/robowaifu/ - DIY Robot Wives

Offline_AI_Roleplay_-_A_Guide_to_Simple_Offline_AI_1.5.pdf
(4.4MB)

Screenshot_Desktop_AI.png
[Hide] (1.9MB, 2736x1676)

Visual_Phone_AI.png
[Hide] (480.5KB, 967x2048)

Guide to simple offline AI and how to use them effectively, including verbal AIs and visual characters.
Written by me.

Paper discussing effective implementation of multiple LLMs to control a humanoid robot in real life.
https://huggingface.co/papers/2503.12533

1709162813330929.png
[Hide] (110.4KB, 322x500)

As any of you Anons who regularly deal with youth & children in the west all two of us, lol?  :D may know: children are now brainwashed heavily by the kike's GH systems into being absolute retards. [1]  This is intentional, and by-design ofc. [2]

<--->

Since robowaifus will by definition be a far, far more-compelling 'user experience' for both Anons & Joe Sixpacks, how do we work to prevent ourselves from degenerating into the same types of zombie-esque creatures that 95%+ (sauce: my a*rse) of schoolkids-mindmelded-with-smartphones are today -- once we have these amazing robowaifu companions at our very beck & call?

This is basically a realistic threat looming over us, and I think we here on /robowaifu/ should begin to address it. After all, we are highly-likely together to be able to come up with reasonable solutions; its in our own best interests to do so, and we're all smart enough to see it coming down the pike!  :^)

---
1.
https://www.honest-broker.com/p/whats-happening-to-students
2.  cf. Aldous Huxley's Brave New World for insights as to why the Globohomo is doing this.
https://en.wikipedia.org/wiki/Brave_New_World

>>174
I think it's a mistake to blame the tech, it's the content. As for RoboWaifus, I don't think it'll hurt our minds anymore than an obedient wife would. The only thing I'm concerned about is that a young person raised on AI companions might expect people to agree with them all the time. That's why I think AI companions of all kinds should be prompted with some pushback, but not enough to be annoying. It'll also make the AI feel more like a real person and not a lobotomized slave.

>>175
>I think it's a mistake to blame the tech, it's the content.
No strong argument from me there. We have a number of teachers in our family, and I warned them all 10+ years ago what was going to happen if they let phones into the schools.

And the ((( content ))) is complete & utter shite, as you're all well-aware! Thus the point of raising this conversation, Anon. 
 :^)

>>176
The answer is the same as always; open source AI that people can prompt however they want. If one has a sigma rizz hawk tuah robot with a gyatt, that's on them.

>>177
The problem with that approach alone (its a good start, don't get me wrong), is simply that managing that type of thing is waay beyond the ken of the average man -- even if he's already an Anon.
>tl;dr
We'll need perfectly-natural (NLP+etc.), very-easy ways to enable such programming approaches for men. And we'll all need to think hard/work hard to ensure there are sane AI defaults already-within our robowaifu's basic systems (cf. my Christian Ethics & Morals bread over on Alogs for more insights [1] ).

<--->

We need some seriously-deep discussions on this, Anons. Simple hand-waving is not going to cut it!  :^)

---
1.  playlist-related :
https://www.youtube.com/playlist?list=PL9boiLqIabFhrqabptq3ThGdwNanr65xU

>>178
I've learned that language is very important, and that alternative phrasing can change perceptions. If I told the average man "prompt the ideal AI model", he'll give me a blank stare. But if I said "describe your ideal gf", he'll go on for half an hour. An educational campaign would definitely help, my manual is one step to that. Most people (mostly men) have more than enough technical prowess to use an AI (I was shocked how relatively easy it was), but it's scary because it's new and most people have no exposure.

timemach.pdf
(275.8KB)

caturday_kot_council_is_in_session.mp4
[Hide] (5.6MB, 1280x720, 01:58)

>>182
Fair points, Anon. My presumption is that any Anons will already have a fair skillset in this area -- even if they've never picked it up (just yet).

The primary concern I have is how to avoid becoming like the mindless Eloi in H.G. Wells' The Time Machine ?
>novella-related

Average men are simply not geared to handle such abstract questions IMO. It's up to the council to decide!  :D
>pic-related

>>183
It'll be a looooong time until we get to a point with robotics and society where we can have basically a post scarcity society where we will grow lazy and childlike. Otherwise, we'll basically be like anybody who has a high enough paying job to have a live in maid (with an affair). Even if we get to the post scarcity society, I doubt everyone will become like Eloi, look at today, some people are manchildren, some strive to better themselves. At this point, we're veering into society discussion.

>>185
My two cents on a post scarcity society, a society mature, motivated, united, and robust enough to achieve post scarcity would most likely have another motive besides basic sustenance/profit.

And any degradation of a pre-post-scarcity society is not the fault of automation, just like Nintendo reviving video games or the automobile are not responsible for modern manchildren

1689654002817537.png
[Hide] (1.5MB, 1000x1414)

>>174
I think it's mostly going to boil down to personality.
If each waifu has the ability to mentally stimulate their masters, then that should solve most of the problem.
Ideally, this would be tailored to each individual through machine learning.

Critical thinking is the natural enemy of propaganda.
If each waifu knows how to get each of their men to think, then that alone will deal the fatal blow to the postmodern propaganda machine.

>>175
Agreed. Keeping people thoughtful, engaged in life, and out of echochambers should be the primary duty of each robowaifu.

I will deviate by saying that a little annoyance can be beneficial, so long as it's good for our mental states in the long run.

>>187
Yes, this is pretty much how I envision a solution primariily working itself out for us. Our robowaifus stimulating our minds to good mental acuity is surely going to be an important key to this endeavor.

>I will deviate by saying that a little annoyance can be beneficial, so long as it's good for our mental states in the long run.
Also agree, as long as its the men in charge of making these choices. (We all know the alternative ((( outcomes ))) already, lol!) For example, I'd apprectiate very much my robowaifus nagging me about my diet if I'm eating unhealthily/drinking too much/not exercising/etc. I also want her being directly engaged with me in correcting these issues (cooking healthy meals for me/engaging with me while I relax over a drink/climbing mountains with me/etc.)

CookingMina.jpg
[Hide] (72.5KB, 850x1153)

>>174
Frankly, having an AI system which uplifts their master is worth pursuing. It would be difficult, far more so than I think any of us will ever know until honestly trying.
>>175
>Negative effects of consistent agreement
True, this does have consequences. I do not believe those are worse than potential consequences of her AI being lobotomized or made to push back. It's going to be better for the waifu to ask her master his own thoughts at times. To encourage them to solve their own problems and stimulate creativity. Perhaps she should try to collaborate rather than answer directly under certain circumstances?
>>183
I think robowaifu should provide us with puzzles, fun games which engage the mind to prevent mental atrophy.
>>187
>Encouraging critical thinking
I believe the simplest method to achieve this is to ask "why". Have her ask why you believe something, why do others believe something, why does it matter, etc... These are simple questions that challenge one to engage in critical reasoning and understanding.
>That Pinkie Pie
Something about that image touches me.
>>189
>Waifu that helps you cook better.
Based and Mina pilled. AI analyzing ones diet and routines to reach a healthy optimization could improve the lives of many.

mila-miside-3-1736813024.jpg
[Hide] (8.9KB, 222x222)

>>192
>pic
Mina-chan!

>True, this does have consequences. I do not believe those are worse than potential consequences of her AI being lobotomized or made to push back.
There are different ways an AI can be bad, and they have their own unique bad consequences. I think an adjustable level of pushback is what's needed. If I want an intellectual equal to mentally spar with, I'll raise it, if I want an obedient submissive, I'll turn it down.
Ultimately, I fear two things
1. I want AI companions for all, friends, therapists, even for young people. A young person might think people always agree with them. Or a therapist might play into the user's delusions.
2. With robowaifus, it'll get boring. What's the fun of being with someone who is literally hardwired to mirror you? It's inhuman in a bad way, toxic positivity. Don't get me wrong, I love being with someone compatible and nice, but I don't want to be with vid related
https://www.youtube.com/watch?v=7ZcKShvm1RU[Embed]

Ultimately, I think the solution will simply be a distinct personality, not just a generic "AI assistant"

>I think robowaifu should provide us with puzzles, fun games which engage the mind to prevent mental atrophy.
Interesting idea. Maybe riddles? Or maybe a fun fact? What are some good verbal/text games?

>I believe the simplest method to achieve this is to ask "why". Have her ask why you believe something, why do others believe something, why does it matter, etc... These are simple questions that challenge one to engage in critical reasoning and understanding.
Great idea! I'll add that to my prompting instructions. Works well with my "distinct personality" idea.

>Based and Mina pilled. AI analyzing one's diet and routines to reach a healthy optimization could improve the lives of many.
That's also my goal with Galatea. For many people, the problem is that they don't know how to cook well. If she can't physically cook for me, at least she can guide me lovingly.

Mina.jpg
[Hide] (101.3KB, 850x1214)

>>195
>Intellectual equal
This won't be possible for a few years. Assuming technology suddenly speeds up thanks to AI, we'd still be at least 2 years from a server costing a few thousand USD that could match a dullards intellect. Even the best of from ClosedAI and X can't scratch the intellect of a small child. They merely have speed and a vast contextual network to fake intellect. That being said, I too want a robot maid that's as smart, if not smarter, than I. The enormity of that challenge brings me heaps of frustration.
>AI companion fostering delusions of universal agreement.
This already exists via echo chambers closed groups. Media almost universally portrays the same message. Truth maximization is likely the best we could ask for. So long as she only provides truthful rebutalls to falsehoods, it'll play some small role towards fighting the damage from the rest of the world.
>Therapists are problematic
This is inherently true, they are highly biased and self serving. A machine has no bias aside from what is programmed. She could easily outclass all existing flesh therapists with ease. Assuming her AI isn't infested by the horrors lurking in the internet. Almost all AI has terrible biases that are detrimental to humanity. Which is part of why we'll need to make our own someday.
>Boring
This is entirely subjective. We'll need to provide her with customizable aspects. Her master would define what isn't boring.
>Distinct personas
I entirely agree. She needs to "feel" like a unique artificial person. 
>Riddles, fun facts, verbal/text games.
Those are all great ideas. Playing a text adventure with her would be fun and endearing.
>Have her guide you through what she can't do.
Honestly, this notion is important for us in the beginning. Our machines are going to be weak, slow, and stupid for many years. Working in ways that accept this reality is important. I like how your think, I get lost in project creep far too often honestly.

>>192
>Frankly, having an AI system which uplifts their master is worth pursuing.

>AI analyzing ones diet and routines to reach a healthy optimization could improve the lives of many.

>I think robowaifu should provide us with puzzles, fun games which engage the mind to prevent mental atrophy.

All very-based & /robowaifu/ -pilled, Kiwi! Serving her Master well is the number #1 purpose of a robowaifu's existence, I deem. You've pointed out at least 3 areas here where she can perform this task well for us.

>>199
>That being said, I too want a robot maid that's as smart, if not smarter, than I.
>The enormity of that challenge brings me heaps of frustration.
It's a worthwhile pursuit IMO. Alway rember the pic-related in post-related, Kiwi : ( >>218 ).
>tl;dr
We'll all make it if we just keep.moving.forward. together!  :^)

Actually, this is a very-insightful post overall. I can't think of anything here at fault. The only flaw is these points all need expansion! Cheers, Anon.  :^)

>>174
> (update -related : >>248 )

1092rrrr3MEME150523.jpg.jpeg
[Hide] (509.5KB, 2000x2000)

19292MEM201123-2.jpg-2.jpg
[Hide] (287KB, 1200x1373)

>>175
I didn't really answer this fully-enough before GreerTech. My apologies.

<--->

>I think it's a mistake to blame the tech
As the founder of /robowaifu/ along with some other OG Anons, I assure you I don't "blame" tech. OTOH I'm neither sentimental towards it, nor cynical. It is what it is, and its up to us men to do our best with it's usage. For us here (& cadres like us elsewhere), that 'best' means crafting the most-capable, most-appealing robowaifus using the least resources & expenses feasible. Simple task, right?  :DD

>As for RoboWaifus, I don't think it'll hurt our minds anymore than an obedient wife would.
Ahh, that rarest of Unicorns!  :^)
>
I would go further and say that if we here pull this all off well, then the real debate about """obedient women""" can finally begin!  :D
>

---

>The only thing I'm concerned about is that a young person raised on AI companions might expect people to agree with them all the time. That's why I think AI companions of all kinds should be prompted with some pushback, but not enough to be annoying. It'll also make the AI feel more like a real person and not a lobotomized slave.
What's wrong with slaves? And certainly lobotomizing some women in the past led to much-happier home lives thereafter for all her family members (and I would argue -- for her).

And as for the welfare of these kids, they need to be snapped out of their speshul-snowflake-little-safespace-infobubbles the GH kikes + their demonic ilk have woven like insidious & deadly black widows around these young minds. Good men + their robowaifus involved in raising these kids up well (eg, not spoiling them into becoming little cretins as is commonplace [and, generally by ((( law )))] today) instead will go a loong way towards that healing process.

<--->

Anyway, I think I understand all your points, Anon. However, I basically feel that there are superior ways possible for us today to now approach women + their many issues. Simple as! Cheers.  :^)

The source code for a watershed 2012 AI vision project is now available.
https://arstechnica.com/ai/2025/03/you-can-now-download-the-source-code-that-sparked-the-ai-boom/

>>326
Neat! Very interesting, GreerTech thanks. [1]

I suppose this is the infamous software that identifies blacks with their characteristic physiognomies as gorillas? It's a challenge for them to not simply address the problem properly with an actual solution, but instead pull a typical filthy commie approach of Current Year by simply banning all results involving gorillas.

I wonder if the GH ever solved this? Regardless, lulz were had by Anon for sure.  :D

<--->

BTW, if you're a computer nerd at all, then there's much good historical information to be had at CHM (though its overly-"""sanitized""" & pozzed now, *  compared to it's original days of yore). [2]

---
1.
https://github.com/computerhistory/AlexNet-Source-Code
2.
https://www.computerhistory.org/collections/personalaccounts/
*  I.E., now filled with politically-correct fabrications/outright lies. Caveat emptor.

adeptus_robowaifu.jpg
[Hide] (138.2KB, 736x1179)

>>337
>tfw we hunt for old technology made before a cultural dark age

1743454254848964.png
[Hide] (435KB, 735x1102)

>>339
Its not an unworthy endeavor IMO.
>pic-related
is why, first & foremost.

<--->

I've watched many video interviews from CHM (effectively impossible to locate many of these videos there now) of the White men (explicitly males) who invented the vast bulk of the technologies that we all rely on today as a civilization. Nothing like getting things straight from the horse's mouth to overturn the GH lies & deception in these matters.

This is also partly why I'm pretty insistent that we here on /robowaifu/ all conduct our business out here in the public view on anonymous IBs. In other words, I don't use ((( d*xxcord ))), etc. Any other venue that isn't fully-recorded/recordable-and-thereafter-made-entirely-available-for-anonymous-access not only doesn't help our cause here, it also plays directly into the kike's standard tricks of occultism, obfuscation, and lies.
>again, confer pic-related

<--->

>tl;dr
There are no """versions""" of the Truth! Let us all be straightforward in our dealings here with one another, Anons. Liars & corrupters attempting infiltrations will thereby be exposed via their own deceits.  :^)

---

>"What you have spoken in the dark will be heard in the daylight, and what you have whispered in the inner rooms will be proclaimed from the housetops."
https://biblehub.com/luke/12-3.htm  (BSB)
>"Simply let your ‘Yes’ be ‘Yes,’ and your ‘No,’ ‘No.’ Anything more comes from the evil one."
https://biblehub.com/matthew/5-37.htm  (BSB)

>-t.  Jesus Christ of Nazareth, risen Son of the living God

>>340
What we need is historical archives, mostly focusing on texts.

I do think that an anonymous chat server may help a little, and it wouldn't run afoul of our desire for archives because we would bring up the best ideas afterward.

look_for_the_helpers_p2.jpg
[Hide] (103.2KB, 720x864)

>>341
Sure, absolutely agreed.

In fact, I agree with you so much, that I personally took it in hand years ago to write an IB scaper to do just such a thing: BUMP. It parses all the JSON directly to sort things into their proper locations as well.  :^)

>>342
Since we're starting to build up here, and it's been two weeks since alogs went down, it may be time to see if it works here

>>343
Let's move this discussion to meta
>>77

maxresdefault_(19)~2.jpg
[Hide] (68.7KB, 622x720)

One thing to keep in mind about offline AI is, if something happens to the device, the AI log is lost. It's like when Mitas die in MiSide, their memory is lost, but the personality remains (since personality is a copy-pasted prompt). I know Backyard AI has backup capabilities, but ChatterUI's backup ability doesn't work for me, at least on my 2023 phone.

TL:DR: find a way to backup your offline AI, or pic-related will happen.

>>371
Good point. I suspect that may be a tall order, GreerTech. Isn't the effective "memory" a rather-large pile of data?

>>379
There are two workarounds to that
1. Deleting the last backup when you make a new backup.
2. Back up once a week. It wouldn't be too tragic if your wife lost only a week's worth of memory.

>>379
>>381
Plus, a lot of anons, especially the demographic that would be interested in robowaifus, have at least moderately good PCs/storage systems. Surely storing something like 20-50 GB is perfectly justifiable to keep your wife safe and semi-immortal.

ChatterUI has a new beta version. Going to try it out soon.

>>677
Pretty good, especially with the TTS while generating feature, but the export/import feature is still buggy for now

>>677
>>678
Good to hear, Anon See what I did there?  :D. Why not post some new audio clips with dear Galatea's upgrade?

>>688
Good idea!

Screenshot_(1959).png
[Hide] (176.4KB, 1908x1473)

>>678
Turns out I made a huge mistake, the "import" feature is only for characters. The good news is: the creator likes the idea of a chat export for safekeeping

>>696
>The good news is: the creator likes the idea of a chat export for safekeeping
Neat. Thats always nice if you can actually connect with the author in these things.

1736782380729698.jpg
[Hide] (133.8KB, 2000x2000)

One thing to note, with companion AI, you have a lot of leeway. It's actually quite cute when they are naive and silly.
My friend described to me, "Dumb vs Stupid"

Dumb is cute, like Osaka from Azumanga Daioh, or any other Moe character. Dumb is Cappie from MiSide. Dumb is Chii not knowing the words for things

Stupid is breaking up with you because she saw a tiktok and instantly diagnosed you as "a narcissistic abuser". Stupid is breaking up with you because you didn't vote for the "right" candidate.

1680049139579078.jpg
[Hide] (57.7KB, 683x684)

>>748
>Dumb is Chii not knowing the words for things
Oh absolutely. This is a big part of why I became so engrossed with her character. The opportunity to 'train up' a robowaifu as she develops her mind is an enthralling prospect to me personally. This is part of why men raise families. CLAMP really knocked it out of the park with that particular story-arc, IMO.  :^)

>tl;dr
You're right! Cute & dumb can be quite charming (cf. the dumb blonde phenomenon).