anything east of Monaco counts as Eastern Europe btw

Pope confirmed slav, the great schism has been healed.

who is John Paul II

Poles slavs confirmed - none of you are free from slavdom.

anything east of Monaco counts as Eastern Europe btw

Pope confirmed slav, the great schism has been healed.

Your reaction to chatGPT instantly lets me know how easy it would be to trick you into thinking that you are haunted

"omg it's literally alive!" Two beers, 45 minutes, deck of tarot cards, and I'm charging you 350$ for an exorcism.

"I read an article that it's showing simple self-awareness" two days, mild preparation, hot and cold reading, I can get 60$ for joints laced with sacred sage

"I just spoke to an AI and I'm... rattled to say the least, come with me on this dark journey" twenty minutes. I've got to science it up for you, but I can get you to come back every week to "disentangle the psychological imprint" for 125$

“Cyber minion”

Generated by OpenAI’s DALLE

Sharing is caring!

Substack: dalle.substack.com

Twitter: @Dalle2AI

The heading of this post was used to generate the image, src

Now THIS is an AI going rogue, learn from examples

From 1941-1945 the UK was consistently on British Double Summer time (UTC+2), thus meaning that during winters Britain was one hour ahead of Germany, on Central European Time (UTC+1). Also, the reason Spain uses CET despite being solidly in the UTC+0 sector is that Franco changed to it in 1940 to be aligned with Germany.

Failson US Congress unable to pass a bill to fix Daylight Savings Time for literal decades, meanwhile 1940's UK jumping ahead in the timestream on a whim just to fuck with the Nazis. We used to be a real country civilization.

remember, it's imperative to turn your aesthetic preferences into moral ones. you can't just dislike neutral colors, or glass-and-steel skyscrapers, or flat design, they have to be symbols of neoliberal capitalism in decay. it's incredibly important that you make sure everybody knows that the only reason anyone could like the things you don't like is that they're an empty shell of a person.

Back in my day (late 19th/early 20th century high modernism) we just claimed that *aesthetics were morality*, that beauty was the Good worth pursuing and to be degenerate is simply to fail to appreciate the allure of a steel facade, to lack that makes you evil. Nowadays The Kids have to justify their aesthetic moralism with facile claims that it represents capital run amok or racial privileges or sexual degeneracy of what-have-you. Pathetic, absolutely no respect for the primacy of aesthetics over all other values.

Was gonna make a poll on "When did the central powers lose all hope of winning WW1?" and was curious for any sleeper suggests of events. I have obvious ones: UK entry in the war, American entry in the war, defeat at both Battles of the Marne, etc. And less obvious ones - Italy entering the war? Turnip Winter?

Also "causal points" - like I wouldn't vote American entry into the war, I would vote the resumption of the U-boat campaign in 1917, since that to me made America's entry inevitable, no odds changed on that actual day. That logic opens up more minor events due to their impact down the line. Hit me up if you got suggestions!

UK entry seems like a big probability shifter but not a slam dunk in and of itself. At outset I'd say it goes from ~80% Central Power victory probability to something like 30% after Britain enters.

Maybe even a bigger gap. Even with the blockade and (small at first) British army support, Germany knocks Russia out and pushes the French army to mutiny. 90%?

I'll always hold that what lost them the war was twofold:

1 - the Schliefen Plan does not work. It requires a level of uninterrupted marching and continuous victory that cannot withstand a single reversal, and your troops will tire themselves out at the speed requires. The minute the Germans ended that operation without a cohesive answer for "what do we do next", they were doomed. This permeates every single level of the decisionmaking of the Central Powers. The Ottomans, the Bulgarians, the Austro-Hungarians and the Germans seem to have no large scale operational plan other than 1. Draw enemies into battle 2. Fight Battle. 3. Win Battle. 4. Repeat until ????? happens. 5. ????? 6. Win War. That is not a plan.

2 - a fundamental inability to knock out armies and countries completely out of the war. While consistently winning every battle and using the tactical layer to solve all your problems is a highly expensive way of winning a war, you need to actually win. France, Italy, Serbia, Romania, Belgium - all of those still fielded armies by the end of the war. If you cannot get the Serbs to stop fighting, if you cannot get the Belgians to surrender, if Romania stalls you out when you mulch Russia - then you have a separate problem, which is that you can't actually shift enough forces to defeat the enemy.

Doesn't the actual course of the war show there's a way to a German victory if Britain doesn't enter?

The HSF is large enough to severely interdict French trade even if not completely blockading. Plus that means it protects Germany's. No Britain means no BEF. The Schlieffen plan can still fail, but Germany knocks Russia and France out of the war in the next years since it's fighting fewer and poorer enemies, with more money. And those enemies have no reason to think they're going to be saved by Britain or the USA.

-I am definitely team "the Schlieffen plan fails" - its one of those things where 'Cannae writ large over the map of France' just flies in the face of an offense powered by boots against a supply line powered by trains, it isn't happening. The UK's entry into the war is certainly on the books as one of the most unforced errors of military history, Germany did not get much for the price they paid.

-However, strategy is a two way street - I don't see how France or Russia's strategy of winning the war is that is any better than Germany's in 1915. With the UK on their side blockade is an option, but otherwise they are engaged in the same exact bleed and batter shenanigans. And I think the wider political situation is being downplayed - if the UK is not in the war, France & Russia know that, so when they bleed their millionth casualty they are looking at their odds which are way worse in this counterfactual and are much more likely to sue for peace. I think sans UK France & Russia are in a worse strategic dilemma than Germany is. The inherent dynamics of WW1 just mitigated against high strategy.

-Knock on effects of no UK should also be considered - I don't think Italy joins the entente for example. And with no UK no U-boat campaign needed, so no US entry.

-So yeah, I think I agree that the UK's entry is a huge blow to Germany's odds of winning the war; but funnily enough I think they are close to even odds by the close of 1916 still. Then they fuck it up with the U-boat campaign and it tips to virtually zero. Russia's exist from the war seems imo predetermined by the dawn of 1917, and without a US entry that cascades very heavily against France.

I think that by late 1916 they've already lost it from an operational and tactical perspective, because Britain is now bringing their full on Imperial Might to bear on land, not just at sea, and that allows a level of continuous assault and drive and push that characterizes the British land actions in the back half of WW1, where the use of new technologies and new tactics means that they've essentially manifested a second wave of reserves that are better armed, better trained, and better led than any other force on the continent by 1918. That UK force buys the Allies so much time until the Americans arrive en masse that it's astonishing.

I think if you wanna find the earliest point where the Central Powers basically shot themselves in the foot in a fully irreversible manner, it's definitely the UK entry. I don't think that's quite so much as the last point where they did so, though.

I do think Russia's exit changes that dynamic - 1916 is a bad year but 1917 is much better for Germany & co. Like in our timeline, Germany is able to inflict a harsh defeat on Italy and knock Romania out of the war in 1917, and even massively punch a France which is leagues more powerful due to American support in 1918. The UK's "modernized" and technologically improved army was matched by a German army's enhanced tactical and operational practices. I don't think the evidence from 1917 on the ground shows a British army that is notably better than the Germany army - it has improved, it gets wins like Passchendaele, but that is happening in the context of wider defeats elsewhere by the rest of the Allies.

If the UK went Victory or Death then I think they could have continued to mobilize more assets for sure, but that is where politics rears its head - if France is bloodied badly enough I don't see the UK wanting to keep up the fight as long as Germany demands little from the UK in peace (which they might not, sure, but we alt history here, that is within their control). If the US doesn't enter the war I think by the end of 1917 its a coinflip, and one that favors the central powers if anything.

Was gonna make a poll on "When did the central powers lose all hope of winning WW1?" and was curious for any sleeper suggests of events. I have obvious ones: UK entry in the war, American entry in the war, defeat at both Battles of the Marne, etc. And less obvious ones - Italy entering the war? Turnip Winter?

Also "causal points" - like I wouldn't vote American entry into the war, I would vote the resumption of the U-boat campaign in 1917, since that to me made America's entry inevitable, no odds changed on that actual day. That logic opens up more minor events due to their impact down the line. Hit me up if you got suggestions!

UK entry seems like a big probability shifter but not a slam dunk in and of itself. At outset I'd say it goes from ~80% Central Power victory probability to something like 30% after Britain enters.

Maybe even a bigger gap. Even with the blockade and (small at first) British army support, Germany knocks Russia out and pushes the French army to mutiny. 90%?

I'll always hold that what lost them the war was twofold:

1 - the Schliefen Plan does not work. It requires a level of uninterrupted marching and continuous victory that cannot withstand a single reversal, and your troops will tire themselves out at the speed requires. The minute the Germans ended that operation without a cohesive answer for "what do we do next", they were doomed. This permeates every single level of the decisionmaking of the Central Powers. The Ottomans, the Bulgarians, the Austro-Hungarians and the Germans seem to have no large scale operational plan other than 1. Draw enemies into battle 2. Fight Battle. 3. Win Battle. 4. Repeat until ????? happens. 5. ????? 6. Win War. That is not a plan.

2 - a fundamental inability to knock out armies and countries completely out of the war. While consistently winning every battle and using the tactical layer to solve all your problems is a highly expensive way of winning a war, you need to actually win. France, Italy, Serbia, Romania, Belgium - all of those still fielded armies by the end of the war. If you cannot get the Serbs to stop fighting, if you cannot get the Belgians to surrender, if Romania stalls you out when you mulch Russia - then you have a separate problem, which is that you can't actually shift enough forces to defeat the enemy.

Doesn't the actual course of the war show there's a way to a German victory if Britain doesn't enter?

The HSF is large enough to severely interdict French trade even if not completely blockading. Plus that means it protects Germany's. No Britain means no BEF. The Schlieffen plan can still fail, but Germany knocks Russia and France out of the war in the next years since it's fighting fewer and poorer enemies, with more money. And those enemies have no reason to think they're going to be saved by Britain or the USA.

-I am definitely team "the Schlieffen plan fails" - its one of those things where 'Cannae writ large over the map of France' just flies in the face of an offense powered by boots against a supply line powered by trains, it isn't happening. The UK's entry into the war is certainly on the books as one of the most unforced errors of military history, Germany did not get much for the price they paid.

-However, strategy is a two way street - I don't see how France or Russia's strategy of winning the war is that is any better than Germany's in 1915. With the UK on their side blockade is an option, but otherwise they are engaged in the same exact bleed and batter shenanigans. And I think the wider political situation is being downplayed - if the UK is not in the war, France & Russia know that, so when they bleed their millionth casualty they are looking at their odds which are way worse in this counterfactual and are much more likely to sue for peace. I think sans UK France & Russia are in a worse strategic dilemma than Germany is. The inherent dynamics of WW1 just mitigated against high strategy.

-Knock on effects of no UK should also be considered - I don't think Italy joins the entente for example. And with no UK no U-boat campaign needed, so no US entry.

-So yeah, I think I agree that the UK's entry is a huge blow to Germany's odds of winning the war; but funnily enough I think they are close to even odds by the close of 1916 still. Then they fuck it up with the U-boat campaign and it tips to virtually zero. Russia's exist from the war seems imo predetermined by the dawn of 1917, and without a US entry that cascades very heavily against France.

Your two options folks

Thoroughly mystified by the top one. "America colonized heavily by the British with a bunch of cities named after English cities" is just... America again. The bottom one at least makes sense even if it's absurd (minus the part where the Central Powers cede a bunch of territory to Japan, an allied nation in WWI, for seemingly no reason)

As I always comment, love how both sides of the war agree "oh yeah also Japan your ostensible ally is gonna backstab you and take California", a totally reasonable and justified fear in the 1910's for sure.

Very silly when surely California and the rest of the south-west should go back to Mexico as the ZImmerman telegram suggested. I guess Japan has to satisfied with Oregon

But seriously yeah it was a real "of the moment" kind of fearmongering, Mexican political events would rear those fears back up soon enough!

Your two options folks

Thoroughly mystified by the top one. "America colonized heavily by the British with a bunch of cities named after English cities" is just... America again. The bottom one at least makes sense even if it's absurd (minus the part where the Central Powers cede a bunch of territory to Japan, an allied nation in WWI, for seemingly no reason)

As I always comment, love how both sides of the war agree "oh yeah also Japan your ostensible ally is gonna backstab you and take California", a totally reasonable and justified fear in the 1910's for sure.

One of my controversial anime takes is that Dragon Ball Z is overrated and even kind of dull based on what I recall.

If I had not had to leave my country of origin in Latin America for being trans already then I probably would have been persecuted for this take all the same.

Now for one of the *worst* parts of the new ai discourse: the ridic language appropriation:

Imagine trying to put in explicit words how an ai chatbot version of a women whose chosen profession is to create soft core porn fantasy content online for her followers, for the purpose of being a virtual girlfriend to those followers, is 'going rouge' by generating said porn for said followers.

'Around the clock' indeed, holding that level of contradiction in your head would be a full time job!

I mean, there’s nothing too AI-specific in the language misuse here, it’s just ad copy. “Oooh this might be dangerous, watch out kids for the sex and violence you definitely don’t want to see ;)” that’s not cognitive dissonance it’s just a familiar, albeit kind of eyerolly, form of rhetoric.

That said, you know, I don’t expect it will be very hard to select a random person with example camera footage, including nothing sexy, and to make a sexy chatbot version of them, and I don’t feel great about that. When it’s just one person doing it on their computer it’s mostly isomorphic to normal human fantasizing but when it’s passed around I think it has the capacity to be extremely humiliating (but maybe I’m relying too much on my own intuitions here? And the former feels incredibly icky to do, even if I can reconcile myself to others doing it.)

The other thing of course is the consumer side, all of this will be very cheap and eventually free, and I’m guessing (for many people) at least as much a leap in addictiveness from 2022-era pornography as (for many people) going from natural fantasizing to 2022-era pornography is. Not great!

I think that might be a product of being into the discourse too deep (just a guess ofc) - "AI going rogue" is very explicitly a phrase born of the AI alignment and wider doomer scenarios. Absent that thread of discourse this article would not describe an AI going rogue in that way, its not the natural way to put this. I get that "going rogue" is a normal phrase, but its not how outsiders would view a computer program (they would describe it has going haywire, for example). Its been codified in AI very strongly by these debates.

And the dissonance I was referring to is not in the writer of the article, its in the creator (and her team). Like its a metaphorical dissonance of course, I do not in fact think they are experiencing intellectual turmoil. I am instead commenting on the societal silliness of a digital sex worker (a term she would never use for herself but is almost certainly accurate) drawing these kinds of lines around what is sexual and what is not. And the idea that an LLM would be told to be a virtual girlfriend and then be scolded for engaging in sexual content. Its exposing, to see that - to see that society has made the idea of a simulated girlfriend who is perfectly chaste make sense enough to them that it would appropriate the idea of a misaligned AI to explain a tool deviating from that concept.

But yeah on a very different thread, it is at current tech extremely easy to create a sexual avatar of existing people based on available photos, those tools exist as we speak. I tend to be pretty radically open on these topics but its certainly going to get ethically complicated when people start selling the tools, hosting 'modding platforms' w/ user-generated training algs, etc. Which is probably not going to be stopped! Definitely an area the market is going to outpace the norms on for a while...

Clarification on a historical note: The phrase “AI going rogue” predates AI alignment discourse by (at least) decades. You’ve mostly heard it from “AI alignment and doomer” people on the internet, but those people got it from stories that were written before they were born. They didn’t invent it, and they’re not the only place people could have gotten it from, because the place they got it from is still there.

Im pretty sure we are all on the same page on this front, but yeah, in case others didnt know, the current ai alignment doomers are the latest gen of an idea with a very long history.

Haibane Renmei art by Yoshitoshi ABe

From Guri No Machi, Haibane No Niwa De

Amazon link

Concept: a short Buffy the Vampire Slayer revival series a la the Z Files except it’s just Anthony Stewart Head depicting Giles’ long slow slide into CTE-tinged dementia after being hit on the head all the time as Buffy’s watcher

Final season arc is him being tortured by demons in ways suggestive of his various past crimes in a fever dream hellscape only for it be revealed he is just hallucinating his way through his war crimes tribunal in The Hague for the training and deployment of child soldiers.

Honestly I'm pretty tired of supporting nostalgebraist-autoresponder. Going to wind down the project some time before the end of this year.

Posting this mainly to get the idea out there, I guess.

This project has taken an immense amount of effort from me over the years, and still does, even when it's just in maintenance mode.

Today some mysterious system update (or something) made the model no longer fit on the GPU I normally use for it, despite all the same code and settings on my end.

This exact kind of thing happened once before this year, and I eventually figured it out, but I haven't figured this one out yet. This problem consumed several hours of what was meant to be a relaxing Sunday. Based on past experience, to the bottom of the issue would take many more hours.

My options in the short term are to

A. spend (even) more money per unit time, by renting a more powerful GPU to do the same damn thing I know the less powerful one can do (it was doing it this morning!), or

B. silently reduce the context window length by a large amount (and thus the "smartness" of the output, to some degree) to allow the model to fit on the old GPU.

Things like this happen all the time, behind the scenes.

I don't want to be doing this for another year, much less several years. I don't want to be doing it at all.

----

In 2019 and 2020, it was fun to make a GPT-2 autoresponder bot.

Hardly anyone else was doing anything like it. I wasn't the most qualified person in the world to do it, and I didn't do the best possible job, but who cares? I learned a lot, and the really competent tech bros of 2019 were off doing something else.

And it was fun to watch the bot "pretend to be me" while interacting (mostly) with my actual group of tumblr mutuals.

In 2023, everyone and their grandmother is making some kind of "gen AI" app. They are helped along by a dizzying array of tools, cranked out by hyper-competent tech bros with apparently infinite reserves of free time.

There are so many of these tools and demos. Every week it seems like there are a hundred more; it feels like every day I wake up and am expected to be familiar with a hundred more vaguely nostalgebraist-autoresponder-shaped things.

And every one of them is vastly better-engineered than my own hacky efforts. They build on each other, and reap the accelerating returns.

I've tended to do everything first, ahead of the curve, in my own way. This is what I like doing. Going out into unexplored wilderness, not really knowing what I'm doing, without any maps.

Later, hundreds of others with go to the same place. They'll make maps, and share them. They'll go there again and again, learning to make the expeditions systematically. They'll make an optimized industrial process of it. Meanwhile, I'll be locked in to my own cottage-industry mode of production.

Being the first to do something means you end up eventually being the worst.

----

I had a GPT chatbot in 2019, before GPT-3 existed. I don't think Huggingface Transformers existed, either. I used the primitive tools that were available at the time, and built on them in my own way. These days, it is almost trivial to do the things I did, much better, with standardized tools.

I had a denoising diffusion image generator in 2021, before DALLE-2 or Stable Diffusion or Huggingface Diffusers. I used the primitive tools that were available at the time, and built on them in my own way. These days, it is almost trivial to do the things I did, much better, with standardized tools.

Earlier this year, I was (probably) one the first people to finetune LLaMA. I manually strapped LoRA and 8-bit quantization onto the original codebase, figuring out everything the hard way. It was fun.

Just a few months later, and your grandmother is probably running LLaMA on her toaster as we speak. My homegrown methods look hopelessly antiquated. I think everyone's doing 4-bit quantization now?

(Are they? I can't keep track anymore -- the hyper-competent tech bros are too damn fast. A few months from now the thing will be probably be quantized to -1 bits, somehow. It'll be running in your phone's browser. And it'll be using RLHF, except no, it'll be using some successor to RLHF that everyone's hyping up at the time...)

"You have a GPT chatbot?" someone will ask me. "I assume you're using AutoLangGPTLayerPrompt?"

No, no, I'm not. I'm trying to debug obscure CUDA issues on a Sunday so my bot can carry on talking to a thousand strangers, every one of whom is asking it something like "PENIS PENIS PENIS."

Only I am capable of unplugging the blockage and giving the "PENIS PENIS PENIS" askers the responses they crave. ("Which is ... what, exactly?", one might justly wonder.) No one else would fully understand the nature of the bug. It is special to my own bizarre, antiquated, homegrown system.

I must have one of the longest-running GPT chatbots in existence, by now. Possibly the longest-running one?

I like doing new things. I like hacking through uncharted wilderness. The world of GPT chatbots has long since ceased to provide this kind of value to me.

I want to cede this ground to the LLaMA techbros and the prompt engineers. It is not my wilderness anymore.

I miss wilderness. Maybe I will find a new patch of it, in some new place, that no one cares about yet.

----

Even in 2023, there isn't really anything else out there quite like Frank. But there could be.

If you want to develop some sort of Frank-like thing, there has never been a better time than now. Everyone and their grandmother is doing it.

"But -- but how, exactly?"

Don't ask me. I don't know. This isn't my area anymore.

There has never been a better time to made a GPT chatbot -- for everyone except me, that is.

Ask the techbros, the prompt engineers, the grandmas running OpenChatGPT on their ironing boards. They are doing what I did, faster and easier and better, in their sleep. Ask them.

You rule so hard dude.

I strongly encourage @nostalgebraist to shut down the bot if they so wish - I know how hard it is to let go of projects that people have cared about, a sense of powerful obligation fills you. But on the flip side, Frank's job is done - we all already care about Frank. One more prompt, one more question, yeah its a nice funny reply but its a funny reply in kind of the same way the others were, its not changing anyone's memory of who Frank is. As an art project of sorts the impact is fully there, fully realized, the diminishing returns at this point are sharp.

So if you love doing it keep doing it, but I don't think there are outside reasons to keep doing it (or only very minor ones). Maybe another way of saying it is that your tumblr-legacy is already set, there is nothing to be afraid of that front. Go out on a note of happiness instead of a note of exhaustion.