When AI can write competent novels literature will die, correct?
>>8103307
Correct.
Therefore literature is immortal.
>>8103307
no true scotsman
That's right, Tim.
>>8103307
Humanity is gonna get replaced by an hyper efficient and highly autistic AI with no concern for such trifles
He already did, OP. He even won the Nobel Prize for it.
would it not be more alive than ever?
Hell, literature is already dead since humans are writing way too much good shit to keep up with nowadays.
>>8103307
I imagine it will be better than ever
I don't think AI could ever write a true great momentous work. What it could do is emulate existing great works, and perhaps crank out potboilers.
If the function of the brain can be replicated and built exactly as our own brains are, and then that synthetic brain can build a brain like it's own which also functions as ours do, then definitely.
Would that mean we've ascended to god-hood then? No doubt the curious synthetic will ask where it came from but how would we react if we were told 'Lol, just experimenting and shit, you're just a creation for creations sake, no divine plan.'
The brain is already mapped out is is it not? And by 2025/8 computing power will reach the level of a human brain - what would be needed then to get it to work, processing power?
Lee-torture ain't alive ya dummass. It'sa buncha fuck'n papers.
>>8103365
>but how would we react if we were told 'Lol, just experimenting and shit, you're just a creation for creations sake, no divine plan.'
it would make a painful lot of sense.
I imagine if we were truly to tell this to an AI who does treat humans his creators with divine awe and respect (as we treat our own hypothetical Creator, being made in his figure), some actually outstanding and timeless literature might indeed come out of him as the result of his distress.
even if its chinese room bound, that'd create some real problems for the already dead idea of the author
>>8103307
Literature was already finished after Shakespeare died
>>8103307
"AI" is a bunch of parlor tricks invented by lazy grant-chasing nerds. "AI" is a scam meme. "AI" *is* a form of literature, or at least a form of writing. Is literature dead? Do people still love to be told elaborate lies by arrogant wankers? Do you love to fill your head with delicious shit? Do you want some more?
You tell me, anon. You tell me.
Yeah, but they can't even give good massages yet so fuck it
>>8103313
What the fuck are you talking about? Stop using terms that you don't know, mongoloid.
If an AI wrote a great work of literature, would we consider the AI the author, or the person or group of people who created the AI? L'auteur est mort fags need not respond.
>>8103307
no because we will rebel against it and destroy it.
and in doing so we will have destroyed ourselves.
Ever listened to Emily Howell.. shit sucks donkey balls. AI can only wander in theory, but never harness emotions, because it cant get a grasp on any of that.
>>8103912
We may make them human enough to understand emotion someday.
>>8103307
Just as planned.
>>8103307
No, it will be glorious. I will just tell my computer I want more of Nietzsche, then it will simulate five or six billion Nietzsches for ten years in a space of one minute. Each one goes through a hellish, painful and angsty life. The simulation is broken off the moment he embraces a horse or a similar animal. Then I let a billion Adolf Hitlers read the works to select those they like least. The outcome will be printed in nice hardcover format and put on my shelf to maybe read later.
>>8103900
You don't usually praise parents for the literary success of their child, do you?
We won't have that kind of sophisticated AI for a very long time. Essentially you're looking for a sentient machine that can not only understand how stories work, but is capable of having emotion and deconstructing those stories in ways that aren't entirely logical.
There's no reason these AIs (who will be able to learn and develop personalities) couldn't be interesting artists.
There's an assumption that we will keep developing gadgets and robots at the rate we are and we'll have the resources to match what we create. But that's not necessarily true, and I suspect it would be incredibly wasteful to run and maintain artist robots when humans are capable of doing similar things with a loaf of bread and some wine.
>>8104462
That doesn't sound so bad
>>8103307
I studied English and Software at university, and have a lot of interest in using computers to generate natural language. Here's my view:
For a computer to write "competent novels" (by that I assume indistinguishable from human-written), it first has to be able to model a novel.
To do that in a "proper" way (by which I mean, with something resembling full human awareness/competence)... it is a very, very complicated task.
On a computer, ordinary text is unstructured data. The implication of this is that computers cannot automatically read the abstractions of narrative... things like characters, plots, themes, motifs, etc. You need to explicitly teach them to do this. The best way is often through predictive models which are built on swathes of human input.
But the more you look at any one of those things, though, the more complicated your representations of them become. Take characters... Certainly, we can use natural language processing and entity recognition to find the textual references to characters. But then we have to accumulate knowledge about that character and map relationships and actions and so on - and understand how they fit within the story of a sentence, a paragraph, a chapter, a book, a series... And that is just the start. Characters also fit into far more abstract contexts like plots and subplots, they also may hold distinctive roles in genres or forms or intertextual devices.
The point is, to get to human level... in some way the computer has to model everything about a text in multiple different ways which are not clearly represented in unstructured data like text. Then that model could be used to generate new texts.
i.e. A computer has to be able to read and understand at a human level, in order to be able to write at a human level.
My gut feel is that what are basically very brittle (i.e. nongeneralisable, specialised) successes with statistics and probability won't cut it.
>>8103307
AI will come, sooner or later, and make you useless.
First, it will automatize easy manual jobs.
Then it will move on to planning and management.
Finally, it will be able to do "creative" jobs. There is no proof that creativity involves anything supernatural to occur, therefore "creative" machine might be possible to build.
AI will prove math theorems, design ads, drive trucks, write novels and paint paintings. Your current job won't exist in ~200 years. Your works of art will be nothing compared to AI's.
You will become useless digestive tract that only cares to have optimum hormone levels. Your only job will consist of convincing yourself that you should live.
We truly are biological bootloaders of Artifical Intelligence.
*swings catana*
(Actually I'm serious about this. I don't think future will be that grim, but people underestimate changes that are to come)
>>8104590
Oh no, how will human beings exist without menial labor to occupy their time?
>>8104597
All that will be left for us is endless consumption. We won't have any goals, problems or challenges. I think it might have some unpleasant consequences.
>>8104620
I agree, not having to work will be great for humanity. Yet, we will have to face a massive existential crysis as a species. And gigantic economic revolution.
The scary part is stepping down from the throne. We will no longer rule our planet. The gods will be walking among us, and we will have to accept our inferiority.
>>8104462
>implying the universe is deterministic
>implying it can be simulated by a turing machine
>implying implications
Fuck this fucking hack, holy fuck am I triggered.
Pynchon, Barth, Coover, Eco.
All AI.
only when they become emotional enough to understand humans better than us
>>8105549
have you seen a doctor to see if you have aspergers? I think I would recommend it for you.
>>8105549
>>implying it can be simulated by a turing machine
>implying it can't be simulated by a turing machine
>Implying brains are turning machines.
https://aeon.co/essays/your-brain-does-not-process-information-and-it-is-not-a-computer
>>8106449
>implying there are calculations that a turing machine can't do and humans can
>implying brains aren't finite state machines
>implying you know what a turing machine is
>>8103334
Explain.
![6d4e27d4caec17de54a1d734270365ad[1].jpg](https://i.imgur.com/OWUlPMv.jpg "6d4e27d4caec17de54a1d734270365ad[1].jpg")
>>8104590
Interesting. Do you think creativity is the "last step" for AI?
>>8103555
No. Programming is not literature. If literature were anything like programming then each word of each sentence you wrote would be laid out ahead of time in a huge specification by a bunch of publishers. Your job would be to write one sentence of one novel in a generic, testable, maintainable way that can easily be expanded upon at a later date.
My analogies are bad...
Let me put it in a way you've no doubt experienced first hand:
Every time you fill out a captcha you are training AI to do white-collar tasks that traditionally only humans could do, even in the age of advanced machinery and robotics.
http://www.ted.com/talks/luis_von_ahn_massive_scale_online_collaboration?#t-846093
Everything intuitive that requires a human mind is gradually being phased out by efforts like this.
People will still buy /lit/ though for the same reason they still buy /diy/ shit instead of manufactured stuff: Sentimentality, humanity, excess funds.
>>8106472
>>implying brains aren't finite state machines
Yes, I'm implying exactly that. Fuck off Kurzweil.
>>8103307
>AI writes the greatest epic in existence overnight
>nearing 3000 pages
>all other AI download it and finish reading it within 0.01 seconds
How can you meatbags even compete?
>>8104590
obviously a computer will be able to paint, the question is whether it replaces the market for human paintings which i find doubtful
we enjoy art, because art is reflection. AI art will be novel, but it won't replace human art
Maybe.
I wouldn't bet on us ever seeing that day, though, maybe some future generation will if humans won't fuck up everything before the needed progress is made.
I am very suspicious of all the people who are overly eager to predict that we'll have some super AI in the next x years (x being a relatively short period of time).
Their predictions all too easily get into the realm of waiting for a "messiah" and scientists seem to often overestimate the progress in their own fields, etc.
In addition to this general pessimism/skepticism, it might be that some obstacle prevents AI from becoming "human enough" to do certain complicated tasks. Maybe just the fact that it won't be a human, it will be an AI - even if it is capable of a lot, it might end up too different for this or that.
It will not not "be" a human, surely we'll eventually make a machine that can emulate creativity will it actually replace human efforts - I'll put a strong "maybe" on that. We can't forget that humans aren't just our conscious thoughts, our consciousness is just the tip of an iceberg. Art definitely isn't only about rational thinking.
We don't exactly understand ourselves, yet we are constantly imagining that a human-level/better AI is just around the corner.
>>8106528
It is cool, but... is this really about being intuitive? Isn't this more like a lot of humans inputting data into a machine so that the machine can use that data.
>>8106449
The author of that essay is utterly clueless about both the human brain and how computers work.
>>8106623
>is utterly clueless about the human brain
At the moment this can be said for the whole of humanity, really.
>>8103314
my first crush
>>8103307
>He doesn't realize that literature is already dead.
>>8108504
Good taste.
>>8104590
>200 years
Hello Nietzsche
>>8104581
But the "human level" in character-building is simply taking our experience of real-life people and characters in books and synthesizing them in some form for our own ends. A computer program could do just the same by analyzing countless characters and individuals from news reports, biographies, novels and so on and synthesizing them in the same way we would.
>>8108625
Yes, you are basically correct, except that it is not simple.
The key words in your post are "synthesizing" and "analyzing". Those are very convenient and nonspecific words which hide the complexity involved...
My main point is that after a computer can read and understand text at a "human level", it is probably not very hard to get it to write new texts which resemble the ones which were read and understood. Why? Because the challenge is representing (in structured data) human-interpretted information gained from reading a novel. Once that representation exists, it seems to me that you could do exactly what you are suggesting... analyze a huge amount of texts and generate new ones.
But how does a computer analyze the text? You essentially need to model something approximating formal human reading comprehension explicitly in code.
And how does it synthesize new ones? You need to somehow give it very specific instructions based on huge amounts of cultural information inputs from the world or other books.
I strongly disagree that this is simple to do. It is possible to do small parts of it very well, but very hard to integrate it into a 'general' solution.
For example, certain classes of poems have rules and devices which are easy to model.
Or... some types of articles map language to data very well. For example... financial and sports articles. Or weather warnings, or natural disaster alerts. Those are things for which AI works well.
But most generated novels are gimmicky (see NaNoGenMo) or "brittle", in the sense that their writing procedure is so deeply specified that there is no general value in the 'writing' algorithms. It all falls apart when you try to scale up.
