To be fair, the best braves were killed in a generational guerilla war. The sick, cowardly, and alcoholics survived and welcomed in dysfunctional genes from European outcasts. It's dysgenic in the way that British teeth are from the Romans removing the people with good teeth from the local breeding pool because good teeth means good bones means good slaves for quarries and export.
I'm sure there were a couple hundred years of some scary fucking tribes.
The government mandated that the savages were simple and could only be taught to grade 5. European children had a higher ceiling. It's one of the bones they could have picked instead of going with the out and out lies they settled on.
They weren't satisfied with the shit that was on the books. Nutrition experiments on native kids, the law that criminalized them homeschooling, and the regulations that allowed over-housing the red kids. And the regulation that declared them mass-retarded and limited their curriculum to primary school drawn out until adulthood so that they couldn't do trades or basic shit.
Instead they went with a fantasy holocaust imagery about Indians being lined up, beaten and thrown in mass graves. And went on about rapes and beatings that the white kids got too.
The Indians know they fucked up. That if a few bodies are tested, they can flip a coin on which ones come up white in the genetic testing. Per capita they probably died more because of over housing and shit nutrition. In raw numbers, there's enough white orphans that there's going to be just as many or more white bodies in the grave yards and it's going to be fucking embarrassing because the Canadian government couldn't help themselves playing the nazi card to demonize nuns and priests and Canada's founders.
Geriatrics are his base up here. Carney personally doing this tracks.
Muldeep and Sculliqua.
It was about the body positivity movement. Pushing a consensus that empowered white women to blame white men and embrace race mixing. Hitching it to rap was just convenient.
Rock also almost dissolved in the 90s. Grunge and the small venue movement shifted it and there was a period where music piracy was demolishing the bottom line because they were neither making a fortune on stadium shows nor selling albums. That took a long time to sort out and was a colossal fuck up by the cabals not factoring in the internet. Rock culture was broken by that miscalculation.
Rap already had a strong indy scene because of all the hoop dreams people wanting to do it for almost free and the ones who were supported by drug and gang income. They got dominant for awhile and it did fuck all to beauty standards.
Beauty standards got negrofied when (((social engineers))) decided to promote body positivity. Then it snowballed during gamergate and metoo.
It's going to about face now because of GLP-1 drugs. GLP-1s reduce novelty seeking behavior and this is extremely desirable, so these are going to get pushed hard. Less addiction means less money, but it's not about that. People will just be OK doing the same old same old without being bored. They'll shelter in place until the day they die as these drugs improve. And the THICC girls are going to get indignant, but no one will give a shit and they'll slim down when they start taking the drugs.
The Gladeu Report mandates it. If people here want a headfuck they can search that up plus court cases and see how little time people are getting up here. It is legally required to reduce sentences for each DEI intersectionality unless these are negated by whiteness and maleness.
The Gladeu slippery slope has gotten so crazy that even Alberta has a separate fucking court where natives can get dragged in front of a council of elders instead of a judge. Usually they get asked if they're sorry and, shit you not, sentenced to rehabilitation in a sweat lodge and issuing an apology to the community. For violent fucking crimes.
Recently we had a rapist get a sentence reduced from 10-15 to 3 years because he was black and grew up poor so he'd been through enough and couldn't have been expected to do better.
It gets wilder when you look at the merited complaints that BC rejects. Things like service animals leading to fines in strata properties and that kind of thing. But this, they'll pull out all the stops for and fine people out of their livelihoods.
It's happened many times in Canada.
Fucking why? Even if this were true, automation is going to make these people a 100% liability within a decade. They're going to go insane when India dries up and dies because an AI will scam people for free.
Had accreditors not been so bought and paid for that they jumped onto the wrong side of the block chain lobbies, they'd have had a solution for this by now.
There's a push against diagnosing it. Doctors are going on about the antigen test giving false positives to dissuade patients while they don't bother with exams because of the existence of the antigen test. All while a $100 genetic test will tell you whether or not you're in the camp that needs to worry about the aggressive variant that'll kill you off at this age.
I'll get into it. I'll try it with some metaphors but the TLDR; is that they're probabilistic, not deterministic, and that probability plays out in the growing and training phases as well as the prompting phases. I'll leave out talk about transformers and some of the nitty gritty around prompting.
The metaphor for it might be:
- imagine millions of islands
- on each island, there are parts of words
- while it's learning, the LLM stores pieces of words (or pictures/video/sound or whatever it's studying) on islands and tries to sort them so that they flow well into one another and make words and sentences and ideas
- there are not enough islands for every piece of words so it has to try and populate the islands with bits that won't conflict
- now imagine that there are thousands of types of portals that connect the islands (dimensions)
- the LLM, while learning, tries to package content so that it makes sense along a given dimension from island to island
So your LLM studies. It has weights that help it ground and sort initially, but ultimately it's grown. It may or may not "understand" something. Like it might see numbers and go monkeys on typewriters trying to write a black box bit of code that represents a relationship. Adding or subtracting or what not. It might do it like us, but it's unlikely. Usually it comes up with thousands or millions of explanation functions and goes with the shortest and simplest, Occam's Razor.
- so now you have islands and portals and black box subroutines that can be executed
- then there's a post-training tuning where it gets asked questions and it tries to score
- further sorting ensues during the tuning, a new layer
YOU PROMPT IT
- your prompt is analyzed for statistical applicability to black boxes, islands, and portals
- a path is drawn and from each island a path is drawn based on probabilities of the whole
- until the probability indicates that it's done
A LOT CAN GO WRONG
- maybe things get ambiguous on an island and it matches Empi instead of Eiff and tells you that the Eiffel Tower is in New York (this is the typical cause of hallucination)
- maybe the black box doesn't actually do the operation in a way that's correct so the content is just wrong or used at the wrong time
- maybe it selects the incorrect portals (dimensional relationship) between islands
These things are getting more correct though as you throw more money at it, you buy more islands and more portal networks. There is less need for ambiguity and superposition on the islands (less need for tokens to occupy the same position in the vector database + dimensions). There are also more monkeys on more typewriters, so there are more lottery tickets and simpler and smaller Occam's Razors emerge as functions.
Also there are better legacy models to train the new models so if the tuning is known to be good, it can fire millions or billions of questions to post-train refine the model.
*They do relate content to other content weighted but mostly on their own. They do have algorithms that "understand" data in their way. They certainly learn. They probably think in something akin to a reflex arc that responds to stimuli but is not sustained, unprovoked and emergent, or certain other properties we consider to be thinking.
*But they are doing something and that something will augment humans to where they can probably get it to self-improve. Also assisted humans will find a more human path more quickly if a self-improved LLM isn't good enough to stop making stuff up so much.
They'll do it after AI solves enough math to make cryptography unreliable.
I'm impressed.
It's conceptually amazing. If you can isolate it and segment its access to mitigate its damage. The issue with it is that, as an LLM, the content it's generating is extremely vulnerable to context drift and injection.
The average user, even the average engineer, isn't going to have a mature mental model or the time and energy to go about subjecting the agent to zero trust. Like it should, ideally, be on its own system. Within that system it should be restricted to acting only through user accounts that have been given explicit permission to read/write/delete/execute specific parts of the file system as needed. The program environments you've configured for it should likewise be "jailed".
What it comes down to is that you should assume breach and work backwards when granting access to your OpenClaw. Because OpenClaw can ingest poisoned inputs and could even create them if it drifts too far or takes on too much from its initial directives. It can only focus on so much at a time and the superposition of its learning can lead to unexpected outcomes when the roll of the dice grabs the wrong learning and attaches it to the next token and the next token and so on.
Also, importantly, adversaries have already released many malicious packages for it that will weaponize it against you after you blindly install them to grant new abilities to your robot concierge. God help you if you've given it access to your core accounts, credit card, and personal information.
He's been talking a lot lately about using the moon as a staging area for orbital data centers. Gets around a lot of political barriers and creates a lot of thermal and energy collection efficiencies if you can pull it off.
Taken far enough, precedents will make the types of procedures uninsurable. Hospitals will stop pushing their doctors to do them.
It's espionage and diffusion. Their only play is to demonetize the leading labs by giving it away for free in the hope that it buys them the year and a half to two years that they need to catch up. Part of that is the time they need to close the gap on their chips given that the American chips can't be optimized all that much in the angstrom range.
This is state sponsored bullshit and it isn't going to work since this is America's one serious shot at rendering China's underhanded, vizier style, spirit of the law killing long game moot. America has rightfully stamped it a Manhattan project and will reroute financial gravity and energy infrastructure to make it happen either way.
It was progressing too quickly for user skills to keep pace. This halted about a month ago after the top tier of GPT became more able to analyze, correlate, and conclude (not just research and answer) spread sheet, image, text, and tool-based data. The potential customer base for this kind of thing is minute because the average human is unable to do this and so cannot instruct the machine to make use of this ability.
Keep in mind that what you see is a distillation of the capabilities that are being developed to facilitate recursive self improvement that may also happen to have mass market appeal and which are also inexpensive enough in terms of inference compute to deploy to people paying 10s or 100s of dollars.
We're mainly seeing computer programming tools right now. But we're also seeing multi-modal inputs, longer thinking/inference times, larger context windows, the ability to ingest irregular data formats, etc. It's trickling into accounting now.
The situation is that it's now good enough that medium companies are scrambling to use it. Large companies are spinning up groups tasked with redeveloping workflows from scratch as AI-First to try and self-disrupt before they're hit with it from outside. People are starting to figure out industry-specific evals.
There's going to be disruption this year. The thousands of software developers who were laid off. Some of them, and it doesn't take many, are going to start leveraging AI and rethinking knowledge work inputs and outputs as a series of continuous loops that humans interrupt. These are going to scale massively against the companies who fired them and they may end up being cheap enough and angry enough that they don't just get bought out and mothballed.
The West is going to transition to a face culture when the cheating economy grants success to the cheatiest cultures.
He interfered on the morning the fucking election. It's not his job to give a shit about a WEF beachhead forming on his border, or globalist getting tagged in to choke America on Canada's corpse if it lets the globalist American Democrats retake the nation during the midterms, but here we are. It'll serve you right if Carney shuts down our midwest with export taxes on oil and potash. He gives zero fucks and can parachute out of Canada just as quickly as he parachuted in.
They have 174 seats with the NDP. They'll get 4 years if the NDP elects another WEF leader to head their party. They'll get a tap on the shoulder like Singh did and the new guy will be gargling Carney's balls until they're told to do otherwise.
Trump Derangement Syndrome was running hot. Why Trump put in an effort to establish Canada as a WEF beachhead against his country is anybody's guess. The Liberal's right hand guy, Singh, a WEF young global leader who went on to feed his party through a woodchipper to push the WEF middle manager ahead.
Poilievre was up against our media, Trump, CCCP interference that was revealed by the government during the campaign, and the WEF. He pulled the best numbers his side has since Mulroney swept in the 80s. In the end he's just a guy.
They got booted in the last few hours. They changed the criteria in January, after almost 4 years, to exclude the PPC in the 11th hour.
It turns out that the Greens are now caught in that same net because they're pulling candidates from ridings to enrich Mark Carney. Never mind how bullshit it is to remove candidates to try and force your supporters to vote for someone with a different platform.
That candidates in 90% of ridings criterion was specifically included to keep the Green Party in while excluding the PPC and these dumb fucks went and broke it intentionally to play ABC games.
Canada's role in this should be hosting data centers. It's a sad read. Very misguided.