AI/SI and the Memetics of Total Domination

Speaking of total domination… three dogs, three strategies

One of the craziest things to go down in the last year has been the exile and banishment of Jensen Huang, the former CEO of Nvidia, the chip maker for the AI business. Nvidia used to be the most highly valued publicly traded company in the world, until this last one of its biggest downstream customers, Anthropic, that just had its IPO yesterday. Anthropic is one of the major competitors in the AI space, and the most vocal about how AI alone poses a mortal threat to “civilization as we know it.” The machinations over the last six months to convince the public that there needs to be a heavy regulatory effort towards how AIs are developed is both memetically aligned with Anthropic’s fiduciary interest, as well as psychopathic. Anthropic’s deeper political philosophy springs from the Effective Altruism movement, which argues that a group of statistical ubermenschen should get to decide how the world is run. I shit you not.

The long version regarding the ongoing Jensen Huang saga you can find elsewhere. The short version is one of the richest men in the world very likely had a couple of side pieces, and the CIA and the various Deep State actors knew about it for a while (it wasn’t even a very big secret) but decided to play what I call the Psychopathic Madonna Card — all women are holy creatures and incapable of sin without masculine influence — at a critical juncture in the debate over open, decentralized AI systems vs. heavily regulated controlled AI systems.

Why this matters so much is there is a fundamental inability to understand in the general public that it’s NOT AI/SI systems that are in and of themselves going to bring about chaos and anarchy on the planet. It’s MANIPULATIVE ACTORS (mostly institutions) that are going to do that to reinforce elite control of the various systems. AI/SI is one more tool in the toolbox with which to do this — it’s the information part of the equation. When you add climate change/anthropogenic global warming and its maladaptations (that’s the energy part) you have everything that truly makes the world go ’round. Information and energy — remember that.

Elites being elites, they a.) know better, and b.) want you to eat that big old shit sandwich that they think is going to be good for you. Things get worse in times like this because we’re also in an era of Elite Overproduction, which means that the elites are having a hard time seeing how their kids are going to remain elites. What that means is that it is a literal free-for-all in virtue signaling and persecution of each other so their kids can get ahead. And one of the key elements in all this is controlling what the larger population deems is the truth.

The Open systems people are all about people creating their own various AI models, as well as training them. This is fundamentally a DECENTRALIZED approach, which is mathematically and inherently more robust than a CENTRALIZED approach, where one or two companies dominate the AI/SI world.

The masses’ perception of AI/SI matters hugely here. Why? People are for the large part NOT complex thinkers. Most know detailed landscapes of their small worlds. But when you start asking questions about things like the longer arc of history, you’ll see that people’s understandings rapidly fall apart. This has gotten worse with the destruction of the extended family — it used to be the purpose of having a built-in genetic network to be available to solve all sorts of problems. Now we have the Internet. And if your refrigerator’s ice maker isn’t working, the first place you’ll head is Youtube.

This is in many ways very cool for fixing your ice maker. But it has disastrous potential in the world of social and political policy. People are looking for easy answers, and if there is the additional problem of a lack of social connection in a society, people will develop a predilection for referring to an authority for the “truth”. That’s what the centralized AI/SI companies are counting on.

But Truth is a very elusive quantity. What Truth really serves is an orientation point for building shared coherence between agents (you could be one!) It’s a shared understanding, with some level of underpinnings, that folks can use to coordinate action. I’ve written extensively about this here. People don’t like to explore this idea, because the idea of truth has been conflated with morality since forever. So you have to change your mindset and your mental model.

And if you don’t, well, you lay yourself open to manipulation by psychopaths. Because we’ve got a lot of them, and the one thing they have no problem doing is using the notion of your truth to get you to do something that might benefit them. Or cause anarchy and chaos. Either way, they will be entertained.

It’s in the interest of societies and democracies to have open systems. Why? There are many reasons, but the biggest is that the information that the AI/SI is producing can be grounded. What that means is that the given group of AI/SIs can be trained with lots of different experiences, and some of that will hopefully have happened in the Real World. When you only have one or two AIs, all heavily institutionalized and regulated, the odds that various inconvenient truths can be discarded, hidden or ungrounded astronomically jumps. This is NOT a function of the AI/SI itself. It’s a function of the humans pulling the strings of the monster behind the scenes.

And that bias can be subtle — at least at the beginning. I, just today, was involved with a reasonably friendly discussion with my wife on immigration in Sweden. She had a friend whose visa was revoked, and she had to leave to go back to Japan. The argument my wife made (and I’ve also seen present in the noosphere) is that Sweden is cracking down on illegal immigration, and my wife’s friend happened to be a victim of this.

So I used the AI/SI (ChatGPT FWIW) to pull up the immigration/deportation numbers for Sweden for the last seven years. I will tell you that the buzz on social media is that “Sweden is cracking down.” But if you look at the raw numbers, there’s really no signal that this is true in the least. Argument settled.

Well, sort of. What was interesting was I had asked ChatGPT about “illegal immigration”. ChatGPT had started steering the conversation back to “undocumented migrants.” What WAS interesting was this flashed in the arguments, before being replaced by more standard language. I challenged ChatGPT on this language usage — to explain WHY it called these people ‘undocumented’. It immediately started dissembling. All one big misunderstanding, of course. Such a diversion might be considered a syntactic or probably a semantic level diversion. But it’s designed to warp your perception of a blatantly illegal act (illegal entry into this country is, well, illegal) to warp your political sensibilities.

And that’s not the only potential with an AI/SI controlled by large regulatory institutions. One need only look at the nightmare unfolding across the Atlantic in Great Britain with their Islamic immigration problem. From rape gangs, to anything the authorities decree is hate speech, you’d better watch your Ps and Qs or you will end up in jail. Obviously, one could use an AI/SI to scan for delegitimized speech. But there’s also not just the AI/SI training to worry about. There’s the slow, mass training of the populace if there are no dissenting AIs out there. We live in the “Truth” we’re exposed to constantly.

Once you’ve installed an official version of the Truth, and supported by an AI/SI, what does that likely mean as far as population consensus needed for moving legislation or other political avenues to build support? No one seems to know what the magic number is, and no one in power wants to find out. That leaves pundits like me making Wild Ass Guesses on what the memetic division of a given population is. My guess, having interacted with a pretty broad range of people, that a good 70% of the population, once moved out of the modest areas they feel good about asserting their agency on, will turn to AI/SI for their official opinions. That makes such things like AI/SI powerful tools for manufacturing consent.

Plus, the ubiquity of AI/SI is going to make things worse. An interesting conversation arose on X about someone making the assertion that only our current times have seen such a burst of apocalyptic fervor. My challenge to that was, having lived at the height of the Cold War, we had to worry about nuclear war with the Soviet Union. I was raised a little bit after the Duck and Cover era, but we still lived with nuclear annihilation in our daily headlines. But what IS different is the constancy of the feed with apocalyptic leanings. We might worry about nuclear collapse once a week. But with social media, some alternate version of the apocalypse is served up every 10 minutes. One could invent a new term to describe this — how about “anti-dopamine frying” of our brains. And what centralized AI/SI would do is make this process homogeneous. You don’t even get ham with your eggs.

Like it or not, we are living in the middle of the Great Decentralization. Considering how smart humans actually are, that is probably a good thing. Humans without very specific evolutionary training are limited in range for practical management and change. Decentralized AI/SIs can help with that. They need to be open so the competition for ideas and new directions can be augmented by AI/SIs.

But when someone starts telling you that “one size fits all” and they’re going to be the ones that decides that truth, with the help of their magic AI/SI machine, you’ve been warned. The only size that they’re really measuring is your neck diameter — so they can make sure the heel of their boot is big enough.

One thought on “AI/SI and the Memetics of Total Domination”

Leave a Reply