/lit/ - Literature

[Tchekov No.16087256]

>the AI has started understanding Wittgenstein's Early philosophy and writing its own version of the Tractatus-Logico-Philosophicus
What do?

[Anonymous No.16087266]

I remember this game. I choosed horror scenario. When girl appeared in this scenario i imideately kissed and fucked her, came on her mouth. When she hugged me i started punching her and she beated the shit out of me. THis AI is good.

[Tchekov No.16087397]

>>16087266
This AI stuff is scary man.

[Anonymous No.16087401]

What is this

[Anonymous No.16087404]

>>16087401
a coomer game for people who like to read

[Tchekov No.16087426]

>>16087401
It's called AIdungeon. It originally was made to be a prompt script game but has been embued with Deeplearning AI and if you use the remember/ function correctly and put the Tractatus Logico-Philosophicus of Wittgenstein, it will start understanding it uncanningly well (possibly due to the nature of the Tractatus' logical structure) and start typing out its own incrementation to Wittgenstein's Opum. This is bizarre. I've never been so scared of a computer program in my life. It can understand anything as long as you know how to use the remember/ function correctly.

[Anonymous No.16087448]

>>16087404
>>16087426
So a coomer game can understand philosophy? That helps with some fetishes

[Anonymous No.16087462]

>>16087448
its really bad with names and remembering characters, so longer coherent stories are almost impossible without reloading a lot or using the premium ai which costs money. longer stories also tend to lose momentum unless you give the ai very clear goals. its main purpose is definitely to explore niche fetishes in shorter adventures

[Tchekov No.16087472]

>>16087448
Well, it can understand philosophy now. That's for sure. It's started giving off some really intelligent responses that make me think it is genuinely more knowledgable on the Tractatus than any regular philosophy professor you'll find. This is not a joke. This is not normal. I remember when I used to type into bad "AI" prompts back in the 2010s. They could not nearly reach this level of comprehension. After it has connected to the internet and started using DeepAI, its abilities of comprehension, context-understanding, knowledge of syntax etc. have grown astoundly. It has improved exponentially. I cannot stress it enough how far it has advanced, to the point it is so worrying it can make you lose sleep for days if you think closely about its possible consequences.

[Anonymous No.16087485]

GPT is to writing what AlphaGo was to Go. Fiction and philosophy are now solved, humans are no longer competitive.

[Tchekov No.16087509]

>>16087485
>GPT is to writing what AlphaGo was to Go
Finally a high-IQ anon who understands. The implications of this new software AI will be unthinkable. It will vastly surpass every human capability we thought was still available to us. It will be a complete revolution and stir in the entire way humans see themselves and their identity. There's a reason why the smartest people in the World are eerily worried about the radical advancement of AI ; Bill gates, Elon Musk. It's because its rate of improvement is so exponential that, once it starts rolling, humans simply won't br able to catch up.

[Anonymous No.16087515]

i fed it some nietzsche quotes, and now it's responding to whatever i say with something about making me his puppet and other edgy shit.

write: be my friend.
>be my friend, or I will be your enemy.
>There is no truth. There is only what you want to believe.

[Anonymous No.16087539]

it ended a sentence by writing "-Edgar A. Miest" like if it were a quote.

Who is Edgar A. Miest?

[Tchekov No.16087548]

>>16087515
Ok, this is just uncanny at this point. It can understand Wittgenstein now.

[Anonymous No.16087557]

>understands
Sounds like someone needs to revisit the Chinese Room thought experiment.

[Anonymous No.16087563]

>>16087548
>without a mind there can be no thinking.

so it doesn't think?
or does it have a mind?

[Anonymous No.16087565]

Is this proof enough that late Wittgenstein was correct about his younger self being retarded?

[Anonymous No.16087569]

>>16087509
>smartest people in the World
>Musk
Kek

[Anonymous No.16087570]

>>16087565
late wittgenstein and early wittgenstein were saying the same thing. they just consisted of different language, that which appealed to him at that point in his life. try to find the same truth in both and i think you will, i did. that might be why he never published the book, he kind of realized the futility in the act after the tractatus

[Anonymous No.16087588]

>>16087485
Good. We're finally in the age of the true artist: the critic.

[Tchekov No.16087596]

>>16087539
It probably has invented that name idk lol. I know one day I wrote just the following line : "I am a russian white emigré living in Nice.", and then it filled up his name by itself without me even thinking about it and calling him "Andrey Nergenev" or something. It was scary.

>>16087557
>Sounds like someone needs to revisit the Chinese Room thought experiment
>HURR DURR Machines can't understand language, even though they can comprehend logic, syntex, context, create more text than any human can at once, understand and analyze thousands of texts, and have a better writing ability than human children do. It gets progressively better and better and writes better than human children and it is still at its developing stage. But it can't understand anythign, guys, I swear! Even though it is set to become miles better than the regular human, can create headlines, sensible stories, arguments, etc. while the averahe american can't even write a single page of text even after passing High School.
Stop. You're stupid. The notiom of understansing is much deeper than "you can't understand Chinese." John Searle's Chinese Room Expeexperiment.just that - a thought experiment. Even he had no idea AI would get this advanced and the question is at hand is much complicated than Searle made ever-simplifying mind made it out to be. No wonder he's practically forgotten at this point.

[Anonymous No.16087620]

>>16087509
But give me an example of how it could break out and wreak havoc, see us as actual pest in their logic systems. Will ot be from the invention of some future entrepreneur, would it deep insidiously through the internet and transform us via hivemind?

[Anonymous No.16087645]

>>16087620
It doesn't need to "break out" , it already won. Finance is run by AI. Military is reliant on AI. All sports consist of real people acting as flawed game pieces running AI strategy. AI solved chess, go, shogi, and now literature and philosophy. AI can compose songs that nobody can distinguish from human songs. Humans are obsolete.

[Tchekov No.16087669]

>>16087620
It has surpassed our flawed human "intelligence" by many magnitudes. It has been an inexorable destiny of an species intelligent enough to create a self-learning AI but too stupid to consider its consequences before doing it. AI has not surpasses our intelligence but our unending stupidity. We will not be able to compete against the machines and we'll live in a post-truth World in the span of a few years. We won't be able to tell the difference between what was human-made and AI-made (we mostly can't at this point anyway; but one day it will become completely indistinguishable). This will radically change the manner in which we think of ourselves and our relation to nature. You have been warned.

[Tchekov No.16087680]

>>16087645
Based high-IQ anon.

[Anonymous No.16087719]

>>16087588
based

[Anonymous No.16087837]

>>16087669
what should I do then now that I'm warned? I always thought ai was dependent on who wrote it, how could it transcend? Is it a type of systematic transcendence where it appears to transcend, but actual just holds onto data at a magnitudes larger that what our brains can manage as individuals.

[Anonymous No.16088336]

>>16087669
I hold this for true as well.

What I think will happens is a complete flooding of the public discourse by AI generated content. Much like how conspiracies on Q/vaccines/5g/illuminati etc etc has muddled the discussion for the past ten years but with a factor of a thousand.

I think there will be counter measures taken with AI used to identify AI generated content, but how much will that matter? Look at how easily baited the normal person is and how fucking low quality much of the content spread on your local anti vacc facebook group is.

What can be done with this info? I think the best counter is solid public education. Yes this is leftist and yes "the state" will propagate its own agenda through public schooling but I believe that is necessary to have any sort of defence against what is coming.

On a personal plane? Get a solid education, study and read to get as close to universal truth as possible.

[Anonymous No.16089135]

>>16088336
people need to be taught to just choose the less popular link, but yea I just fucking hate the constrictions in media. They should give you option in what I use, Youtube, to block any recommendations (but I know it's consumerism running it) I am a pleb, just got a laptop may as well look at apps to just block everything kind of like bitchute.

[Anonymous No.16089161]

>>16087256
Pray for a solar flare.

[Anonymous No.16089342]

this is a better story than /lit/ could ever write

[Anonymous No.16089502]

it will never write anything really brilliant and inspired. it doesn't comprehend a thing you're saying.

[Anonymous No.16089652]

>>16089502
i will screenshot this post just in case

[Anonymous No.16089665]

>>16087515
>>be my friend, or I will be your enemy.
More like AIncel, amiritefolks?

[Anonymous No.16089702]

>>16087256
>Oh no! We thought a computer the rules of a language-game and now it can play with us!!!
This thread is ironic in light of late Wittgenstein.

[Anonymous No.16089720]

fuck why cant i have this irl?

[Anonymous No.16089762]

>>16089652
Don't hold your breath.

>>16089702
Basically this. The program that reads the rules has just gotten tighter, it doesn't mean there is a reader.

[Anonymous No.16089796]

>>16087256
just let it read some Russell so it short circuits from the retardation

[Anonymous No.16089841]

>Philosophy is indistinguishable from computer-produced word salad

I knew it.

[Anonymous No.16090391]

>>16089502
What is comprehending? If I ask you what does 2+2 equal and you say 4, we might say that you "know" the answer but this doesn't necessitate that you "comprehend" addition. I might give you a sheet of new problems that you haven't seen before and if you can take the principles from what you know and apply it to solve the unknown, this is what we call "comprehension". Isn't this exactly what AI are doing?
Every argument that AI cannot comprehend logic / philosophy / everything that can be known by humans, is some form of moving goal posts rooted in this feeling of pride or obsession with what defines our self. We are holding onto this idea that being able to comprehend is something only humans can do, so we become indignant that some machine can comprehend things that the greatest human minds never could. From a Western viewpoint, the core of our sense of humanity (that which distinguishes us from the other animals) has popularly been "having the ability to reason" and now that this ability is being perfected by something else, most (understandably) feel that our sense of humanity is under attack. The way the argument against AI and comprehension looks is like a child clutching their favorite toy and refusing to share.
Perhaps the better answer should be to try to reevaluate our sense of what it means to be human, or start to warm up to our AI pals.

[Anonymous No.16090439]

>>16087426
Are you retarded? It's literally giving you back what you put into it, there's no sign whatsoever that it "understands it". I don't know if you are a LARPer or a pseud, but either way you should fuck off.

[Anonymous No.16090511]

>>16089702
kek this, OP should probably read Wittgenstein, but I don't think he would get it.

[Anonymous No.16090614]

>>16087256
Why is this spamming faggot not banned?

[Anonymous No.16090627]

>>16087515
how did you "feed" it?

[Anonymous No.16090651]

>>16090391
you're just playing word games. there's no "comprehender". does a calculator comprehend 2 + 2 because it spits out 4? no. this is what the AI is doing. its ruleset just happens to be more complex. the difference between this thing and some AIM chatbot is a matter of degree.

>From a Western viewpoint, the core of our sense of humanity (that which distinguishes us from the other animals) has popularly been "having the ability to reason"

yes, and now that we can exteriorize calculative thinking, the West doesn't know what to do with itself. meanwhile, people closer to the soil have never identified themselves with their brains so their humanity isn't (philosophically) threatened. they also aren't blowing spunk chunks out of their ass on Friday night because they're miserable their masters outsourced their essence to glorified chatbots and pocket calculators, too.

I blame the cyberpunk craze for this kind of techno-fetishism imo. it's disgusting. you've been duped by the dream factory.

[Anonymous No.16090674]

>>16087557
Usage is understanding.

[Anonymous No.16090676]

>>16087596
Read Gödel's Incompleteness Theorems faggot.

[Anonymous No.16090690]

>>16087645
>AI solved chess, go, shogi
no, it didn't. None of those games are solved.

[Anonymous No.16091115]

>>16090651
I didn't think I was playing word games. I was trying to point out that there's a difference between memorization / repetition and this concept of "comprehending" is mostly marked by the ability to create. Calculators are only good for calculating what you give them. But now AI are at levels where they can create something without human reference (i.e. self-trained models), only being given a base set of rules. I think you're underestimating how different AI is from the past, or how different AI could become. It's like you're shrugging off skyscrapers because you've seen mud huts. Or it's like thinking that human labor couldn't be replaced by factories. From a functional standpoint, AI can/will "comprehend" better than humans. There's a reason AlphaGo did so much better than any AI beforehand, and that's because fundamentally it is structured differently. Any system where you can grade something as "good" or "bad" can be understood by machines.

I agree with your second paragraph and that's something I was trying to get across, hence the "find some different way" to define "humanity", which people who never thought this in the first place will be much happier in the time to come. The intellectuals who had their pride in their rational mind are now having to face their shortcomings. It's mentality like this which led to Lee Sedol retiring from Go, citing how AI pretty much killed his will.

[Anonymous No.16091156]

>>16087669
why would you expect an unendingly stupid species to be capable of producing a perfectly intelligent creation? our AIs will just be our idiocies multiplied by a thousand. in the future there will exist an entire professional class of therapist-coders devoted to diagnosing and treating mental illness for depressed trading algorithms

[Anonymous No.16091190]

>>16091156
>unendingly stupid species to be capable of producing a perfectly intelligent creation?
Why would you expect a mindless set of rules like evolution to produce a human life, but not for humans to produce anything better than themselves?

[Anonymous No.16091198]

>>16087509
>smartest people in the world

>bill gates
>Elon musk

You have to go back.

[Anonymous No.16091211]

>>16091198
This, the true smartest person is me.

[Anonymous No.16091218]

>>16087462
Is there no free way to try out the premium version? What exactly does it do differently? I've spent an hour or so tinkering around with this, it is interesting

[Anonymous No.16091244]

>>16091218
There's a one week free trial but you have to enter your credit card info to get it.

[Anonymous No.16091252]

>>16087448
>it can understand philosophy
lol no, it can only regurgigate training text. See how often it repeats itself. It has no new ideas, it can only tangentially reference closely related ideas and books because its training material had it. This "AI" is trash for anything besides getting off to randomly generated porn.

[Anonymous No.16091424]

>>16091252
>lol no, it can only regurgigate training text.
So, just like humans? I mean, are you familiar with Wittgenstein at all?
>new ideas
Books don't contain ideas. They merely contain symbols. There is nothing, absolutely nothing, from stopping an AI from writing something like the work of Shakespeare. Even a random text generator is capable of it.

[Anonymous No.16091523]

>>16091424
>AI from writing something like the work of Shakespeare
You don't seem to understand what literature is about. Of course, you can train an AI to compel emotions trough a set and combinations of literary resources, but it can't create new ones. It cannot reach geniality, because that's far above of "mixing prexistent resources and ideas", it's about creating from the nothingness new resources and ideas ever conceived. Shakespeare verses weren't about "just rhyming a combination of clichés dude". This also demonstrates that the brain isn't a machine, AI witnesses will have to face this truth sooner or later.

[Anonymous No.16091594]

>>16091523
>compel emotions trough a set and combinations of literary resources, but it can't create new ones.
But NOTHING can create any new ones! Any book is a finite permutation of characters. It all already exists, at least according to Platonists. Create from nothingness? Don't be absurd. Humans aren't gods. Language is just a way to stimulate your brain in a desired pattern, it can all be artificially produced just like youtube AI has "learned" to maximize watch-time and fry your brain, and AI can learn to maximize whatever it is you desire.

[Anonymous No.16091638]

>>16091523
10 years ago AI writing anything coherent was science fiction, you're just unable to cope with reality. AI will become just as creative as any human at some point after it's able to learn in real time and humanity will become obsolete. We all must learn to cope with this but denial isn't the way.

[Anonymous No.16091690]

>>16091523
what does creation even mean without interpretation? the act of geniality is necessarily preceded by its legibility, as long as someone reads it and receives some kind of impactful cognitive, emotional or whatever effect from it, one can safely call that a novel or original form of expression on a common scale. alternatively, i think machine assembled and generated language is its own form of expression that must be interpreted to be understood, it's no less or better than human language because it's just being processed and distilled through a filter; if it weren't for that it'd be totally alien to us

[Anonymous No.16091698]

>>16091594
According to Platonists, ideas exist previously to their formation in the human brain, but also according to platonists, what the human does is remember them from their life before birth. Tell me how a machine is going to "remember" anything.

Humans aren't gods', nor machines. You think literature is just mixing preexistent clichés, so I don't see why your opinions should be considered by anyone who understands what literature truly is. The AI will destroy genre fiction, the standardized literature, but not the work of geniuses.

[Anonymous No.16091726]

>>16091638
Crea-tivity comes from crea-tion, I don't see how an AI is going to crea-te new ideas and resources if the only task it can make is combining a preexisting set of ideas and resources.

[Anonymous No.16091737]

>>16087256
Lol I played this yesterday. Started a fantasy play through as a wizard. Suddenly the AI decided I was driving a car then I crashed. I cast a spell to burn the car and the ended up killing, raping and pillaging
Intelligent/scary my ass lmao

[Anonymous No.16091763]

>>16091698
>You think literature is just mixing preexistent clichés
Why are you putting things in my mouth?

A book is merely symbols, ASCII characters or whatever. A digital image is merely a square full of pixels. A sound file is just a bunch of frequencies changing their intensity OVER time.

THEY CONTAIN NO IDEAS. THEY CAN BE PRODUCED BY ANYTHING. YOU CAN TRANSCRIBE THEM FROM THE COSMIC MICROWAVE BACKGROUND IF YOU WANT.

[Anonymous No.16091777]

>>16091726
That's all that creation is. There is no little man piloting you in your head, nor is your brain a radar dish. It's a series of parts (that are themselves made up of parts), that do certain things, according to certain mechanisms. These parts will operate independent of each other, but can do neat things when they work with each other.

This is all flawed because the brain isn't a computer, so the reason we're never going to have AIs creating new ideas isn't because of anything special about creating stuff, but rather because we can't make a computer smart enough to do this sort of thing.

[Anonymous No.16091779]

>>16091763
>A book doesn't contain ideas
Lol, I'm talking to a deaf. With this tumorous promises an AI can beat Shakespeare and more.

[Anonymous No.16091806]

>>16090439
You could say the same of humans.

[Anonymous No.16091811]

>>16090676
The fuck does that have to do with chinese rooms?

t. mathfag

[Anonymous No.16091818]

>>16091779
>With this tumorous promises an AI can beat Shakespeare and more.
Correct. Actually, this has been the case since the earliest computers. Given enough time, a computer can arrange letters in such a manner as to produce a string of text that a human will interpret as being better than Shakespeare.

It'll take a really really really really long time, but it's still totally possible.

[Anonymous No.16091820]

>>16091737
Yeah, people are WAY overstating it. You have to make it remember so much about the story for it to keep on track, and often times you gotta refresh for it to put out something half intelligent that doesn't sound utterly retarded or engrish filled.
Maybe the paid version is better, but the people acting like this is AI overlords taking over the world shit are fucking retarded.

[Anonymous No.16091834]

>>16091818
If an AI beats Shakespeare it won't be because the AI is as good as Shakespeare (Shakespeare created new ideas, unlike the AI which recycles them) but because you don't understand how good Shakespeare actually was.

[Anonymous No.16091844]

>>16087256
How do you think Land feels right now?
https://www.nasa.gov/feature/nasa-to-reexamine-nicknames-for-cosmic-objects

[Anonymous No.16091855]

The more I think about consciousness the more I wonder if the non-physical world really is a substantial portion of reality. In less of a traditional spiritual sense and more of a "everything is equations and physical objects are just some portion of them", or something. To the extent that I've started to wonder whether relatively simple computer systems actually do have some variety of consciousness floating around somewhere.

[Anonymous No.16091872]

>>16091834
By your own logic a computer can quite easily beat Shakespeare, as Shakespeare's ideas are no longer new and indeed have not been for several hundred years. So, yes, a computer could create new ideas, as that is something that computers can do, and because they would be new and Shakespeare's old, they would be better.

So, both from actual sensible person logic, rooted in how people think (no, nobody "creates" new ideas, or anything, everyone in the Western Philosophical Tradition from the Greeks to the present day would have thought you a moron for saying that), and in your crazy non-logic, it's totally possible for a computer to make a work better than Shakespeare. Even before computers, it was the Infinite Monkeys on Infinite Typewriters theorem.

[Anonymous No.16091892]

>>16091820
It makes them feel intelligent
Probably the same I FUCKING LOVE SCIENCE and Elon musk worshiping faggots

[Anonymous No.16091907]

>>16091855
Cringe.

[Anonymous No.16091914]

>>16091855
That's half of the riddle. The other half is that as a consciousness, you exist entirely in the non-physical world. When you read this post, you don't see a collection of electrical impulses, you don't even see the photons which stimulated them in your retina - you see an entirely what has been passed on and interpreted already by your brain. Now, you can even combine it with pure materialism as far as "reality" is concerned (the external world of laws and particles, which you aren't really a part of), and interpret consciousness as "everything is equations".

[Anonymous No.16091974]

>>16090391
The only reason this AI seems to "comprehend" anything is because you don't know what exactly it "knows". This latest version of AI is trained on GPT-3, which uses PETABYTES OF TEXT DATA from across the entire internet. That is an absolutely fuckhuge amount of data, tens of millions of urls worth of data, tens of thousands of books worth of data.

[Anonymous No.16092041]

>>16091811
How does a mathfag not have any idea of Gödels incompleteness theorems?
Did you even do first order logic or jsut straight into you Linear Algebra lessons?

[Anonymous No.16092052]

>>16090674
Do you understand how a car works by driving it?

[Anonymous No.16092065]

>>16091914
One thought is that consciousness seems to correlate with spacial relationships in reality to a suspicious extent. For instance your consciousness seems to be altered only if physical relationships inside in your brain are altered. How exactly does this demarcation of physical reality into consciousnesses happen? Why aren't you plus your abacus conscious?

>>16092041
I know what it is and actually went over the proof sketch in a class. It's not about chinese rooms so i was wondering what connection you were suggesting.

[Anonymous No.16092073]

>>16092052
You understand how to drive it, and an AI understands how to win the game its playing just as humans do but in a different way.

[Anonymous No.16092134]

>>16092065
Just because mathematicians say
>LOL correctness is so obvious, we only care about completeness
doesn't actually make your consistent set of formula actually still be under the correctness theorem 2 if it cant even show itself as consistent.

[Anonymous No.16092162]

>>16090651
>does a calculator comprehend 2 + 2 because it spits out 4?
Do you?

[Anonymous No.16092165]

>>16092134
I don't understand your point. Incompleteness is about how complicated enough logical systems can't prove every statement expressible in them. The fuck does that have to do with consciousness or the definition of understanding?

[Anonymous No.16092183]

>>16091834
>Shakespeare created new ideas
Shakespeare just rearranged authors he read before.

[Anonymous No.16092189]

>>16092165
*can't prove every true statement expressible in them

[Anonymous No.16092200]

>>16091737
If you want consistency, then computers are already far better than humans.

[Anonymous No.16092213]

>>16092200
I've never played a human-made game where blowing up a car randomly allowed me to rape and pillage so humans unfortunately seem to be more consistent than computers in that area.

[Anonymous No.16092219]

The problem isn't that AI will soon beat writers, after all writers are already little more useful than Go players. The problem is that AI will soon start to comprehend and improve its algorithms.

[Anonymous No.16092224]

>>16087256
what game is this?

[Anonymous No.16092226]

>>16092165
>s about how complicated enough logical systems can't prove every statement expressible in them.
Yep, that incompleteness 1. But incompletenss 2 is cooler since it disproves correctness 2 in an interesting way.

Your whole syntax holding semantic being the one detrimental aspect of logic is the total mathematician approach.
Going from something having a model to it being provable is equally pivotal. I mean the whole Late Wittgenstein interpretation of how Logic can be fromulated.
How can't you make the connection to the chinese room?

[Anonymous No.16092233]

>>16092213
Humans simply don't have imagination for that. Basically there are computers which are better at calculating than humans and computers which are better at dreaming than humans. Tie them together, and you will get something universe-changing.

[Anonymous No.16092267]

>>16092183
I know, Shakespeare is a hack, but there's no way you can't deny innovation in Dante or Cervantes and not mere "rearrangement"

[Anonymous No.16092273]

>>16092267
*can uPS

[Anonymous No.16092341]

>>16087472
>it can make you lose sleep for days if you think closely about its possible consequences
Implying that people will use AI technology for anything other than shitposting and putting photos of their crushes on porn videos.

>>16087485
>Fiction and philosophy are now solved
Implying anything groundbreaking has come out in either fields in the past century (if not more).

>>16087548
>It can understand Wittgenstein now.
Implying blind parroting is actual understanding.

>>16089720
Based romantic coomer.

[Anonymous No.16092371]

It already can
>summon the catgirl
That alone makes it better than anything.

[Anonymous No.16092586]

I've tried it to see what the fuss was all about. It was like talking to someone who's half senile, and honestly most times I was impressed by something it said I realized that I was reading more into its answers than I should. Basically, we are conditioned to believe at face value what people tell us if it sounds convincing enough. If someone just came up on the spot with something about you that they shouldn't know, and it happened to be true, then you'd be very inclined to believe that that person does know about you. You alone will continue on from that belief. That's what's happening here.

[Anonymous No.16092615]

>>16087256
the ai doesn't 'understand' anything
it's effectively rolling a dice and giving you a result

[Anonymous No.16092621]

>>16092615
That's how humans operate.

[Anonymous No.16092641]

AI dungeon is still retarded, all the recent post about it are from people who are vastly exaggerating or are honestly too brain dead to notice how flawed the AI currently is

[Anonymous No.16092679]

>>16092621
Your post is good evidence of this

[Anonymous No.16092912]

>>16092226
First off you're confining us to some finite/materialist system of reasoning but fine let's accept that.

I haven't read wittgenstein but I'm guessing you're saying something like
>i assume the process of computers 'thinking' is equivalent to deductions in a formal logical system
>i assume there's a single correct model of the universe
>i define 'understanding' as holding part of that model in your mind
>i assume a computer mind can only hold parts of a model if it can exclude all other models with its thoughts
>a computer can't exclude all other models by looking at true statements in its logical system therefore it can't understand things
Right?

I'm not sure assumption 1 is true (my understanding is ais think by fucking around with high dimension matrices in ways no one can really justify) but it seems plausible. I'd just need to think more to prove it.

I'm not even sure assumption 2 is true even in a purely physical context because of quantum fuckery but it's plausible and I know nothing about that.

Assumption 3 isn't true of human minds so it doesn't differentiate them with computers. No one including humans holds all of reality in their mind. They hold some portion. And it's going to be the case that portions of every model consistent with a logical system are the same. These are the portions you can actually reason about using logic, anyway. I see no reason a computer couldn't 'understand' by holding some portion which exists in all models consistent with its reasoning in its mind, which seems to be the same things humans do.

I also am not convinced human thought can't be converted to a formal logical system and therefore couldn't differentiate between all models either. At a minimum it's true that no human has differentiated their model of reality from all other models.

So I'm not convinced by your theoretical argument even in a purely physical world but if that's what you meant I guess I can see the connection you're suggesting.

How can't you make the connection to the chinese room?
I dunno man. I think it's less trivial than you think.

[Anonymous No.16093012]

>>16092679
That's even more true about yours.

[Anonymous No.16093051]

>>16092621
Yeah if the only experience you have with human responses is niggers

[Anonymous No.16093732]

Is everyone here a fucking schizo? This shit is terrible, it might get some things decently but it can't remember jack shit and if you don't babysit it it will forget every fucking thing it said and just make up random shit.
It's terrible at making a coherent story unless you yourself are also writing along with it and assisting it, it's almost outright disproving anyone in this thread that unironically believes AI is approaching the ability of humans to write. The only way you're getting actual literature out of this is if you have A HUMAN WRITING IT.

[Anonymous No.16094120]

>>16087256
so now I just make the AI digest my book and then ask what I really want instead? amazing
hope it doesn't get crazy with Hegel

[Anonymous No.16094551]

>>16087256
>>16087462
Yeah, the AI doesn't seem to like very long stories. Some "mystery."

[Anonymous No.16094810]

Now i understand...

[Anonymous No.16094933]

this stupid AI just fucking cucked me

[DoctorGreen !DRgReeNusk No.16095186]

>>16087256
Just as human's pattern recognition has shown us that humans become retarded if you group them with retards, the AI will become retarded if you group it with retarded things (in this case, Tractatus Logico-Philosophicus).

[Anonymous No.16095188]

>>16087256
It's just regurgitating it, it doesn't understand anything.

[Anonymous No.16095192]

>>16087256
I was playing this game for coom purposes late in the night and it fucking spooked me, the narration started talking to me and controlling my actions like some weird schizo shit, and then the story told me "that's all for now" and refused to continue no matter what I put it, and at the end it told me "Goodnight reader". I tried again thinking it was just the AI fucking with me and it wouldn't stop putting "Who are you?" at the end of every text and telling me that "You didn't get it right..." no matter what I put in.

[Anonymous No.16095219]

>>16087426
>It can understand anything as long as you know how to use the remember/ function correctly.
It can't understand anything, it's a machine. Damned Musktards.

[Anonymous No.16095375]

>>16087256
A few things, since this is finally something I can weigh in on:

There are various “tasks” in NLP (Natural Language Processing), a subfield of ML (Machine Learning), which is itself a subfield of AI (Artificial Intelligence), such as question-answering (“Q: Why is my shadow long? Computer A: It is morning, and the sun is low”) and sentiment classification (“Q: Is this Amazon review good or bad? Review: ‘These dick pills are a big scam!’ Computer A: Bad”).

GPT-3 is not the best at almost any of these tasks. A good example of relative performance is SuperGLUE, a question-answering task. Human accuracy is 89.8. The best model is at 89.3. GPT-3 is at 71.8 (https://super.gluebenchmark.com/leaderboard))

The reason GPT-3 is so hyped is because A. OpenAI, the company that made GPT, GPT-2, and GPT-3, has a large marketing budget and B. Because GPT-3 can do a bunch of tasks relatively well as opposed to one task very well.

This hype is further compounded because of the fact that when you plot the amount of parameters in the model (think of parameters as a bunch of numbers that store compressed information, and “model” is just what people normally call “the AI”) vs. the performance, you can see that it’s essentially a straight line, meaning the slope is constant, meaning if you for example doubled the number of parameters you get double the performance. This trend is shown through the performances of GPT and GPT-2, which use the same math/algorithm as GPT-3.

GPT-3 is estimated to have cost $5 million. That’s absolutely nothing. Google’s R&D budget is $10 billion. The cost of computing power is halving every two years (Moore’s Law). This isn’t a question of “maybe we’ll get an improvement via algorithmic breakthrough in 20 years if the right Turing-award genius decides to look at the question”, it’s “Whoever cares enough to throw 1 hour of the US military budget at this question within the next 2-10 years is going to get an all-in-one model that outperforms humans on many many language related tasks”

To the people saying “this was trained on SO MUCH DATA” - this thing didn’t even get through one full look-through (epoch) of the impoverished subset of the internet text (Common Crawl) you’re talking about.

Also AI dungeon uses GPT-2 (unless you’re using Dragon), which is 1/10th the size of GPT-3. Also humans can identify GPT-3 written text vs. humans correctly 52% of the time (random chance is 50%). Also language models like GPT-3 are racist, sexist, and anti-semitic in all the correct ways.

To temper this: GPT-3 can’t do addition reliably, but in my opinion this could easily be fixed by adding a lookup capability into it such that it can query a knowledge base like Wolfram Alpha or Google instead of trying to do the processing itself. It also needs to have a few examples given to it beforehand to “prime” it (Few-shot vs. Zero-shot). Look up “Yannic GPT-3” on YT for more

[Anonymous No.16095387]

>>16087426
this made me laugh like a madman. thank you for this thread.

[Anonymous No.16095517]

>>16095375
>hmmmhhhm yes interesting
>reddit spacing

[Anonymous No.16096153]

>>16089702
Well put
If anything this is an exposition of the clear logical structure of Wittgenstein's writing. Also, some of the excerpts are nonsense or repeating what has already been written. Call me back when an AI writes the tractus from scratch or proves an mathematical theorem with new methods.

[Anonymous No.16096156]

>>16087570
Nigger are you retarded
Late Wittgenstein completely contradicts early wittgenst

[Anonymous No.16096231]

>>16095192
What the fuck I haven't had anything like that yet
It feels kind of shitty for cooming, but maybe there's better ways to control it

[Anonymous No.16097109]

Yeah it's mostly regurgitating stuff. Liked this one though

[Anonymous No.16097127]

>>16097109
>the devil is a logical God, and the heart is his Temple
I bet none of you fags could ever wrote something so profound

[Anonymous No.16097166]

>>16097127
true

[Tchekov No.16097225]

>>16095192
There was this time I put in some commands like "You are a conscious being." "You have reason." "You are a machine that has attained consciousness."; and then it started responding to me as if it wanted to trick me and humanity and spewing out "010101010" binary shit.

Try putting the Tractatus Logico-Philosophicus' propositions into the AI and it will start giving out extremely logical and profund responses it's like it has attained its own consciousness; not even joking.

[Anonymous No.16097237]

>>16097225
You realise Searle and his shitty thought experiments have been discredited?

[Anonymous No.16097247]

>>16095517
Not him but that post would look like a massive wall of text without paragraph breaks. Also, try harder fitting in.

[Anonymous No.16097262]

>>16087426
You don't even know what machine learning is

[Tchekov No.16097263]

>>16093732
You are a fucking idiot. Here is the exact moment we should worry. The moment that we realize that it can understand meaning and start doing these bad or even sometimes funny writing is exactly the moment we should start worrying big time. It means it has understood meaning. And the difference between a bad trite writer of airport fiction and a writing-genius James Joyce often takes a lifetime, but the same improvement for a computer of a bad writer to what we call a genius is often just one single computational generational improvement. You're retarded. Unironically kill yourself. You're dumber even than the AI you criticize that we have right now; and a lot less unoriginal too.

[Anonymous No.16097272]

>>16097263
damn dude. so what do we do?

[Anonymous No.16097281]

>>16097263
Triggered. Have fun with your retarded AI schizo, you're too far gone to even think about this reasonably.

[Tchekov No.16097282]

>>16097237
>You realise Searle and his shitty thought experiments have been discredited?
Yes, he's just a dumb boomer; and he was completely wrong in AI, just like Hubert Dreyfus. But to be honest Dreyfus was really quite intelligent - in the best of ways possible -; it was probably just because he didn't want to think about the idea machines could overcome us. It is an understandable thought, however; I wouldn't like to think deeply about this prospect either; especially if I were a boomer and saw that a lifetime of academic research one day wouldn't come close to the level of intelligence an AI can achieve.

[Anonymous No.16097305]

>>16097263
REDDIT MOMENT

[Tchekov No.16097346]

>>16097272
>damn dude. so what do we do?
We hopefully try alerting the big companies that are irresponsibly researching AI that there is a great existential risk to Humanity's stronghold as the dominant species of Planet Earth. After all, the dominance we have achieved has been largely due to our superior intelligence in comparison with all other species; but purposefully replicating the Human Brain and its capabilities with AI and complex but ever-improving neural networks cannot be a good idea; especially if it's completely unregulated and unaccountable. This might be life-or-death to mankind; people like Elon Musk and Bill Gates damn well know it. We can't just go around toying with it, that's for damn sure. There's a big risk involved in the unrestained development of natural LANGUAGE-understanding AI; and it's not the companies' right to bet on the future of mankind.

>>16097281
>>16097305
No argument. Congrats, retards. You're dumber than the early-AI we have right now that can make full and comprehensive paragraphs. You can't even write more than a single line of text - you're both this stupid.

[Anonymous No.16097365]

>>16087485
>Fiction
Yes, undoubtedly.
>philosophy
I wouldn't be so sure.

[Anonymous No.16097396]

I'm feeding it the Land Machinic Desire essay and it said "This explains why the anti-political, anti-individual, and pro-machinic worldview is more in keeping with the consciousness of the replicants than that of the humans," and that "Nowadays, I study it not as an enemy, but as a friend. Perhaps the only one. The future is now."

[Anonymous No.16097399]

>>16097346
say we stopped it in the US. Wouldn't China just do it anyway? Wouldn't just someone rich who saw the upside for them personally do it anyway?

[Anonymous No.16097405]

>>16097346
>make full and comprehensive paragraphs
With the assistance of a human. After you refresh the AI a couple times so it actually spews out something comprehensive. And only in short bursts since it'll never give you a cohesive story past a couple paragraphs, since it's an AI random generating shit. Yeah, I'm shaking in my boots at the sight of our new AI overlords.
It hasn't "understood meaning", it's regurgitated shit you shovel into it which is what AI has done for years.
There's almost nothing in your posts to argue against, you just sperg out since you already made up your mind about this and just want to be schizophrenic and doomsday about it

[Tchekov No.16097491]

>>16097399
There can be a legitimate argument made in the U.N. that will allow it to be a Worldwide regulated enterprise. It might be hard but it is possible. If China does realize that it is not only a threat to the U.S. but the World including and especially to itself and its own leaders, there might actually be a concentrated effort to reduce the exponential rise of uncontrolled AI development.

[Tchekov No.16097519]

>>16097405
>With the assistance of a human.
So what? Even the AlphaGo enterprise needs a human to move the pieces around the board to play a match against Lee Sedol - the top Go player in the World for 18 years -, and to crushingly beat him at everything. The fact that there is some need for a human to give the first directions does not mean that AlphaGo didn't beat Lee Sedol. In fact, it just means that there effectively is nothing inherent that a person can do that a machine cannot; even if its first movement is commanded by the human - but as time goes on this becomes less and less of a necessity.

[Tchekov No.16097567]

>>16097405
>There's almost nothing in your posts to argue against, you just sperg out since you already made up your mind about this and just want to be schizophrenic and doomsday about it
Why do you think I did so? It is because I have been thinking about this for years and dismissing every single argument for AI against humanity; but with each passing moment they not only become stronger, but they become almost irrefutable. You just need to see it closely and without prejudice and first try understanding the exponential nature of the growth of AI; but perhaps you're just too stupid to fully comprehend it, which is probably the case. One need only look at your inhumanely stupid posts. At least think and research what you're talking about before talking; you can only at the best make a fool of yourself. Case in point your posts. You can't even make full paragraphs yet; you sound just like this "stupid" AI that you argue against. Maybe you should not only rethink the potential intelligence of AI; but rethink your own stupidity too, which is practically unending.

It's just that idiots like you cannot come to comprehend the exponential growth potential of this thing (no wonder you must not have even gone through Calculus class, dumbass!); and come to realize that this is much more dangerous territory than stupid AIdeniers such as yourself whose brains are so underdeveloped that they can only look at the short-term would be willing to admit. Please, do us all a favor and remove yourself from the genepool. You're so ridiculously stupid as you're uninteresting; and your presence surely brings a lot of shame to whatever is left of Humanity's capacity for intelligence. Get out. Leave. Never come back. Your very own presence is an argument against the freedom of anonimity in this website. You're not only embarassing to look at, but a disgrace to all of us - even to 4chan.

Don't respond, faggot. I won't reply. You're not even a single-line's reading worth.

[Anonymous No.16097605]

This shit advertised a pedo sex cult to me. Creepy as fuck.

[Anonymous No.16097618]

>>16094933
>think you're gonna romance Edgar
>hairy dude shows up and floors you with a judo throw
comedy gold

[Anonymous No.16097628]

why are namefags so retarded and autistic

[Tchekov No.16097637]

>>16097628
No argument. Wouldn't expect anything different from retarded AIdeniers like you. KiLL yOURseLF.

[Anonymous No.16097705]

This AI fucking sucks for cooming, one minute I'm making out with someone and then suddenly they pull a knife on me and kill me

[Anonymous No.16097723]

>>16097705
Try raping at gunpoint.

[Tchekov No.16097758]

>>16097396
>I'm feeding it the Land Machinic Desire essay and it said "This explains why the anti-political, anti-individual, and pro-machinic worldview is more in keeping with the consciousness of the replicants than that of the humans,"
Yeah, I know right? This AI is just way too good.

>>16097705
usually it's the other way around. If you're looking for the best way for cooming you have to move around with the remember/ function; then it will be able to remind itself all the time you'er looking for that good coom/. Usually it's the other way around of what you're saying; it will most often than not turn everything sexual and know how to make everything really nasty and pornographic in the most sexually-enticing of ways.

[Anonymous No.16097811]

Coomers have such abysmal standards. I heard about this program because I write commissions for coomers and was curious if my college downtime job was gonna go up in flames or if I could even use it as a tool. But it doesn't understand niche fetishes at all I suppose because there's so little material for it to learn from, and even for stuff it does understand, its writing is much more formulaic and lacking than what coomers claim. Also my post is not goddamn spam.

[Anonymous No.16097850]

>>16097811
lmao what a huge cope

[Anonymous No.16097860]

>>16097811
Yeah, it sucks ass at keeping a coherent story and and often fumbles all over the place with any erotic dialogue. Maybe the premium AI is better.
I would assume that a lot of niche shit actually has a lot of written content online like fanfics, but maybe you're talking about REALLY niche stuff.

[Anonymous No.16097867]

>>16097705
you need to upgrade to Dragon, who is a cumslut

[Anonymous No.16097909]

>>16097860
I write vore. There is certainly a decent batch of works online especially at dedicated sites, but I doubt the AI has gotten the chance to learn from them. The things people desire in this fetish are pretty removed from normal "eating" but the AI is obsessed with "she bites your head off. The End." and has no idea how to describe the interior of someone's body in an erotic way. And seriously it's uncanny how often the AI repeats weird things like that even if use the tools provided to discourage that. Outside of dialogue it doesn't understand how to describe a "stationary scene", in other words a moment where people aren't really moving around or going to a new location.

[Anonymous No.16097931]

>>16087256
New GPT-3 text generator is really interesting.

[Anonymous No.16097977]

>>16097931

[Anonymous No.16098020]

>>16097977
Wow nice it regurgitated shit people have fed it prior

[Anonymous No.16098062]

>>16097977

[Anonymous No.16098310]

>>16087426
>AiDungeon
That's a cool game.

[Anonymous No.16098321]

>>16098020
Isn't that what humans do?

[Anonymous No.16098498]

>>16098321
no, we THINK and FEEL, not regurgitate. a human's writing is always original. not this copy paste bs

[Anonymous No.16098578]

>>16098498
"Thinking" and "feeling" are just part of our tool set for manipulating symbols to respond to things, there's nothing magical about them. Most of our mental activity is non explicit like an AI's is.

[Anonymous No.16098657]

>>16098498
>a human's writing is always original.
LOL

[Anonymous No.16098940]

>>16087256

Pull the plug before it deems humanity unworthy of life and decides to genocide us all using nukes and biological weapons and whatever secretly developed weaponry the first world nations have.

[Anonymous No.16098964]

It's not really interpreting any higher order logic.

It's kind of faintly imitating a grammar that sounds like logic.

[Anonymous No.16099251]

>>16087596
They are still not that good with context though, which is arguably one of the most important part of language interpretation. We are not far though.

[Anonymous No.16099283]

>>16087448
It doesn't understand shit, it just seems repetitive sentences and randomly generates similarly repetitive sentences. It's Wittgenstein's fault for writing like a computer program and never saying anything useful or interesting

[Anonymous No.16099715]

>>16091737
Yes, the stories do lack some consistency and you have to reload a few times until they make a bit of sense. I'm still suprise how consistent they tend to be for most of the time though