyone are taken into account 24/365 on millions of subjects all the time. The goal is giving everyone access to millions of lifetimes worth of skill sets with zero training time. Moderators? You can get hardware with a modest NPU right this moment, download any number of LLMs, and decide for yourself. Moderators will be people capable of affording the hardware, which are already in the hundreds of millions; those billionaires can fuck off, the genie is out of the bottle and it's spreading fast (last I've seen, there's an estimate of a "Moore's law" where AI is growing at 100x efficiency per year, most people don't even remotely realize what that means).
Study material: https://en.m.wikipedia.org/wiki/Paradox_of_tolerance
Your words spit in the face of everyone who died in a hunger strike. Truly disgusting.
The bright side is, that you run it in a container. Beware of privileged mode, don't give it unnecessary mounts or networks, and there's very little some malicious code can do.
If you're using it for a build system, tough luck but you need to manage the keys to avoid TOFU, and ideally pull from your own registry.