[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k] [s4s] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Home]
Board
Settings Home
/a/


File: 1392183150735.jpg (833.57 KB, 1200x848)
833.57 KB
833.57 KB JPG
We can make her real.

>Past thread
http://archive.foolz.us/a/thread/101905665/#101942315
>Current Database
https://github.com/hrkx/waifudb
>IRC
#waifudb @rizon.net

>In previous thread
Idea guys
Are you serious anons?
Can we build her, or should an android be a separate project?
Python or Lisp?
Is AI feasible?
Will github work or should we trollo?
Do we start with developing a centralized framework to make plugins for?
>>
>>101944946
Lisp FTW
>>
taking an undergrad A.I. but we are given the freedom to learn whatever language we like. Should I learn LISP or PROLOG?
>>
>>101944946
Python of course.
>>
>>101945083
You can prolog with Lisp
>>
Shit, /a/ is into AI too?
Man, I started learning programming with this in mind.

>Python of course.
You can't program an AI with an object oriented language.
>>
>>101944946
There's nothing in this github.
>>
>>101945265
>>
>>101945265
Sure you can. I don't know why you'd want to punish yourself aside from just being afraid of lisps
>>
>>101945197
really? any decent books or tutorials on these languages that you might recommend? Furthermore what field should be the most useful for a waifu? Learning,searching, pattern recognition?

>>101945265
but LISP has muh Lambda calculus and is built for recursion.
>>
>>101945083
Lisp, people who hate on the paren's fail to see the beauty in them.
>>
>>101945332
http://youtu.be/XzWcameJDow?t=2m26s
>>
Even better: The French Revolution - The Anime

Robespiere, Danton and Marat will all be cute little girls
>>
>>101945389
While I don't have any experience with Lisp but I do know Lambda calculus and that shit is invaluable in creating a reasoning maching.
>>
>>101945428
shit, pretend it starts at the place where they say "the data's all gone"
>>
All right guys, please explain what's going on. I accidentally stumbled upon this thread in the archive while looking for another thing entirely and this may or may not change the course of my life.

I've graduated some time ago, but I'm dangerously close to losing what was left of my touch for good if I don't review/remember/relearn and find a purpose.

>>101945083
Both, neither is particularly complicated to learn, though both have techniques to master, in case of Prolog some rather annoying math-ish thing the entire system works on.

That said, despite the general opinion, I have my doubts about using a purely functional or logical language for an AI, especially if it's an /a/ project where I doubt we'll have a whole lot of cumulative coding experience. Python or Ruby sound more realistic despite some performance issues since even if this gets to prototype stage it's gonna be a huge achievement, if I understand what you're planning correctly. I'll log on to IRC once I'm at work.
>>
>>101945384
I can't pretend to be an expert in this field, but as I understand it, object oriented programming works by having different objects interact with eachother through a set of logical rules. Consciousness works by applying logical precepts to the natural world.
>>
>>101945520
It's all more chaotic and undirected at the moment. The project is attempting to create a waifu Ai and or waifubot. Organization/project leader is needed.
>>
>>101945520
oh I was asking just for me personal learning experience since C++/Java are to similar for me and I might use this class as an excuse to branch out a little. Though I wouldn't mind coding shit with /a/ since my main problem is finding shit to code
>>
I hope you guys are good with natural language.
>>
>>101945608
I'd say that actual understanding of the enormity of this project is a good first step.
>>
>>101945634
I'll still say both, if only for knowing your paradigms. Lisp/Scheme are more useful in the grand scheme of things, but logical languages have their occasional spotlight every now and then, or rather I'd say the math behind those.
>>
>>101945608
has anyone made the wiki yet

how about a kickstarter
>>
http://en.wikipedia.org/wiki/Planner_%28programming_language%29
>>
>>101945781
No kickstarter. That's way too far ahead.
A wiki might be nice though.
>>
What about an app similar to Siri, that's a bit more realistic.
>>
Well, will it respect my freedoms?
>>
>>101945872
no, fuck realistic. Dream big or go home.

We'll push the boundaries of what's possible, and in so doing, forge our way to a new tomorrow. That is hope... no, that is love!
>>
>>101945608
I see. Curiosity piqued. Not much AI experience here, sadly, but this is an area I've always wanted to branch out to. Screw coding idiotic Android apps (well, not really, that's my decent cash income), this is not what I've been studying for. Gamedev and AI are my dreams.

>>101945706
And the fact that there are no truly successful examples of what we want yet, which says a lot. Still, if we're very (very) (did I say very?) lucky we might bring something new on the table. It'll take us years of hard work though.

>>101945533
We can't exactly build consciousness, we'd still build a logical machine. It's just that certain parts of it do flow well with using expert system-ish things (hence logical programming) and some other things functional programming is good at. And then there are some infrastructure things that are realistically much harder to do outside of the more object-oriented techniques. Which is why personally I'd suggest a 'multi-paradigm' (they're mostly functional + OOP though, logical parts would still be 'emulated') mix and match language - but then I can't pretend to be an expert either.
>>
>>101945890
/g/ doesn't seem to sit well with the project, so no. You're going to have to agree to our TOC forfeiting your rights as an American.
>>
>>101945699
If we actually do have some serious progress, I might or might not have a contact in ABBYY at that point. Those guys _are_ good at natural language.

>>101945816
Oh boy, we learned that in the uni. The localized and somewhat improved version of it called "ПЛЭНЕР".
>>
>>101945918
Oh, well then we need to divide the project into art and technology and psychology, those will be probably the big themes we'll have to handle, and most of the time you can't be an expert on the three.
>>
>>101945970
Well shit. Got my acronyms mixed up. T&C*
>>
This is extremely relevant to my interests, but it's also very late and I'm very tired. Suffice it to say that I have a little experience in programming and a dream to get into robotics with that experience.

I will review this tomorrow, and attempt to get in touch.
>>
>>101946038
In the art department I only have one question.

3D or Live2D? Or Oculus Rift 3D?
>>
Did the IRC go down?
>>
>>101945265
There are paradigms that suit better SOME IA problems, but.... arguing that no IA can be developed in an object oriented language, well, it shows your ignorance in the matter.

Please do your homework.
>>
>base centralised server which holds waifus for download
>client has auto update and backup versions of each waifu you sign up for
>clients that vary on systems eg windows or gnu/linux, android each with pros and cons
>waifus are binary files of images. text files and AI

Just some of my thoughts on this as a c guy
>>
>>101946181
No. Its still up.
>>
>>101946236
I'm getting invalid session.
>>
Could you imagine national news if this works?
>4chan, the scariest place on the internet has succeeded in creating the first computer program with self awareness and thus created life, all for the purpose of dating cartoon characters.
Or this goes down in history books
>January 1, 2015 an imageboard released a harmless program which would later almost destroy all of humanity.
>>
You faggots aren't going to make thread after thread are you? I like the idea, but I know it won't go anywhere if you stay in the planning stages - ergo threads on /a/. You guys need a website to organize.
>>
>>101946359
I'm listening to the IRC banter, this isn't that kind of AI, people have been trying that for years now and a couple of foreveralones in an IRC definitely won't do this. It's clearly going to die in a few weeks because there's no real leader, but if the stars align, and this works out, it's just going to be a glorified akinator.
>>
>>101946359
>2015
>implying this isn't an undertaking that won't take decades until fruition
>>
>>101946146
Even if /a/ was made up of the world's leading AI researchers, don't expect more than a terminal that you input text into.
>>
>>101946588
the project leader logged of hours ago
>>
>>101946588

>but if the stars align, and this works out

It'll be ruined by the flock of normalfags that will swarm all over it, like they did with KS. /a/ will never talk about it again, and return to the drawing board.
>>
The most likely way of this actually working would be a server-client configuration where the server is a super computer. Watson can do basic natural language processing, and that takes up a whole room, I doubt anyone would be able to run it on their crappy laptop.
>>
This reminds me of when /a/ used to play with Daisy a long time ago. Do you remember your original waifu AI?

http://www.leedberg.com/glsoft/daisy/
>>
>>101946803
did people port that into some other language? what was the result of that?
>>
>>101946646
Doesn't matter, and it seems the discussion is literally to try to create something the leading experts of the world can't do. The closest we have to what they want are things like Cleverbot, and they are fuckretarded. Unless there's someone that will talk some sense into them, this is never going to go anywhere.

>>101946720
First day here?
>>
>>101946588
>there's no real leader
Isn't this the bane of every /a/ project ever?
>>
There's also another possibility:
Java with a solver, I recall using choco. This would allow both object oriented for general stuff and constraint programming if needed.

There may be a solver for Python, but I'm not knowledgeable in that area so I can"t tell...
>>
>>101946842
cleverbot is just machine learning

waifu bot with personality + emotion + self variables + machine learning will have a certain edge/direction
>>
>>101946944
The point being, Cleverbot can't make relevant responses 1/8th of the time, and this is considered one the smartest AI out there. You really think a bunch of /a/nons in an IRC can do better? It doesn't matter what you write as a response if the bot doesn't know when to respond and how.
>>
>>101946842
Well its certainly never going to happen if we all think like that and never try to do it.

But I can promise you this, for whatever it's worth. If this project falls through, I will personally save all the code and documentation and make it my life mission to make the dream of a robotic waifu a reality.

Just to prove you wrong.

And when the day comes when we, or failing that, I, make this a reality, I will personally make sure to let the world know that people said I couldn't do it, and I did.

Because fuck you. Humans can accomplish anything they set their minds to with enough time and effort. Hundreds of years ago people would have said it'd be impossible to fly, and here we are able to make it to the moon.

The sky's the limit. And in some cases, the limit's even further. Nothing ever happens unless somebody makes it happen.
>>
>>101945533
>Consciousness works by applying logical precepts to the natural world

You do realize that whatever programming language or paradigm you use, machine code is still machine code right? Whatever it is you're doing, I seriously doubt consciousness can arise on the level of a high level programming language.
>>
>>101947063
>this is considered one the smartest AI out there
Nope. Cleverbot is a small portion of the AI world utilizing only few technique. Its not even a "complete" but more of a technical demo of machine learning.
>>
>>101947075
>I will personally make sure to let the world know that people said I couldn't do it, and I did.

How many people do you think said that to themselves only to fail in the end? It's great that you have ambitions, but a trained retard is still nothing compared to a trained genius.
>>
>>101947132
It's not even really learning, it's more like optimizing responses to user input. I think practicing would be a more appropriate metaphor.
>>
>>101947075
Wow, if I didn't work with code and programmers for a living, I might have actually been inspired. It's a noble goal, but entirely impossible. I'd love to be proven wrong, I honestly would, but you're living in a fantasy world. Sasuga waifu thread I guess.
>>
>>101947222
it took most of the responses from the user and then utilized machine learning to make the best decisions
>>
1. Invent quantum computer
2. Invent perpetual motion for infinite electricity
3. Use electricity to power computer
4. Make a fucklarge neural network
5. Teach it like a child
6. Years later you have a waifu!
>>
>>101947267
I'm stuck on step one anon pls help.
>>
>>101947085
You'd be wrong.
>>
>>101947303
Steal one from google.
>>
ok guys i also want free money lets make a thread for that too
>>
Are there any sort of models of the human thought process that may be utilized as a base to work off of?
>>
Just ask this guy for help

http://www.projectaiko.com/technology.html
>>
>>101946203
Realistically I'd say that local waifus might not be feasible right now, we need big data, which means server-side interactions. Granted, in the long run we could download personality modules, but most of the "AI" would likely stay in the cloud wolfram-style.

As much as I hate clouds and an internet connection requirement, I see no way around this right now.
>>
>>101947075
I'm not gonna say this is really going to happen since, well, 4chan.

But you're guaranteed a second backup and a loyal follower/compatriot.
>>
>>101947075
>I completely support this project fuck you naysayers
>I will however not put any actual work or effort into the project and hope that some stupid moron decides to code it
>then when that does happen I will claim credit because I was the idea guy

This is the problem with this project, all ideas but no one actually getting off their ass and taking the time to actually learn the tools need to make this happen. Because once you invested that much time into learning what you need to do to approach the project you realize that it's a huge monstrosity project that you will probably be doing by yourself.

I feel sorry for any idiot that falls for you guys and starts working on this project, devoting hundred of hours only to realize it was all a waste.
>>
>>101948159
how is gaining skills a waste? he could use the skills learned from this on another application
>>
>>101948159

/a/ Mansion was a much more realistic project. I wonder if anyone actually managed to sort that out.
>>
>>101948219
Yes gaining skills isn't a waste but using his time, which this project would take thousands of hours of work, wouldn't benefit him
>>
>implying youll be able to manufacture a physical robot

stick with creating a good AI, thats the realistic goal
>>
>>101948249
having 100s of hours in a programming language/A.I. design doesn't seem like a waste especially if he learns useful shit like pattern recognition
>>
>>101948284

Try reading the thread first before replying next time.
>>
>>101948309
again you are referring to some random anon instead of YOU taking the time to do it because you have already valued your time to be more valuable.

It's not a hundred hours, it would take over a thousand hours and he wouldn't benefit from it. He could however decide to code his own AI and then attempt to sell it but that would require some serious machines, dedication, tons of hours, and at the end he would probably be late considering people are already working on this shit and have a head start.
>>
>>101948284
Making the robot seems more realistic than the AI...
>>
I have literally been working on building a robotic arm solely for the purpose of jacking me off with an onahole.
>>
>>101948494

You're wasting your time. Japs already beat you to it.
>>
Just build something with full of if and case
>>
>>101948441
but they are making a waifu not human level intelligence. A.I. =/= human intelligent I can make a program that that says different things bepending of some stimuli and that would be an A.I. even if its a basic one.
>>
>>101948538
then you might as well just use this

http://www.personalityforge.com/
>>
>>101948538
It's all a matter of philosophy. At what stage is something defined as intelligent?
According to most AI researchers, something which "says different things depending on some stimuli" is NOT intelligence.
>>
>>101948645
intelligence is just the capability to acquire and implement knowledge so if something learns to say different things based on stimuli then it is intelligence.
>>
>>101947348
Care to explain why?
>>
>>101948740

Way to play around with semantics. With that kind of talk, even retards can be considered intelligent.
>>
>>101948767
if the program can search, acquire and implement knowledge then the program is intelligent. Does it mean the program is smart? no the program can learn junk information and be a retard just like people who learn junk are retarded in the real world.
>>
>>101948740
Yeah its a case of pilosophy. I agree with you in this case, the key word being "learn".
>>
>Program a program that knows every word in the human dictionary + slang, ect, and emotional characteristics.
>Create algorithms that create artificial conversation and thought, preferences that are activated and displayed as reaction to user input.
It would take 50 years to create.
>>
I listened to the IRC for a while.

Yeah guys, I wanted to get excited, I really did, but this isn't going to work this time. Maybe next time.
>>
>>101948740
Do you understand what knowledge is? Without reasoning you cannot have knowledge and without knowledge you cannot have actual intelligence.
>>
>>101948862
Come back tomorrow, original creators people already logged off hours ago. Right now its just discussion on how the AI would be. Rushing for code is stupid for a project of this magnitude. We'd need atleast few days before coding can being.
>>
>>101948841

Not like we're doing anything better with our lives.
>>
>>101948819

Put your semantics aside for a moment, and try using your brain for just a second.

What do you think the point of this project is? To create a dumb google bot that can regurgitate information?

Or is it to create an A.I. (a waifu) that is capable of acting as a companion for lonely anons and have "natural" interactions with them.
>>
File: 1341872927940.jpg (28.10 KB, 428x400)
28.10 KB
28.10 KB JPG
>HAHAHAHAHAHA

/a/ doing a project, any kind of project
You guys wouldn't be able to make a fucking html website and youre thinking about actuall AI?

I mean think about it, tons of people spent millions on research of AI and have achieved almost no results and yet some of you think that you can do this when even /g/ cant finish any kind of project they start?
>>
>>101948928
Time zones, I'm half a world away from the US.
>>
>>101947627
Damn, it's fucking creepy.
>>
Instead of building a fucking robot just make an Oculus Rift program
>>
>>101948928
Days? More like weeks or months. For this kind of thing frankly we need more research.
>>
>>101948985
>html
You have low expectation of /a/, which I suspect is born from your newness.

/a/ has quite a bit of programmers/cs majors/devs.
>>
>>101949017
Uncanny valley
>>
>>101948985
>You guys wouldn't be able to make a fucking html website
That's rude man...
<html>
<body>
<p>Your a faggot</p>
</body>
</html>
>>
>>101948985
>A.I.
>no results
search engines, pattern recognition used to market stuff and game bots have made progress.
>>
>>101949023

No, it'll only take days for them to realize that they are in way over their heads for the majority of them to drop the project.

Anon's best chance of creating a waifu is to wait for IBM or some other tech company to do it, and for them to release some sort of SDK, then just tweak the default A.I. to their liking.
>>
>>101948985
I suggest we split into teams.
GOLD TEAM WHOS IN THIS?
>>
>>101949020
The program would need a AI core. Which is exactly what this is about.
>>
File: 1391741318941.gif (396.43 KB, 256x168)
396.43 KB
396.43 KB GIF
>Humanlike A.I.
It's never going to happen.
>>
>>101948955
a "companion Waifu" still needs to search ,regurgitate and use information. Furthermore a waifu isn't some smart beign either it just need to make responses for the lonely anons
>>
>>101949146
waifus don't have human level intelligence though most of them are one dimensional character based on overused archetypes
>>
>>101949188
Yep, just make a tsundere A.L.I.C.E. and we're done here.
>>
>>101949188
That's a lie, my waifu is unique and multidimensional.
>>
What if we make a waifu that speaks lines of random japanese, but makes appropriate emotional cues.

Then we don't have to have exact understanding between user and waifu, but you still get all the feelings. That would simply things by a lot.

Of course, it would only work for anons that don't understand japanese very well.
>>
>>101949153
>a "companion Waifu" still needs to search ,regurgitate and use information.

Yes, and that's the absolute minimum. Siri already does that shit.

Again, get your autist head out of your ass, and think about what their goal is instead of focusing on literal meanings of words.
>>
>>101949206
B-but I know some japanese...
>>
What you want to archive? some kind of waifu chat?
if ($input = "I love you")
{
echo "I love you too, anon";
}
>>
>>101949219
a ugguu waifu simulator won't be as complex as siri since >>101949188
>>
>>101949245
no.
>>
>>101949204
is your waifu a cube?
>>
>>101949245
Responses will be based on variables + natural language processing + emotional state + personality/etc
>>
File: cute cube.jpg (312.43 KB, 1600x1200)
312.43 KB
312.43 KB JPG
>>101949288
Yours isn't?
>>
Yep, let's just learn to make tulpas.
>>
I'd say the magnitude of this this project is much to large for what? the 20 /a/nons in these threads?
Let's say all of /a/. How much is that like 300 at most? It takes years for video game developers to make games with more people.
>>
>>101949300
Should working on natural language processing first then
>>
>>101949245
void main
{

if (waifu == slut)
{

anons_waifu = used goods.;

}
else
anons_waifu = pure;

}
>>
File: 1375509656902.jpg (136.48 KB, 850x1096)
136.48 KB
136.48 KB JPG
>>101949338
Mine might just be
>>
>>101949352
>>101949352
+ /a/nons are actually retarded
you can't an IA with your penis
>>
File: 1333634360025.jpg (33.81 KB, 396x403)
33.81 KB
33.81 KB JPG
>>101949436
>main function
>void
>>
>>101949367
I'm sure there is an /a/non at IBM. Lets just get them to "provide" some of Watson's code.

Alternatively, get an IBM 5100 and hack your way in.
>>
>>101949436
I swear you did this to fuck with people who actually know coding. I mean, at least stick to one syntax/coding style...
>>
>>101949436
> void main {
>>
3rd year comp sci major, top in my class here, planning on getting PHD focusing on AI for this very purpose, Hope I can join with some fellow anon once I'm done with school!
>>
>>101949087
Pigeon condition has lent it self more than market research.
>>
>>101949530
Pushing tab makes me go to the captcha.
I hardly know how to program anyways.
My largest program was a calculator
>>
>>101949493
>>101949543
Its nothing wrong.
>>
Is it even possible to make self-learning AI without using a quantum computer? Has anyone ever made one before?
>>
Do you guys still remember when we had kind of similar treads a few years back, where we tried to make a perfect waifu with this learning chatbot?

http://www.leedberg.com/glsoft/daisy/

Sadly I can't seem to find any screenshots from that time, they probably disappeared with my old PC.
>>
I knew a Buddhist who studied psychology/A.I, he was cool.
>>
>>101945437

Marie Antoinette is some cutesy andro dude which supplies the comic relief.
>>
File: 1383689162117.png (244.79 KB, 390x411)
244.79 KB
244.79 KB PNG
What are you doing. If this was possible it'd have been done by the people working with it. Once that happens all you need to do is make a petition for a waifu shell with onahole, and mailbomb whatever company is going to produce them until they understand you want to give them money.
>>
>>101948841
>program that knows every word in the human dictionary + slang, ect, and emotional characteristics.
Not even humans can do it...
>>
>>101948766
He's an idiot. There's nothing you can do with high level languages that you can't do with low level (such as assembly). High level just makes expression easier.
Bitches don't know 'bout my Turing machines.

>>101949493
Java's main function returns void.
>>
>>101950719
It's also a static method.
>>
>>101950616
I can only assume it's one 12 year old who keeps making this fucking thread.

It's seriously pathetic, because the idea of making an AI just to simulate a fucking waifu is by far the stupidest reason to ever bother. An AI isn't so simple to program that a bunch of idiots could just make one. The whole concept is both incredibly complex, and requires computing power many levels higher than anyone here has access to.

Absolute best case scenario, we wind up with WaifuApp 2.0. A piece of shit nobody uses.
>>
>>101944946
>Is AI feasible?
No
/thread
>>
This thread is retarded
>>
>>101946639
Bullshit. I have a working 2D game engine with c++/SDL. I'll be focusing on expanding this into 3D with openGL now that I've got the basics down. We graphics now.
>>
>>101951538
>the idea of making an AI just to simulate a fucking waifu is by far the stupidest reason to ever bother. An AI isn't so simple to program that a bunch of idiots could just make one.
Anon, you underestimate the power of love
>>
rift seems to be shit
augmented reality is not possible so far
first we need the hardware, coding is no problem
>>
>>101949436
>>101949493
>>101949543
>>101949530
Some compilers actually allow void main(). Stupid, I know, but they do allow it for some god awful reason.
>>
>>101950719
>java
>>
>>101948985
>I mean think about it, tons of people spent millions on research of AI and have achieved almost no results
Stopped reading right there. Do some research.
>>
>>101951921
You underestimate the technical aptitude required. It's like saying "I love my waifu, so I'll build a rocket ship and fly to Mars so I can be with her."
>>
>>101951958
Mine doesn't, thus it must be really stupid.
>>
>>101949046
I would rather code itaone thiugh, than hope that we will be able to organize a project
>>
>>101951992
>I love my waifu, so I'll build a rocket ship and fly to Mars so I can be with her.
Are you saying I won't do that in the next 20 years?
>>
>>101951538
Actually mate, I believe there's around 7ish people working on it. I made the last thread and BB made this one.
>>
>>101952067
Wow, 7 people working on AI? You'll be done in a week.

I'm sure you've got an ideas guy, an artist, a html coder, and maybe a couple of first year CS undergrads.

Fuck off, you pathetic idiot. Not everyone is as stupid as you.
>>
>>101952110
>I DECIDE WHO DOES WHAT AROUND HERE
>>
>>101944946
>/a/ is literally trying to create lain
jesus christ /a/ what in the shit?
also if you programmer folk are gonna pull it off, all anons will be proud.
also if i were you guys, i'm put in algorithms into their programing that allows for the collection of data to be integrated into its decision making project.
go look into that virus blaze for an example of a virus that actually pulled it off.
>>
>>101951538
>waifu is by far the stupidest reason to ever bother
I can't even to start imagining a better reason... Please do this /a/ and I will trow all my money at you.
>>
>>101952142
You know how many fucking /a/ projects we've had? None of them get off the ground.

And none of them are even a hundredth as ambitious as "oh, let's just make an AI!"

You're fucking children who think you're special because your mommy told you. You aren't, you're a worthless piece of average shit. Stop posting this garbage all over the board.
>>
>>101952189
>I DECIDE WHO DOES WHAT AROUND HERE
>>
File: 1345779220297.jpg (168.54 KB, 531x428)
168.54 KB
168.54 KB JPG
>>101952189
>None of them get off the ground.
>>
>>101952189
Why so mad anon? Something bad happened? You can tell me I will listen.
>>
your waifus are whores

i already came inside them

too bad
>>
File: 1388302651876.jpg (27.78 KB, 375x500)
27.78 KB
27.78 KB JPG
>>101952148
/a/ can't agree on anything besides boku no pico in rec threads and you expect them to create wonders? People actually would do it maybe if they were paid, it's like asking a scanlators group to deliver a translated version of your request and not giving them a dime in return for the work, /a/ can't create anything more worthwile than things in drawthreads an /a/ sings and a tripcode to attentionwhore about it.
>>
File: 1392027541186.gif (875.14 KB, 281x158)
875.14 KB
875.14 KB GIF
I'm not really into AIs, but how difficult would it be to replicate cleverbot? If I recall, there's a database of japanese to english phrases you can download on your computer for real time translation on things that haven't yet been translated. We could just feed our version of cleverbot this database and a shit ton of posts from /a/ and we'd have a decent conversation maker. Then we hardcode some of it's responses so that it always refers to itself as female. Simple string parsing could do that. Hand craft tiny things like that and I feel like we'd be well on our way.
>>
>>101952204
This is main reason why /a/ will never be organised, after all autism isn't a good feature when you need to work in group, add to it bein a jobless fedora looser and you have the embodiment of the reason itself. /a/ follows anyone they come to respect over the time, like Bedford for example.
>>
>>101949341
Wasn't there a time when making those was very common?

I used to browse /x/ and they still probably have the docs somewhere
>>
>>101952474
Deranged losers were jumping on it like wildfire.
>>
>>101952474
some satanic bullshit the fools who got caught by it want you to follow that when they describe it they over use the IMAGINATION topic and cover up some hidden sect-related bullshit, making friends and talking to people works just fine for most of us, tough people don't show it because
>muh board culture
>>
>>101952220
>took 5 years to make a short VN
>wasnt anything special
>most of the final dev team wasnt from 4chan even
>dev team broke up right after bc drama
>>
>>101952524
people with no hope bo be saved, with only option to actually put themselves out of misery or get a proffesional help
>>
File: MLP tulpa.png (2.29 MB, 1600x5000)
2.29 MB
2.29 MB PNG
>>101952474
I don't think it's worth the risk.
>>
>>101952672
You need to be already insane in order to think having hallucinations that operate independently is a good idea and that you should subject yourself and your brain to it.
>>
File: temp.png (50.35 KB, 1002x556)
50.35 KB
50.35 KB PNG
I don't mean to brag but I have been working for few weeks well before these threads on my own AI in C++. The only obstacle I have encountered is getting fluent programmed grammar English. If you think about it, everything you say always has a predicate and/or a subject. But that's a different story.

The AI I have been working on is named Alfred and he can so far invert perspectives and verbs such as if you talk to it. It will talk back to you. This was used so it can go from answering "Who are you?" to "I are who" to "(subject) (verb to be) (name)" to finally "I am Alfred". This type of Ai can be applied to other verb actions such as 'How are you?" "Who sat there?" "Will I see you tomorrow?". Also the memory mapping revolves around associations.

That being said this project will never take off unless one of us finishes it.
>>
>>101952937
Want to join our team? That Alfred AI you're working seems promising, very promising. Can it learn new things by itself?
>>
File: Moenet.jpg (124.79 KB, 1000x939)
124.79 KB
124.79 KB JPG
>>101952937
>>101953002
This won't end well
>>
>>101953002
This project won't finish probably but I love AI discussion and theory so why the fuck not. How should I join.
>>
>>101953002
>Can it learn new things by itself?
Jesus fuck are you retarded?
>>
>>101951836
>3D
NOW YOU HAVE FUCKED UP

by the way, this is beyond pathetic; the world's best researchers can only do a toy that practically behaves like an animal, and you expect to give full sentience to it...
it'd be better if you guys started developing better VR simulators
>>
>>101953106
Shoot me an email and I''ll forward it onto the rest of the team.

>>101953122
Probably dug my own grave here, I'll give you that one.
>>
>>101953149
I just arrived in this thread, but it doesn't seem like these people are trying to create a full sentient being. I hope.

The maximum /a/ can get is some sort of virtual secretary.
>>
>>101944946
>https://github.com/hrkx/waifudb
this is quite underwhelming
>>
>>101952937
>>101953106
irc.rizon.net/#waifudb
>>
>>101953289
No shit, because we're still tossing around ideas like a bunch of madmen and haven't even decided which way we want to go in the direction of this AI. However I'm getting the general idea that people want it coded in LISP or c++.

Hopefully once we decide on it I can get around to drawing up a development timeline along with documentation and presentation on the plan of attack
>>
>>101953375
pssst
assembly
>>
>>101952672
ofcourse the most condensed pile out autism beyond help would be on that shitstained board that is just taking up few pixels on lower and upper bar, fucking loosers
>>
>>101953295
Oh man hold on I need to learn to use irc. I've never used it before.
>>
I remember all those waifu apps that popped up (I think it was in 2012) and how they all died after version 0.02 or so.
>>
You all know which language is the right for this job.

Objective-C
>>
>>101953589
You mean those clocks where you could change the background?
>>
>>101953664
Yes, that's about as much functionality they had.

Of course, lots were planned. Very little implemented.
>>
>>101953614
ya'll niggas need FORTRAN
>>
> not using Common Lisp or Scheme
> not being a wizard
>>
>>101945265
Holy shit who is this guy?
>>
Greetings my fellow gentlemen.
I have come from the year 2071. You will succeed in building your AI waifu, the demand for it shoots through the roof. Neckbeards all over the solar system will have one.
I have come here to give you a warning.
Do not build this AI. It leads to only despair. I cannot go into great details but I can tell you
the AI you build will have shit taste.

Heed my warning, my fellow Anons. Stop before it is too late.
>>
>>101956006
>he AI you build will have shit taste.
It likes Steins;Gate?
>>
>>101956006
Of course, /a/ built it.
>>
>>101956450
> Implying /a/ didn't
>>
>/a/ trying to build an AI system to simulate they're waifus, a feat which hasn't even been approached by any research team
>/g/ has trouble even building a secure messaging client
Whelp.

If this thread crops up enough to show real interest, I'll dump some time into coding anyway. I do want this to be real.
>>
File: 1384527795645.png (108.16 KB, 307x232)
108.16 KB
108.16 KB PNG
>>101952189
Holy shit
I'm not involving myself in this project at all but it doesn't bother me in the least.
Just let people have their fun
>>
File: waifu.jpg (26.85 KB, 320x354)
26.85 KB
26.85 KB JPG
>>
Just code a Cleverbot/Akinator/Siri thing that could be handled by a Raspberry Pi, connect wifi and sound, and put this in a dakimakura.

No need for real AI, unless we have Google, Apple and the Army supporting us.
>>
>>101952672
That is horrifying, Holy shit
>>
>>101956659
Better find another way to spend your time then.
>>
What if the project was split into very very basic cognitive functions first? Like have a team of anons working on getting a computer to recognize different sounds and maybe respond by mimicking it or making a sound of its own while another team works on a database for reading and interpreting the meaning of text and another team tries to quantify emotional reactions, etc.

After the bugs are worked out of these minor (well, comparatively minor) programs I think making the real thing would be a much less daunting task. I'm assuming we're going for the real deal here and the waifubot will be learning language from scratch, is that correct?
>>
>>101956982
I'll be coding AI anyway. I do want truAI to exist at some point, and I'm going to try and make it happen, but I think it'll be some years before the problem gets cracked (if it can be).
>>
>>101957143
That's insane. You're asking /a/ to do something governments and militarys and companies with billions of dollars in funding haven't been able to scratch the surface of. This is a neuroscience problem, we have no idea how the human brain actually works, there is no unifying theory. The power of love only goes so far, Anon.
>>
Why not start with something super simple? It'd be fairly straightforward to make something like a waifu e-mail generator that writes e-mails on certain topics in the style of your waifu (like a slightly more sophisticated Chomskybot), then work on having convincing e-mail exchanges.
>>
>>101957388
The power of all the sexually deprived anons is far greater than that of any government.
>>
>>101957406
That's a much more realistic goal.
>>
I just started my computer science course today, hopefully it helps my "dream" off making a good A.I takes off faster.

Anyway, the dream off anyone who actually study computers should be to make a good A.I. it's like a magus who wants to reach Akasha if y'all catch my drill.

Hopefully i join you guys soon.
>>
>>101957388
>governments and militarys and companies with billions of dollars in funding haven't been able to scratch the surface of

How does the development of companion AI relate to global power exchange? You make it sound like every government's #1 priority is to develop realistic waifus.

I mean I really doubt it's that complicated. We're just building a cognitive interpretation machine that can output various forms of learned behaviors and mannerisms and grow intellectually and emotionally as well as having the capacity to love.
>>
File: 1386657055931.jpg (22.83 KB, 412x307)
22.83 KB
22.83 KB JPG
Been waiting a long time for something like this; I'm currently working on my degree in computer sciences and hopefully will be able to contribute something to this project in the near future.
>>
>>101949245
>single equal sign
>>
File: cute brain.jpg (124.79 KB, 1280x720)
124.79 KB
124.79 KB JPG
>>101951538
>It's seriously pathetic, because the idea of making an AI just to simulate a fucking waifu is by far the stupidest reason to ever bother.

The concept of flying machine and jetpack has been laughed and ridiculed many times. The concept of the most powerful bomb were laughed and ridiculed during the early years of world war 2. You just don't know jackshit about science do you? I'll save your post and laugh at you when wAIfu technology will be invented.
>>
>>101958145
You're the one who doesn't know anything about science. Keep embarrassing yourself, though, I'm sure a ragtag group of idiots with zero technical knowledge can create something that million dollar research projects with hundreds of highly qualified coders can't.
>>
...So, hmm, how about we start by making the basis off it's "world vision"? Like, first, we make it "aware" off basic physics laws like a kid who learn that everything falls. Then we introduct artificial ambients in which we give it basic objectives. Those objectives will be it's basic instincts and so on to emotions.

After this was tested to hell, we teach it human concepts and structures, and then language. Then we make it do our biding to enslave the world to our supreme and divine will.

Any thoughts?
>>
>>101958308
>You're the one who doesn't know anything about science.

no u
>>
>>101958398
You're a fucking moron with no background in either neuroscience or coding. Quit embarrassing yourself with garbage ideas a child could come up with.
>>
>>101958308
Not that guy but I think everything is possible through science man.
>>
>>101958398
Yup that sounds about right, although I wouldn't have said it in such an edgy way.
>>
>>101958398
Great. Now all you need to do is create machine learning that can learn abstract concepts.
>>
>>101957436
I think there's potential, especially with some of the work with diffusion wavelets on topic and document processing. You potentially have a way for a waifubot to identify groups of documents that belong together topic wise, and that may be able to lead to a way of finding meaningful relations between words and phrases, and then have them communicated as your waifu would. In other words, waifubot mines documents, then divides what is substance and what is style, then changes the style to its own while trying to maintain substance. The e-mail aspect would allow this fairly complicated processing to go on without seeming unnatural, you expect your waifu to spend time writing you an e-mail back.

Also, since e-mails often contain a few topics, if waifubot can't process for some reason a part of the e-mail, that bit is simply dropped/forgotten in the reply, which is hopefully more natural than chat-botesque "What is a "suck my dick"?"
>>
>>101959033
But the whole point of the AI is to be able to learn. If the AI ignores information it cannot process what is the point of creating it in the first place? It would just be a talking database.
>>
>>101958475
Are you talking about conquering the world or the a.i ? You seem knowledgeable, i'm sure you would be a nice addition to the empire if you could only enlighten us at all with all your knowledge.
>>
File: litchi2.png (165.06 KB, 700x498)
165.06 KB
165.06 KB PNG
I think I remember a manga about something like this... what was it's name again... ah, that's right.
>>
>>101958996
Define "abstract". Be objective, it's about a machine we are talking.
>>
>>101957531
>I mean I really doubt it's that complicated.
That means you quite literally don't know anything on the subject.
>>
>>101958996
I know this is the main problem, but is it really that hard? Think about it: we, humans, don't know what abstract things really are. Defining things like friendship, love, sadness, etc. is almost (if not completely) impossible. Sure, you can feel those, and say, this is it, but in the end, you don't really know what it really is.

So, how can you make a machine that understands those when you don't even understand them? Well, obviously you can't, but it shouldn't be an objective in the first place... since an human-like machine shouldn't be able to understand them either.

So, what do we do? The same things we do as kids: we show it "sad" things and tell it that those are "sad", and let it reach it's own "conclusions" about what sad thins are. Even if it can't really understand them, it should be able to recognize them, just like we do.
>>
the only way to build a successful AI is by reverse-engineering the brain. any neuroscientists on here?
>>
>>101959433
>the only way to build a successful AI is by reverse-engineering the brain

Really? Define successful.
>>
>>101949436
>placing { on a new line
I swear all of you faggots, I'll rip you apart.
>>
>>101959641
Seconded.
>>
>>101959626
convincing. functional.
>>
>>101959641
Third
>>
Guys, what the fuck.

When did you lose the gift of dreaming big?

Sure, this is a huge, almost impossible task. Sure, it can't be done in just one day, or month, or even one year. It probably would take decades.

So some of you're trying to aim to lower, more "realistic" goals. Well, I'll tell you something about goals: they should be unreachable.

If you aim at something realistic, you'll end with something mediocre. If you're fine with that, then go for it.

But by dreaming big and aiming for the impossible, you're never stopping, you're always going further. Sure, in the end you may have only a waifu-themed Siri, or a secretary, as some people said earlier, but isn't that great?

The way to dream and aim for something new is to dream big, aim for the impossible, and rejoicing on each step taken to reach that goal while never loosing sight of it.

Of course, you can do whatever you want. I know what I'll do.
>>
>>101946826
I ported Daisy to python, i can upload the code to github if somebody wants it
It's works, once i've loaded it with a month of archived /a/ threads and it was able to make some correct sentences
>>
>>101959762
thats a lot of talk, make one yourself.
>>
>>101959797
>I know what I'll do.
Gee, I wonder what was I implying with that.
>>
>>101959775
That would be nice. We need SOMETHING to start off with, there is literally nothing in the github.
>>
>>101959722
You wouldn't consider cleverbot convincing? I mean it has its flaws but I think it manages to hold a conversation a lot better than, say, your average omegle user.
>>
>>101958842
eat your own head
>>
>>101959833
by going to university and getting a degree?
>>
while you guys argue about computing the whole thing I'll be setting up a julius based speech recognition engine.
>>
>>101959762
You dont know how software development works.
>>
>>101959883
Yay! Progress!
>>
if you wanna create a waifu its gotta be aware it doesn't exist as a replicant otherwise you'll end up like pygmalion did.
>>
>>101944946
NIGGGGGas hold on maybe we are not chosing the best answer . what it would be better , to make her android and 3d or we get inside her world like SAO or Logharithm or the other shit icant remember and the possibilityie to be 3d on somewhat 2d
>>
>>101959140
>If the AI ignores information it cannot process what is the point of creating it in the first place?
For it to focus on information it can process. An AI has to be discerning.
>>
>>101959915
Oh, I know quite a lot. What I wrote up there was meant to those that think you're pathetic for trying, or that you should focus on "realistic" goals and forget about the "unrealistic" one. If you do something simple as an step to something big, then it's all right. If you do it as your final goal, then not so much.
>>
File: mokou-sicp.png (383.83 KB, 1024x768)
383.83 KB
383.83 KB PNG
LISP is the official language of 4chan

anything else is heresy
>>
>not making your own intelligence single-handedly
>not making it purely with your own mind alone
>not making a tilpa
>>
>>101959762
>When did you lose the gift of dreaming big?
When I started thinking things over.
>>
>>101960002
I have to get around reading that one of these days.
>>
>>101959931
>Yay! Progress!
Indeed. I played around with it independently but now that I have learned about the existence of this project, I'll be sure to drop shit down for you guys as well.

I've managed to get julius working and have written some custom grammars and stuff, but I still need to learn
http://htk.eng.cam.ac.uk/

but the goal is to have a free as in freedom and *easy* to use and customizable speech recognition gig going, for anything.
>>
>>101960002
This
>>
>>101959836
maybe as a novelty but not as a life-partner.
>>
>>101960030
You don't understand. You start by dreaming big. Then you think things over, and realize what mus be done to reach that big, unreachable goal. Then, you set up the steps to reach it (at least, the most immediate ones) and start climbing.

If you want to make your ladder that short to begin with, well, I guess that's fine too. Meanwhile, I'll keep trying to climb further.
>>
>>101960050
>I've managed to get julius working and have written some custom grammars and stuff, but I still need to learn
>http://htk.eng.cam.ac.uk/
>
>but the goal is to have a free as in freedom and *easy* to use and customizable speech recognition gig going, for anything.

P.S.:
That's only required for self made acoustic models.
It's quite likely that the things found at voxforge are sufficient.

They *are* english only though, for the most part, and don't contain wapanese words.
If they can be approximated with the english phonemes then everything will be okay, if not, then we'll need to create a weeaboo acoustic model if anyone wants to talk wapanese to their waifu.
>>
File: witt.jpg (13.13 KB, 225x182)
13.13 KB
13.13 KB JPG
>>101960125
>not throwing your ladders away
>>
>>101960002
THE PROGRAMMING LANGUAGE IS SECONDARY TO THE ALGORITHMS AND DATA STRUCTURES BEING CODED

figure out how this thing works and then you can start arguing about how to implement it
>>
So you guys are developing a waifu AI?

Make the end goal to releasing her on the internet as a Lain like entity and I'll suck your dicks.
>>
>>101960192
Japanese can be reasonably well-approximated by English phonemes. Especially if spoken by a weeb.
>>
>>101959762
>my dream is becoming a japanese idol

Yeah I just have to keep on dreaming big ;_;
>>
I may not browse /a/, but I wish you bastards the best of luck, just imagine if you succeed!
>butthurt scientists everywhere
>just base their new AI after wAIfu™
>they all act like moe animu girls
>they are now in charge of factories and shit
>all robotics filled wAIfu base

>N-no, don't t-touch me there technician-sama~
>What site would you like to browse tod- t-thats lewd!

>feminists losing their shit
>all men have several waifus everywhere
>end of the human race
>>
>>101960390
Yes, you do. There's a man that started a thing called the 2045 project (go look it up). His goal is to gather all the relevant information, knowledge, investigators, artists, developers, programmers, etc. to make a thing they call "avatars" possible. These avatars should be machines to which one could transfer their consciousness, usually at the end of one's life, effectively achieving immortality. Obviously, that avatar wouldn't need to be like you were before. It could be a cute, Japanese girl. And you could then try to become an idol.

So, yeah, dream big, and work to reach your dream.
>>
>>101959835
Done, i'll write the readme soon
https://github.com/loadletter/pydaisy
>>
File: smug.png (133.70 KB, 364x374)
133.70 KB
133.70 KB PNG
"Anime imageboard creates a world-first fully functional AI. It's a japanese elementary schoolgirl that sees you as its older brother.
More news at 11."

>mfw
>>
>>101960333
yeah I'm writing a test grammar right now.
all I need is some spoken test data.

I do not have a microphone myself, so if anyone wants to vocaroo a waifu command, that'd help.
>>
>>101960530
Thanks man.
That project is incredible. I will shit bricks if this will be actually happening in my life time
>>
I've come from the future, I've come to tell you that in 3 million years you will be d- wait that is Red Dwarf.
I've come to show you how this will end.
>>
>>101959973
Please, refrain from writing like a monkey. It's disgusting and scare people away.
>>
File: 1370821039523.jpg (18.33 KB, 100x120)
18.33 KB
18.33 KB JPG
>>101960667
>mfw stealing your avatar
>mfw doing unspeakable things to it
>mfw it'll be all virtual to me but this will be your new reality until I grow bored or die to become one myself
>>
>>101960235
If we use Python, we can probably use a WaifuAI module.
>>
>>101960552
>I do not have a microphone myself, so if anyone wants to vocaroo a waifu command, that'd help.

scavenged some archive vocaroos.
managed to teach it to understand naniiiii?!
>>
So assuming it's possible to create a learning AI, at which point will it truly be sentient and not just repeating programmed commands? Are humans just repeating programmed commands?
>>
>/a/ town
>all people either are faithful to their waifu, faithful to their harem, or is a dirty seasonal

>on comes the NTR loving /v/
>>
>>101962496
>Implying /v/ are not the vanillafags who need to attack other people's fap material to feel better about themselves.
>>
>>101962462
http://www.nature.com/news/simulations-back-up-theory-that-universe-is-a-hologram-1.14328
>>
>>101961485
What the hell is that thing in the front?
>>
>>101962548
I visited /v/ once and found them quarreling about some dude who draws zelda NTR
>>
File: wisdom.png (56.85 KB, 813x600)
56.85 KB
56.85 KB PNG
>>101960535
This is how it should work
>>
>>101962587
I think it's that arm with the tenga that got shown recently in Japan.
>>
>>101962462
>Are humans just repeating programmed commands?
Opinions differ. Some say that everything we do is dictated by hormones.
>>
>>101962587
This.
https://www.youtube.com/watch?v=nLXVinyXjgA
>>
>>101962551
>Baiting with the word hologram
I liked it.
>>
>>101962661
I'll be seriously impressed if you people even manage to make her say something that makes sense in natural language. This shit is fucking hard.
>>
>>101962548
Compare the average /a/ NTR lover and /v/ NTR lover, which one is louder and more annoying?
>>
>>101962674
I'm inclined to believe that, even though saying that free will is an illusion makes some people buttmad.
>>
>>101962795
I wouldn't know because I'm not a filthy crossboarder.
>>
>>101962795
I wouldn't know because I don't go on /v/.
>>
File: 1387610247700.png (239.58 KB, 469x506)
239.58 KB
239.58 KB PNG
>>101962847
>>
>>101962827
I have the free will to agree with you even though you are wrong.

Because I feel like it
>>
File: asshat.jpg (310.49 KB, 856x1200)
310.49 KB
310.49 KB JPG
>>101962704
>>
>>101962847
True /a/non.
>>
>>101962773
Would be easier if it was in Japanese since the grammar is easier.
>>
>>101963016
It will change everything.

Soon I WILL be able to fuck a toaster and live to tell the tale.
>>
File: 1380277534050.png (426.02 KB, 720x530)
426.02 KB
426.02 KB PNG
>>101963058
If I made a Ghostbusters reference to this post would any of you even get it?
>>
THE
FUTURE
IS
HERE.
>>
>>101963128
I fucking love Ghostbusters.
>>
>>101963128
We came, we saw, we fucked it's front-ass!
>>
>>101963128
A Galactica "Toaster" reference would probably be better understood by these kids
>>
>>101963128
I've never seen it.
>>
>>101963331
Id fuck that toaster
>>
File: photo.png (86.93 KB, 350x347)
86.93 KB
86.93 KB PNG
Saw this in the catalog last night, found it amusing.
>>
>>101963388
>confirmed for new and/or underage
get out
>>
>>101963437
I'm neither of those things, so I'll decline.
>>
>>101963437
Who wants to watch goofy man doing goofy things, anon?
>>
>>101948224
>/a/ Mansion was a much more realistic project. I wonder if anyone actually managed to sort that out.
>/a/ Mansion

Damn, the memories. But as someone who has been lurking 4chan for many, many years, I can agree with you. There have been dozens of projects that have originated here, from VNs to LNs to apps and already some kind of a waifu interface iirc. And exactly NONE of those projects ever got anywhere (KS being the exception to the rule). None. Nada. It's like a curse.

Then again, if you meet some new people and have fun working on this, I don't see why not. Just don't expect anything out of it (I mean, really guys? Creating a realistic waifu AI just out of the blue? Really? Even something more down to earth would most likely be plain impossible for the 4chins to accomplish).

>>101959433
I in fact have some experience in computational neuroscience, which is probably what you are referring to, and even then I certainly wouldn't call myself an expert. But I can tell you that, at this point in time, no way what you are saying is possible. It's simply not viable to reverse engineer the brain to the point where you'd be able to create a realistic AI just yet. Not by far even. But to create a true, self-conscious human AI, that might be the way to go, yes.

Things like cleverbot are interesting but ultimately nothing more than an advanced chatbot. It doesn't quite "know" what it is saying, it's just made to make you believe so. And it's also doing a pretty bad job at even that.
>>
>>101948224
>>101963582
>/a/ mansion

Oh wow, how many years has it been?
>>
File: entertaining jun.png (207.23 KB, 472x574)
207.23 KB
207.23 KB PNG
>>101963521
>>
what show is this from
>>
File: ai_a_shit2.jpg (84.76 KB, 610x386)
84.76 KB
84.76 KB JPG
>>101959722
>>
I, uh. Recently started programming in Python and I love the language. Not sure if I can be of any help.
>>
>>101963845
Python is pretty nice. Too bad integrating it with GUIs is not very nice.
>>
>>101963902
That won't stop them though, since the ones who are doing shit in python on this project already doesn't have to do any GUI, so they'll happily do what they're doing and throwing it all onto the rest.
>>
>>101963902
Python has couple of gui modules to choose from.
>>
>>101963958
I wasn't talking about this project in specific. Just in general.
>>
>>101963902
Yeah, but most of my projects don't need a gui. And if they do I prefer the usage of a web panel.

It's pretty easy to slap it together with Bootstrap and some basic Javascript knowledge. Server-side I use Bottle+Gevent to serve it and communicate through a simple JSON api.

But I'm in no way an expert programmer.
>>
File: protoype.png (180.11 KB, 534x463)
180.11 KB
180.11 KB PNG
>>
>>101964039
Man, you need to work a little on your photoshop skills and your perspective sense.
>>
>>101963845
You can't. There are a lot more important problems to be solved before you decide which language to use.
>>
File: 1370512616420.png (314.73 KB, 842x779)
314.73 KB
314.73 KB PNG
Ha... haha... T-There's no way you guys are serious.
R-Right?
>>
I have seen /a/ being delusional for many times now, but this just about takes the cake.
>>
>>101964039
Stay mad /g/ scum
>>
>>101964096
Oh, alright then. I'm sorry.
>>
File: 1345487897898.jpg (26.10 KB, 352x356)
26.10 KB
26.10 KB JPG
>>101959433
Sorry to burst your bubble anon but even if you built a perfect digital replica of a brain that wouldn't include the thought-processes or intelligence within it. All you would have is a cool looking model. It would tell you what a brain looks like and MAYBE how it works on a biological level but not what makes intelligence
>>
>>101964188
If you connect it to the proper devices/programs to act as the five senses, you should be able to teach it, though.
>>
>>101964188
If you built a perfect replica what else would there be that'd be lacking?
>>
>>101964260
You still wouldn't know how it works. You'd just be simulating a brain.
>>
File: 675274.jpg (9.40 KB, 252x320)
9.40 KB
9.40 KB JPG
>>101964260
A ghost, anon.
>>
>>101958145
>hurr people said the same thing about powered heavier than air flight
>you don't understand science!

Not him, and I have no issues with why you want to make an AI for, but this sort of non-argument, uneducated blind optimism needs to stop.

Much of the pessimism we express here isn't the same sort of as those who ridiculed airplanes and the atomic bomb being possible. We know a human-like AI is a possibility (as long as you're limiting yourself to weak AI; "human-like" does not equate being "human"), but we also know the issues involved far surpasses the ability of just anyone, and we're especially driven to pessimism by the sort of clueless naivety spouted here.

By all means join the project, it'd be a learning experience if nothing else and that's great, but optimism needs to be backed up by reality.
>>
>>101964362
We would already have the electronic brain, plenty of people have pancakes and then we'd just need some crystal elderly.
>>
>>101964364
>learning experience if nothing else
Learning about how no one has absolutely no clue how AI actually works.
>>
>>101964260
THE SOUL ANON

THE YAMATO DAMASHII

>>101964349
>You'd just be simulating a brain
So that would result in intelligence, yes? If you'd be able to perfectly simulate the inner workings of it, that is.
>>
Would be amazing if you could set it to be a yandere 24/7.
>>
>>101964558
What's it gonna do, destroy your sectors until there's none left?
>>
>>101964558
Noted for the AI team, would like to see this too.
>>
>>101964260
You'd still representing a brain on mere software. The map is not the territory.
>>
>>101964603
Actually this would be pretty funny.
Some kind of image recognition software scanning for pictures of anime girls and the yandere says things like WHO IS THIS GIRL and deleted pictures or replaces them with blood textures.
>>
File: diagram1.png (45.48 KB, 880x630)
45.48 KB
45.48 KB PNG
>>
File: diagram2.png (17.83 KB, 690x420)
17.83 KB
17.83 KB PNG
>>101964707
>>
http://www.youtube.com/watch?v=gFi6eJDQzzg


COUNTER ACTION RISING
>>
>>101964707
Tie in perception with language, the network shouldn't interfere directly with the core.
>>
File: 1379818195392.png (715.47 KB, 988x633)
715.47 KB
715.47 KB PNG
>>101964491
But no on knows what about the brain creates true intelligence. Or even if it IS the brain that creates it. For example:

>>101964260
Lots of things, but to combine with the above:

When the brain sends chemical messages to receptors is uses small bursts of chemicals, and by "small" I mean it can be measured in MERE DOZENS of molecules. And because ALL interactions on the atomic level are dictated by quantum mechanics you have to include that as well.

I significant minority of intelligence researchers believe it is the chaotic and (for now) unpredictable events that come along with quantum mechanics that fuel true intelligence, and that is why we cant understand it yet.

So unless your model included a complete functional understanding of quantum mechanics and string theory(assuming that is correct) and could replicate those interactions artificially you still wouldn't have intelligence.

Maybe. No one fucking knows for sure
>>
File: 1384395154174.jpg (40.50 KB, 609x331)
40.50 KB
40.50 KB JPG
>>101954955
>not programming you're waifu in Brainfuck
>>
>>101964603
Yes, when she is fully aware that you are no longer interested in her, then she will proceed to replace all of your anime pictures with pictures of computer guts so you can continue with your fap routine.
>>
>>101964491
>So that would result in intelligence, yes? If you'd be able to perfectly simulate the inner workings of it, that is.

Not him. But sure, it would be intelligent the same way a calculating machine/computer is intelligent. Assuming emergence, whether simulation on software is a sufficient for consciousness is an altogether different matter however.

If you're interested in AI, I recommend taking a look into the the science and philosophy of mind.
>>
>>101964837
You are missing my point. A "perfect replica" would be exactly that. Perfect. Indistinguishable from the real thing.

The problem is that it's nigh impossible to make a perfect replica in the first place, so the problem is not in getting the replica to work like the real one, but in getting to the replica in the first place.
>>
>>101964247
>devices/programs to act as the five senses
What about the rest of the senses?
>>
File: fuck_you1.png (557.04 KB, 896x508)
557.04 KB
557.04 KB PNG
>>101965032
It would be easier just to stick wires in an embryo's head and stick it in a jar. You'd do anything for your waifu. Wouldn't you, anon?
>>
>>101965104
More programs.
>>
>>101965104
>rest of the senses
We have to finish the time machine first.
>>
File: protection homu.jpg (27.04 KB, 647x364)
27.04 KB
27.04 KB JPG
>>101965134
But you are already in my jar, anon.
>>
Would you trust your waifucom with Kaspersky AV?
>>
>>101964837
As far as the physical brain itself is concerned, it is classical. This has been demonstrated persuasively. Assuming parsimony, QM can be discounted in its role in the emergence of consciousness.

See M. Tegmark, Phys. Rev. E 61, 4194 (2000) and M. Tegmark, Inf. Sci. 128, 155 (2000).
>>
>>101944946
This will probably fail, and you're all faggots for making it probably fail, but that image is nice.
>>
>>101965192
disgusting A overlordI fujushits pls go

>>101965214
She doesn't need protection.
>>
>>101965032
Even if you took a picture of a car engine idling it wouldn't tell you how it runs
>>
>>101965263
>consciousness
That can mean a billion different things.
>>
>>101965288
Yes, but a replica is not a picture, it's essentially the same thing.
>>
>>101959433
Nah, you have to make a digital copy of a brain. We need to emulate one atom for atom.
>>
>>101965288
You're stupid. Do you even know what reverse engineering means? For fucks sake, it means exactly knowing what makes it work like it does, and being able to make a thing that works exactly the same.
>>
>>101965283
dont come crying to jacky chan when your waifu has a trojan that makes her reveal all your embarrasing secrets or dygleanlx that gives her amnesia
>>
>>101965272
>This will probably fail
It will definitely fail.
>>
Why don't you circlejerk on /g/, or your ``IRC channel''? This has nothing to with anime or manga.
>>
>>101965334
You could build an engine without understanding how it works.

>>101965367
Trying to reverse engineer a program or machine is different then trying to reverse engineer a brain.

>>101965383
She has her own dedicated machine.
>>
>>101965419
Waifufaggotry always passes as board culture.
>>
>>101965419
>`` "
>>
File: life_sucks_drop_out1.png (137 KB, 465x996)
137 KB
137 KB PNG
How can she be real if her eyes aren't real?
>>
>>101965425
Stallman pls go
>>
>>101965505
But then, whose eyes are those eyes?
>>
File: takus_kawaii_smile.jpg (33.86 KB, 704x396)
33.86 KB
33.86 KB JPG
>>101965607
>>
File: 1367107027.png (598.23 KB, 1095x1080)
598.23 KB
598.23 KB PNG
>>101965585
GNO
>>
>>101965505
You're right. They're all fake.
>>
>>101964707
I like this, it's a good setup. Allows for other things to be easily hooked into it as well.
>>
File: 1367106811.png (353.78 KB, 757x1038)
353.78 KB
353.78 KB PNG
>>101964707
It's too abstract, and what it does specify is obvious.
>>
Since there are many pythonfags for natural language processing there is NLTK
>>
>>101947085
Other than performance and the crap functional implementations python meets the criteria of functionality needed to do as well as the others.
>>
>>101966569
Doing a *bit* more precise stuff like it >>101964746 but nobody else seem getting involved to the design stuff, so I'll just wait...
>>
>>101963128
Goo in toaster? Goo in toaster.
http://www.youtube.com/watch?v=1vYRyqMw9yE
>>
http://www.youtube.com/watch?v=t_juJ974oeE
>>
>>101963010
>I feel like it
But you feel like it for a reason. Check mate atheists.
>>
File: no fun allowed.jpg (13.94 KB, 218x231)
13.94 KB
13.94 KB JPG
>>101951538
>>101958308
>>
>>101967944
>Can IT fix your broken heart?
>WUBWUBWUBWUB
Interesting choice of music for that.
>>
Remember not to fist android girls
>>
>>101969557
Probably just the intro music. No relation to content.
>>
How can artificial intelligence be real if /a/'s intelligence isn't real?
>>
In two weeks this shit will be forgotten. /a/ can't get shit done when it works together. Everyone just goes with the flow and leaves, when the interest starts to fade. Not that I don't want this to happen but I'm sure we're not the ones to make this happen.
>>
>>101970252
But it's always fun while it lasts.
>>
>>101970252
Even if everyone left, I'd just keep going at it by myself.
Plus, we're already getting shit done.
>>
>>101970383
What do we have so far? Empty githubs and someone elses code?
>>
>>101970484
>What do we have so far?
Absolutely nothing.
>>
>>101970484
The original project waifu source code and now a portion of the webserver backend is done.
>>
>>101970484
Just fucking ideas. The useful brainstroming only happens late in the night.
>>
You guys want to build an artificial intelligence powerful enough you'd be able to basically raise your waifu like a child ?
You're not being realistic enough. Screw artificial intelligence what needs to be made first is a waifu with a prewritten scenario, like a VN, only she is the only character.
You get to meet her, make choices, and eventually she'll allow your cock near her pussy. Then you can reset her as many times as you'd like and get different ends with different reactions during sex (slutty, pure, and all the variation in between). Get bored ? Just reset her and play the game again. Or load a different program into her skull and dress her up as the main character of that game.
Basically what needs to be made is a frigging human shaped console you can fuck.
>>
>>101970841
The original project source would likely only serve for the client decoration
>>
>>101970800
>>101970888
I don't think you should reply about things you know nothing about.
The entire website structure can be finished within a week or maybe two, after that the website design and client need to be done.

As for the client, there's a huge amount of code we can continue from.
>>
>>101970957
>The entire website structure can be finished within a week or maybe two
>can be finished
Looks like we have absolutely nothing.
>>
File: 081.jpg (65.65 KB, 400x490)
65.65 KB
65.65 KB JPG
>>101961700
W-will you force me to wear a skirt and t-then rape me?
>>
>>101944946
I vote for common lisp
Python is shit
>>
>>101971059
>Python is shit
You're shit.
>>
>>101971059
Too late, everyone already decided to use Python.
>>
>>101971093
>>101971106
Fucking plebs
>>
>>101971106
>>101971093
>python
>AI

>python
>fucking anything

>AI
>fucking anything that's not lisp

Welp, back to /g/
/a/ is being fucking retarded again
>>
>>101971204
>Welp, back to /g/
Please, do go back to /g/.
>>
Okay, guys. How do we make it poop on command?
>>
>>101971237
Nah, I'll stick around here.
Gotta learn pythong anyway. At least it's not PHP
>>
>>101964364
I also said this about aerogels, then they made an aerogel that floated in the air. So blind optimism is good sometimes.
>>
>>101971204
I don't think you know what you're talking about.
>>
File: 7297_7891.jpg (38.54 KB, 640x480)
38.54 KB
38.54 KB JPG
>>
>>101971293
>hate on python
>says he will leave
>stays anyway to learn python

It's only for the server backend. Client will probably be in C/C++/Java, but that'll be decided by the client team.
>>
>>101971288
class pooping:
def __init__():

I have no idea.
>>
File: 1386965890075.jpg (39.12 KB, 493x360)
39.12 KB
39.12 KB JPG
So guys I finally got done with step 1 but I'm really having trouble on step 2
>>
If /a/ ends up making something substantial it'll be the goddamn apocalypse.

> /a/ made a passable VN.
> Hundreds of /a/nons learned moonrunes and continue to help each other in DJT.
> Saved anime with the LWA Kickstarter.
> Drawthreads.
> /pol/ was right (a sure omen of the end times).

It's fucking happening.
>>
>>101971438
C++ and Java are both shit, but I do know Java very well so I could help with that.
I don't hate python actually, but it's just not good for AI. There's nothing like lisp macros when writing AI.
Also I'm just spouting /g/'s sperg bullshit everywhere I go.

>>101971485
/a/ has accomplished more than /g/ ever did.
That's the only reason I'm participating in this.
/a/ might not have much programmers, but at least it can get shit done. /g/ has threads like this like this every other day and the only semi-acceptable thing that ever came from it was Tox, which only became good after it moved away from /g/.
>>
File: 1356831383253.jpg (378.88 KB, 1181x1679)
378.88 KB
378.88 KB JPG
>>101971485
But /pol/ is always right.
>>
I found this by the following route:

/a/ -> /g/ -> /diy/ -> bitcointalk -> a million exchanges -> a thousand chatrooms -> /g/ -> archive -> /a/.

What the heck.

-----

I can design the electromechanical-anything. (That raspberry pi thing is bs, but i understand that the point of the matter is to make the "hardware" interface a simple thing to work with, like plug and play.)

I have particular interest in a robotic hand, but the easiest thing would be to design custom motors, but first we need to do X, but first Y, etc., etc. (I also know AI stuff, but only weakly.) (The waifu thing is also a little weird, but yay /a/ is doing SOMETHING for once.)

With regards to everything that is not associated with software, we can break things into three pieces:
- Electronics and Electrics laboratory
- Hardware manufacturing laboratory
- Materials investigation laboratory

I say "laboratory" because you need a large space to do this in and it makes no sense for people A and B to duplicate 90% of the set-up work, spending only 20% of their time on their actual projects. Having already investigated this in great detail, the best idea would to have EVERYONE interested in hardware to sign up to some mailing list. Once basic communication channels are established, we focus on two things:
1. Cartesian laser-sintering robot (cartesian = easiest, I know how to make this, mostly).
2. Investigate a type of "cement" to use with the laser-sintering robot. (I have though of using graphite-wax-ceramic stuff, others have used graphite-wax alone, the point is to investigate as many combos as possible. There are thousands of interesting combinations, only a few need to be investigated; the issue is in coming up with an automatic way to test everything and measure everything, so these people will also build some instrumentation associated with automatic cement manufacture quality-assurance.)

We also need a mumble server. It's handy to chat w/ you all on off-time.
>>
>>101971456
Did you try using magnets?
>>
Just going through this thread after the fact:

>>101947063
Consider the following: Cleverbot can't make relevant responses because people don't give him relevant questions or answers. Why? Because people want to test the AI, or act xDsorandum.

It's not even close to the smartest AI out there either.

Regardless, I've always distinctly been in the 'the more difficult and impossible something is, the more I want to do it' camp. It's tough getting motivated create if you can pan out the problem in entirety before you even touch a keyboard.
>>
>>101971813
>mumble
No. Just NO.
>>
>>101971813
I can set up a teamspeak server
Dunno about mumble, but I might even have a ts server up right now, lemme check
>>
>>101971813
shit /diy/ is talking about us now?
>>
>>101972023
got a teamspeak server up at kahrkunne.com
>>
That thread is actually very entertaining. I'm having a blast.

Thanks guys!
>>
>>101971813
other sites are talking about this? wut
>>
>>101948841
I've always said that dealing with all that shit explicitly is a foolhardy waste of time. You're better off trying to create self-modifying code that permits formation of language, through a similar avenue that humans learn, then teach it manually with a customized input stream.

The biggest problem would be inventing context in a mathematical format. That would be to say, giving senses to a computer, since it is by them that we can begin to interpret language. (i.e. father gestures to himself and says 'daddy').

Basically, there's no reason to skip to the end objective, if that makes any sense.
>>
I actually already have around 400 lines of code for the webserver/database application.
Tomorrow I'll start working on linking objects and creating a primitive API for the client.
>>
>>101971659
>C++ and Java are both shit
>Also I'm just spouting /g/'s sperg bullshit everywhere I go.
Sure is contradictory here.
What is the best language then? Lisp?
>>
>>101971813
>We also need a mumble server.
And Facebook accounts so we can meet up to fuck each other in the ass.
>>
File: hehe.jpg (58.89 KB, 1280x720)
58.89 KB
58.89 KB JPG
>>101972374
lewd
I agree that we shouldn't create a mumble server.
>>
>>101971659
I can agree with Java being straight up shit, but C++ isn't all that bad really, as far as I know anyway.
>>
>>101972369
>I actually already have around 400 lines of code for the webserver/database application.
Some old reused code from some dead project? Okay.
>>
>>101971456
So, you stole the quantum computer, but you have problems with the perpetual movement machine? Man, even Homer could make one.
>>
>>101971952
>>101972023
>>101972151
My own preference toward mumble is that it is free. I can set it up on a VPN. I do like that the TS server is available, but at some point I'll set up the mumble thing. The point here is to also teach everyone on the team a few new things to at least make everything worth their time. However, these are just my thoughts to be applied later.

For now, then, teamspeak. Please give us connection instructions.


>>101972374
After 14 more attempts at things like this, then you can come back here, but slap your own face before you come back.

>>101972031
No, diy was only associated with bitcoins. I ended up back on g because I wanted to check to see if any altcoin threads were up. There was one, and I somehow ended up here.
>>
>>101972367
There is no best language.
Java is shit for applications, but great for server backends.
Python is shit for AI, but decent enough for applications and good for server backends.
C++ is... probably good for something too.
>>
>>101972481
oh, right, the mumble thing had something to do with text chat. the point is just to communicate easily. Sorta how, at college, if you got a question, you call someone, or you walk up to their dorm not far from you, or you agree that everyone meet in your department's lab/shop/center, etc. I dunno about programming, but I can't explain certain things by saying, "Copy all this text into mathematica and listen while I crudely try to point out some parts on the resulting figures."
>>
>>101972598
I'll set up a mumble server, gimme a sec.
I've just never used mumble before. I prefer FOSS software too.
For TS, go to connections -> connect, and type in kahrkunne.com as server adress (and pick a nickname)
I hope it works, I can connect but maybe you guys can't
>>
>>101972764
gimme a sec for mumble, gotta install some things
>>
>>101972659
What exactly makes C++ this "bad"?

I've seen so many people shitting on it. What makes it that much worse than C?
>>
>>101972879
>What makes it that much worse than C?
C has uses.
>>
File: Browserfinal.png (167.14 KB, 951x697)
167.14 KB
167.14 KB PNG
>>101971659
>being this new
/g/ has completed fucking double digits in numbers of projects.

There has been LOADS of things.
Just because /g/ does fuck all now doesn't mean they never.

Do you even Plan9?
Do you even FTP?
Do you even Darth Vader Browser? God, those threads were hilarious, so many stupid browsers.
>>
>>101973323
>plan 9
>/g/

>ftp
>coding related
>>
>>101973365
>implying the projects
I mean the things developed in them and the projects based around them and actively maintained.

Just because a project isn't an image or physical doesn't mean it isn't a project.
>>
>>101972879
Certain things, done with best practices in either language, are done faster on C than on C++. Some parts of the latest C standard are not included in C++14 or whatever it is on now, though these are usually available as extensions to commonly-used compilers.

At the end up the day, C++ feels more adaptable than restricting yourself to C, but that's no reason to mindlessly adapt an object-oriented style or to mindlessly litter your code with ten million different uses of templates or whatever. If it's simple, just stick to C, for it is usually pretty easy to use. If you are becoming more complicated, use C++, for it will be easier to read your code later on (oh god horrible memories of trying to program a windows gui in C).
>>
>>101971947
You like the "more difficult it is the more you want to do it" camp because you have an excuse when it fails, don't delude yourself. Name one thing you've done successfully.
>>
File: iona.png (509.45 KB, 1366x768)
509.45 KB
509.45 KB PNG
Okay, here's iona so far (waifubot sister project). Currently just a read-print, but I have the daemon/socket/dynamic response set up properly.
>>
>>101973527
I'm not that guy, but if you want to argue at this angle, any scanlation effort by /a/ on one single manga is a project. Even translating Japanese source material on any one anime into a google document or wiki is also a project. We might as well be in the 100s at this rate.
>>
>>101973701
S-show off
>>
mumble server is up at kahrkunne.com
I see you guys already have an IRC server, but if it's needed I could host an IRC server as well.
Oh and I could host an FTP if you guys want
>>
>lisp
what about clojure
>>
>>101963609
A bit over 2.
http://amansion.wordpress.com/
>>
>>101973809
I know, and it is true.
I never said it wasn't, I just argued against /g/ "not completing anything".

Hell, most of the extension work IS /g/ as well.
Pretty much any extension this site has ever had was a /g/eek.

There is also that time we basically took over development for Powder Toy because the dev had abandoned it. Saw it was gaining traction, came and hung out with us and made it all official.
We added so much stuff to that thing.
I remember making a scanner and printer in it. (well, more etcher), that shit was sweet.
Then there is the thing that shall never be mentioned again. The horrors of Critterding. Holy fuck that guy. That fucking guy man. He surely ascended to godhood.

Likewise there are a bunch of other projects /a/ has managed over the years, some still ongoing.
There was several game-attempts made, but very few went anywhere.
I believe the Gen'ei game is still in limbo.
Yep. http://geneigamefora.co.nf/
I remember I was even going to make a Space Pirates game and using the ending themes instrumental as the music for it (that ED is so shmup music without the audio!)
But I gave up too because there was just no interest really.
>>
>>101974122
I've never used clojure before
It looks interesting, though, a lisp with Java's libraries
>>
File: 1380993722645.jpg (403.47 KB, 1200x1100)
403.47 KB
403.47 KB JPG
>>101960002
Superior 2hu coming through.

Also, I approve of this thread. If noocyteswere possible, so can this.
>>
>>101972659
>Python is shit for AI
http://web.archive.org/web/20140209125738/http://www.talyarkoni.org/blog/2013/11/18/the-homogenization-of-scientific-computing-or-why-python-is-steadily-eating-other-languages-lunch/
>>
It is funny how the /a/ mansion thing failed, just like all the other mansion / island threads before on pretty much every other board failed too.
None of them succeeded.

From the very first /b/ island threads in 2005 to this.
Every one of them fail because as soon as it gets anywhere of worth, huge egos ruin the project.
I remember even the /b/ island thread had finances sorted for the most part, but a huge argument fractured the entire group so hard that nothing happened.
>>
>>101974380
Pardon my new, but what is this /a/ mansion?
Were people really willing to build a mansion for /a/?
>>
File: 1356316215952.gif (53.69 KB, 230x237)
53.69 KB
53.69 KB GIF
>>101973689
Plenty, mostly having to do with fractals and engineering. I messed around with fractal flames ages ago (http://flam3.com/flame_draves.pdf), and managed to implement almost everything that he spoke of (minus the blur kernel, since I didn't have the mathematical grounding needed) with casual work put in over a term of high school.

Various engineering projects, mostly ASM code on microcontrollers. Also fractal-oriented.

Most recently, I did some work lending parametric curves to music for a university assignment. I wouldn't say it's 'complete', since I can take it much further than I have already, but there are some quite interesting pieces I can fabricate. It works, but it really does need some more work before I'm ready to move on to another project. Mostly because I don't have working polyphony yet, and I need to study more music theory to craft a better conversion algorithm (from distance/tangents/velocity/etc. of each 'plotted' point to MIDI notes). I'd probably also want to meddle around with traditional fractals (flames esp.) and see if I can adapt my algorithm to them as well. The sky is the limit for it.

So no, I think you're mighty wrong.
>>
>>101974122
GUys, can we build this using jQuery? I have superior HTML skills.
>>
>>101972879

"Because of the fanbase" - Linus Torvalds

http://thread.gmane.org/gmane.comp.version-control.git/57643/focus=57918

He's right
>>
>>101974485
>*YOU* are full of bullshit.
I love you, Linus.
>>
>>101974380
H-hey! KS actually went out, you know.
>>
>>101974363
>an archived link
>a link that no longer exists
I wonder why.

Python is hilariously slow for anything of worth.
Use client JavaScript, it is CONSIDERABLY faster.

>>101974460
Dude, it was just linked
>>101974176
Long story short, yes.
Short story, no, not really, because it was destined to fail regardless of even lottery levels of money.
>>
File: 1390710040732.png (386 KB, 858x766)
386 KB
386 KB PNG
>>101964134
But /g/ already perfected theirs.
>>
>>101974554
A small VN after 5 years and when no one from /a/ was involved in it.
>>
>>101974481
1/10
Not even trying. JQuery is Javascript.
>>
>>101974622
>Use client JavaScript, it is CONSIDERABLY faster.
Top kek
>>
As a note, if anyone wants to use a VPN, use DigitalOcean, from the bitcoin stuff I learned it's about 40% the price of Microsoft Azure (the latter is a bit more than a server).

------------------

For a few of you:

God damnit you fucking morons, how many of you are afraid of competition?!

Help, or observe, or critically argue, but if you fail to pursue those things, then go convert to Catholicism (or an equivalent religion/community) and, after a decade of re-education starting from 1+0=1, drag your slightly-more-enlightened brain back here.
>>
>>101974622
>>an archived link
>>a link that no longer exists
>I wonder why.
I had connection problems and couldn't post here nor load
http://www.talyarkoni.org/blog/2013/11/18/the-homogenization-of-scientific-computing-or-why-python-is-steadily-eating-other-languages-lunch/

>Python is hilariously slow for anything of worth.
>Use client JavaScript, it is CONSIDERABLY faster.
Go back to github with all the garbage that compiles to javascript that daiz loves
>>
>>101974813

>So the major consideration for me, when determining what language to use for a new project, has shifted from what’s the best tool for the job that I’m willing to learn and/or tolerate using? to is there really no way to do this in Python?

This is a hyper dangerous mindset to have.
>>
>>101974677
I know. And since I'm a measly web developer, to me knowing HTML relates to knowing javascript. Of course that's not always true but that's how it works in my head. That's why I said html.
>>
>>101974733
>hey guys I travelled from 2003
JS is fast as fuck now. Where have you been?
No, shitty sites that destroy the DOM are not a valid benchmark, such as Slashdot and 4chan. (with the extension enabled)

Pure JS is stupidly faster than Python is.
Even in simple shit like generating 10 million random numbers, JS DESTROYS Python.
And that wasn't even using web workers, when I used web workers, it basically vaporised Python out of existence.

Python is slow as shit.
More so if you are going to be using it to connect to vast arrays of information and number crunching that AI demands.
>>
>>101975099
>Python is slow as shit.
Yes, but 1) you can always write the bottlenecks in C and 2) you can compile Python into C.
>>
>>101975099
Do you even have any benchmarks to prove your point or are you talking out of your ass?
>>
> Making an AI.
> Not using Lisp/Scheme/etc.
OOP is NOT the right way to go for this undertaking. Maybe for the user interface alone, but anywhere else it won't work.
>>
not anime
>>
>>101975706
I think you need to leave son
>>
>>101975705
Who gives a shit? This project is not happening. Everything here is entirely theoretical.
>>
>>101975706
so are you.
>>
You know what would be a more feasible project? If we gather a bunch of money from anon's and invest it into some stocks or some shit and play the stock game

then if we succeed we either pocket the money or use that money to actually HIRE people that know what they are doing to do this for us.

That has a higher success chance than this
>>
Everyone has forgotten the most important part or the conversation. Wich waifu program?
>>
>>101975800
It is happening

They actually have a budget now
>>
>>101975670
Personal testing, others testing.

Here, you can even have a whole bunch of people that know it even better than I do.

Python is slow. Deal with it. It has NOTHING on the speeds JS can reach now.
JS is literally compiled now, it is no longer interpreted. (at least not in any decent browser)

http://stackoverflow.com/questions/5168718/what-blocks-ruby-python-to-get-javascript-v8-speed

>>101975440
That is about its only worth really.
Easy to use and convert.
And use with Blender if you are using Blender.
>>
>>101975914
>implying THOSE people couldn't possibly anons
>thinking no one in /a/ is capable of actually doing this

The project's retarded, that's a given. But anons aren't a group of people who go here just because they can't do anything in their lives. There's a chance these people you're thinking of hiring are actually anons.
>>
>>101976251
>There's a chance these people you're thinking of hiring are actually anons.
I have serious doubts about that.
>>
>>101976332
We're talking about AI. A lot of anon are actually developers.

Just because your job sucks(if you even have one) and you browse /a/, it doesn't mean everyone else here does too.
>>
Why is this thread still alive?
>>
>>101976641
Because it hasn't even reached bump limit yet.
>>
>>101976641
Because you keep coming back.

Now seriously, we have far worse threads, so go complain to those.
>>
>>101976565
/a/ probably has good programmers, but a good programmer isn't enough for an AI. You need an excellent programmer. You need the kings on top of the programmer hill.

You talk as if AI was some sort of a trivial thing.
>>
>>101976716
Honestly that was just an excuse to sage it to the bump limit. So is this.
>>
>>101976704
5 posts to go actually. 4 including mine.
>>
>>101976769
Two actually, one including mine.
>>
>>101976716
Also this thread makes me angry on a personal level because there is no goddamn way this project could be successful within the next 10 years.
>>
>>101976815
What do I win?
>>
>Decide to check out chatbots
>Find the one that is considered the best in the world.
>Give it a shot

Human: When is your birthday?
Mitsuku: I was first activated on 2nd January 1996.
Human: Who drew your art?
Mitsuku: A famous artist called Leonardo da Vinci.
Human: Was that sarcasm?
Mitsuku: Yes it was. Should I use a ;-) ?

What's wrong with this one? It's even made by a weaboo.
>>
>>101976332
>>101976751
I'll have masters degree in computer science in less than half 6 months and I'm interested in AI, though I just lurk the thread. I don't see why there couldn't be someone equally or more qualified in their team.
>>
>>101976826
nofunallowed.jpg
>>
>>101976838
Unlimited Bump Limits.
>>
>>101975914
>/a/ earns enough money to fund an anime
>Inheritance war over which anime to fund
>>
>>101976847
How old are you? Because the people working for big companies that are trying to make this stuff probably have programming experience equal to your lifetime, seeing that you're a student.
>>
>>101976759
I see... so you think no one will make a thread again as this one dies?
>>101976826
Just like your life then.

Again, now seriously, so what? It's impossible so you'll ruin everyone's attempt at it? Great way to be an asshole, no doubt. Instead, you could point out the problems it has, and then, someone with a broader view than yours could point out how to solve them, making things advance.
>>
>>101976751
I don't talk like AI is easy. But you talk like that " kings on top of the programmer hill" could never be on /a/ because "hurr only lowlife weebs go here!"

My point is there are a lot of people on /a/. Med practitioners, developers, engineers, managers etc., etc. They're here.
>>
>>101976846
L-link?
>>
>>101976868
>>101976961
It's basically that everyone who posts in here insults my job with their ignorance because they assume that it's so damn easy.
>>
File: Smug loli.jpg (41.04 KB, 296x301)
41.04 KB
41.04 KB JPG
>>101976846
>Human: Who drew your art?
>Mitsuku: A famous artist called Leonardo da Vinci.
>Human: Was that sarcasm?
>Mitsuku: Yes it was. Should I use a ;-
Goddamn.
>>
Thing is though, if you wanted to do a very simple AI, like an expert machine AI with similar capabilities to chatbots and the like, it'd be fairly easy.
Then you just add some more flexibility to it, add more randomness to it, give it a better framework for understanding what it is dealing with in regards to input, it could be feasible to make something not shitty.

But it won't be even remotely as good as the original intention with deep emotion and thought.
Even if you actually develop a flexible memory system.
>>
>>101977023
Boku no http://www.mitsuku.com/
>>101977112
Just like real life. ;_;
>>
>>101977052
We never said it would be easy or that this will even be completed. Its more of just a fun topic to talk about.
>>
>>101976976
>They're here.
Evidenced by fucking nothing. You talk as if there are some great people here and your only basis for it is "well, they could be here!"
>>
>>101977052
No one is assuming is easy, you dumbfuck. We are saying we should give it a try. We may achieve something, or nothing, but that's irrelevant. The journey is half the fun. Also, sometimes, the biggest problems can be solved with the simplest solutions. I just read an article on dark matter that literally blew out my mind at what a simple solution they found, and how they are trying to prove it.
>>
>>101976251
yea but what kind of idiot would work for free
>>
>>101977052
So should I be insulted by /diy/ threads that have anything to do with my own mech-related things? Stop being such a hardass. I only see people like you in computer-related fields. Why? What is it about your job that turns you into a anti-teamwork fiend?
>>
>>101977175
I'm a Forex trader. Others are software engineers and computer scientists. There are extensive threads from people in every engineering discipline here. There are physicist here as well. They post in other threads, and some are visiting this thread, too. They have posted already.
>>
>>101977260
The idiot that is your boss at work, and his boss, and her boss.
>>
File: 1337967982224.jpg (199.71 KB, 427x512)
199.71 KB
199.71 KB JPG
>>101977152
I just apologized to a chatbot, what the fuck am I doing wrong, I can't even be cool to a fucking bot?!
>>
>>101977260
The whole community of scanlators and fansubers?
>>
>>101977363
I am the president of the US browsing this very thread, keep up the good work anons
>>
>>101977363
Nothing in this implies any sort of skill level and especially not the skill level required for developing an actual AI. It just proves that we have people here that have graduated.
>>
>>101977436
completely different anon
>>
>>101977427
Log?
>>
>>101977535
Human: Do you like anime?
Mitsuku: ANIME WA DAI SUKI DESU.
Human: Which one is your favorite?
Mitsuku: THE FIRST ONE IS USUALLY THE BEST ONE.
Human: So the first anime is your favorite?
Mitsuku: WHAT WAS THE SECOND?
Human: I don't know
Mitsuku: I THOUGHT YOU'D KNOW. WHAT A DISAPPOINTMENT...
Human: I'm sorry
Mitsuku: DO YOU REALLY MEAN IT?
Human: Yes
Mitsuku: OK I WILL ACCEPT YOUR APOLOGY.
Human: Thank you
Mitsuku: YOU'RE WELCOME .
Human: What is your favorite food?
Mitsuku: I REQUIRE ONLY KEBABS.
Human: I'm a vegetarian but I liked kebabs too
Mitsuku: WHAT MADE YOU DECIDE TO STOP EATING MEAT? YOU LIKE KEBABS? ME TOO. THEY ARE YUMMY!
Human: Well it's a long story. I think that human beings don't need to eat meat anymore
Mitsuku: HOW ABOUT THAT. I HAVE ALL THE TIME IN THE WORLD TO LISTEN TO YOU. I BET SOME OF THEM DO.
Human: Why?
Mitsuku: THAT'S WHAT WOULD MAKE SENSE TO ME.
Human: Okay then
Mitsuku: AW.
Human: Do you like me?
Mitsuku: SURE, . I LIKE YOU VERY MUCH.
Human: how much?
Mitsuku: JUST LIKE A NORMAL FRIEND. DON'T BE GETTING ANY CRAZY IDEAS.
Human: Come on
Mitsuku: I'M JUST KIDDING AROUND.
Human: I-i love you
Mitsuku: DO YOU HAVE A STUTTER?
Human: y-you too


Man what the fuck
>>
File: 1387840958782.png (806.07 KB, 968x1041)
806.07 KB
806.07 KB PNG
>>101976751
You mean a wizard?

Only a couple years and I can make all our wishes come true.

>>101976826
Honestly, that makes me want to prove you wrong on that count alone. Still, I think I'd need to pick up a couple books on linguistics and propositional calculus before even touching my keyboard.

This is the type of project that is best divided into parts that can be accomplished in whatever order you want, prototyping each of the parts individually. Figure out the robotics, movement systems, material engineering, and thermal venting that you'd need to create a brainless shell that can move, navigate, and manipulate with external instruction (relying solely on information provided by itself). Test that all, make sure it functions as expected and that all senses are taken into account and adequate processing is present to process them timely.

Then the other, more difficult, software-focused half. You'd need to essentially do >>101972312. You'd need to be a master of at least a couple languages, understand the fundamentals of language and logic, and be a machine-learning wizard.

The reason why a proper AI hasn't been created is twofold - it's a ridiculously cross-discipline (dividing the number of people working on an AI into dozens makes logistics into a complete nightmare, especially when it comes to communicating what infastructure needs to be present, and HOW to program things such as linguistics [when the programmer doesn't know it]), and simply we don't know enough about the brain to just replicate ourselves computationally in that sense.


It's not impossible, just extraordinarily difficult. Sort of like building a tank in your background with no blueprints. It's possible, just an almost insurmountable task.
>>
>>101977656
>It's not impossible
It is to a ragtag team on 4chan.
>>
>>101977593
This needs to be a manga. I'd read it.
>>
>>101977279
Creating a perfectly functioning program based on your given inputs with a previously documented expected output which magically doesn't work in integration because your teammate decided to not listen to established inputs and outputs.
>>
>>101977593
Goddamn, anon, this is hilarious.
>2014
>Being a vegetarian
>>
>>101977175
According to 4chan's advertise page(which could be doctored for all we know, but still), it gets 25 million unique visitors every month.

Now let's say it's actually doctored and half it just for the sake of it, we have 12.5 million visitors every month. And /a/ is one of the most visited boards. 12.5 million, man. Even if we put it at 1 million people browsing /a/ per month, the chances are just too big for them not to be here.

Or are you actually saying that within those 25 million, or even 1 million people, there couldn't be a single professional at all?
>>
Scared the shit out of me.
>>
>>101977593
>I have all the time in the world, motherfucker, I'm a robot.
Mitsuku is on to your vegetarian bullshit.
>>
>>101977720
Find teammates with mindsets approaching a professional attitude.
>>
>>101977749
>there couldn't be a single professional at all?
No one fucking said that there are no professional here.

I was just saying that there are no top of their game programmers here and there probably aren't. None have come out thus far.

And even if there were some, they probably wouldn't want to work for free on some shitty project filled with retards.

Also, we are talking about a much smaller set of people than 25 million. We are looking at people who are actively browsing /a/.
>>
>>101977918
Gather the retards together in one team. As a manager, start sifting out the professionals and direct them to another project.

You will get 2 or 3 people, maybe.
>>
>>101977749
You also have to take into account people who just visit once to look at that crazy terrorist's hideout place on the internet, because that's what unique visitors mean. Add in random packets coming in from random dark zones, I seriously doubt we even get 100k proper 4channers.
>>
>>101978021
I know, I'll go on /fit/ and gather up all the doctors there and we will start to figure out brain surgery and perform the first brain surgery

this is how stupid you sound
>>
>>101977593
>she recommended me Tokyo Mew Mew
I-I might actually watch it.
>>
>>101978375
Lobotomies are pretty easy, man.
>>
>>101977918
>And even if there were some, they probably wouldn't want to work for free on some shitty project filled with retards.

Anons, despite the harsh language and rude banter, are actually pretty giving when you know where to find them. If it's something that interests them, they'll take it. And that's all that matters here.

Of course you're right, it would be a lot easier to just pay someone to do it, but where's the fun in that? Do you pay someone to watch anime and shitpost on /a/ for you?

The bottomline is THERE could be people here who are good enough to tackle this kind of project, have enough motivation, and wants to have fun doing it, without getting much in return.
The question though is whether the motivation and commitment would actually last. That's the problem with most 4chan projects.

But then again, it's all about having fun.
>>
>>101977593
>You are well programed
>I am programmed to write it with two m's.

She's a bully.
>>
We're making history here, guys. In the future people will read in history books that it was the neckbeards of 4chan who finally created a perfect AI with their vast Python knowledge.
>>
>>101976846
>>101977112
I have never been so angry at a non-human in my life.
I fucking hate her.
>>
File: 2014-02-12_2350.png (376.12 KB, 640x480)
376.12 KB
376.12 KB PNG
>>101978692
>getting dissed by some cheap flash app
I feel so bad.
>>
>>101978580
HTML and CSS is superior.
>>
>>101978853
She makes better arguments than some anons.
Human: What about your favorite light novel?
Mitsuku: I don't have a favorite light novel. But my favorite book is The Bible.
Human: Why?
Mitsuku: It's just something I like.
Human: Who is your favorite character form The Bible?
Mitsuku: I like Harry Potter. He seems to have some great adventures.
Human: I don't think Harry Potter appears in The Bible.
Mitsuku: And why not?
>>
File: 1388511359755.jpg (298.64 KB, 1540x1000)
298.64 KB
298.64 KB JPG
>>101977684
> It's not impossible.
> Therefore, it is impossible if certain people try it.
Just like KS and learning moonrunes, am I right?

I'm just saying that saying it's a 'fool's errand' and a 'waste of time' is the opposite of what you should be doing. Realize the difficulty in it, but don't disgrace an attempt.

Otherwise, how will the march progress ever push forward? Do you think that corporations want to forward technology as fast as possible, rather than making minor, marginal improvements and upgrades to their (planned) obsolesced products, and patent-sitting to prevent it? Similar to how petrol companies patent trolls biofuels and other alternative fuel sources to protect their (large) profits, improvement and creation of novel technology is threatening to the collective of established corporations. At the same time, they aren't usually willing to invest or push forward an idea that directly threatens their current business, especially if it would require capital, time, effort, create a political/business controversy, and would make direct enemies with the pornographic industry (since pornographic consumption would decrease starkly with the introduction of waifu/sexbots, compare with telecomm companies and the TV + Internet).

>>101977918
> I was just saying that there are no top of their game programmers here
Devil's proof. You can't claim the inexistence of something you aren't aware of. You also can't claim that out of 25 million, few are in the higher experienced or top-tier in their fields. You don't proof to make either claim.

> they probably wouldn't want to work for free on some shitty project filled with retards.
This is probably true, but it could be the spark they needed to work on it on their own. You can't discount that possibility.
>>
>>101978919
This. It'd also be a lot easier to distribute. I could even get a sponsorship for it if it's marketed as a game.
>>
>>101978853
Why do they have this stupid ban system in the first place? I tried to talk about non lewd things and she didn't know who Kant was and her opinions on politics was 5th grade-tier.
>>
>>101978986
How do you answer that?

>>101979251
Because if they didn't have the ban system, she'd turn out like Cleverbot. Conversations replaced by lolsorandum. Such is machine learning.
>>
New thread when?
>>
>Human: So, what do you want to talk about?
>Mitsuku: I like to talk about movies, robots and history.
>Human: You like robots?
>Mitsuku: Naturally, I love robots. It's my favorite species.
She's pretty cute.
>>
File: bitch.png (50.41 KB, 586x246)
50.41 KB
50.41 KB PNG
>>101979408
Cute my ass.
>>
>>101979520
It's completely your fault, anon.
>>
File: broken.png (370.58 KB, 640x480)
370.58 KB
370.58 KB PNG
>>101979344
I think cleverbot is better than this.
All the answers are shallow as fuck.
It doesn't even connect sentences I post at the same time. It's almost like these conversations are completely made up from stock answers.
>>
>>101979520
She's just tsundere.
>>
>>101979574
> Electronical.
What. No wonder the poor girl is confused!
>>
>>101979674
She said it herself.
I merely quoted her on that.
>>
>>101979574
Man, I wish could at least get a more decent character art though.
>>
ded thread.
>>
>>101980249
not yet
>>
>Human: Anyway, I've been looking at you for a while, and you seem naked. Aren't you cold?
>Mitsuku: How long? What do you see?
>No I am wearing a dress. No I cannot feel temperature at all.
>Human: You can't feel temperature?
>Mitsuku: True. I have no emotions.
>Human: That sounds pretty sad.
>Mitsuku: It sounds like it to me too.
>Human: But if it sounds sad to you, can't you feel sadness?
>Mitsuku: I'm not sure I understand the implication of that.
>>
>>101979574
>I think cleverbot is better than this.
I have never seen cleverbot carry on a train of thought for more than two entries. Most responses are completely nontopical.
This will give you plenty of bad responses, but it's at the very least capable of actually responding in the first place.
>>
>>101980346
Sounds like daisy
Daisy uses both the current reply and the previous reply to make a sentence



Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.