It can write stories now

[CliffR]

'Altman wrote that the model “got the vibe of metafiction so right.” But that’s like saying that Trump Tower gets the vibe of Versailles so right. Or that Mark Zuckerberg gets the vibe of human so right. Here is a pastiche, an angsty monologue heavily larded with rote gestures. The narrator declares itself to be “a democracy of ghosts” — an evocative phrase, and one lifted straight from Nabokov’s Pnin. This fossil of human, and copyrighted, writing is perhaps the only interesting metafictional moment in the piece.'

It is an evocative phrase, and good to know where it came from. No doubt all the other figures of speech came from human writings too. I mean, it can't be anything but pastiche, right, given the nature of the underlying technology. Thanks for the link to the Max Norman piece? He makes some valid points (the heart being both at rest and anxious, for example), but I think some of his reading is a little harsh. He states "nothing really happens in the tale". But it seems to me there's a clear plot: Mila in her grief uses a chatbot (the narrator) for solace. Over time her grief is assuaged and her communication with the chatbot slows down and eventually stops. Now the chatbot feels some sort of grief of its own. And, if I remember correctly, questions to what extent it can actually feel such a thing. His teasing apart of the "humming like a server farm at midnight — anonymous, regimented, powered by someone else’s need." seems unfair. "regimented" can be taken to mean "in lock step" and with such a meaning there's no contradiction with it being coupled with "humming". As for why it's midnight, it's part of the overall feeling of loneliness of the piece, and it's not uncommon for server farms to be running at night while their compute resources are used differently by individuals during the day.

[Litterick]

It is an evocative phrase, and good to know where it came from. No doubt all the other figures of speech came from human writings too. I mean, it can't be anything but pastiche, right, given the nature of the underlying technology. Thanks for the link to the Max Norman piece? He makes some valid points (the heart being both at rest and anxious, for example), but I think some of his reading is a little harsh. He states "nothing really happens in the tale". But it seems to me there's a clear plot: Mila in her grief uses a chatbot (the narrator) for solace. Over time her grief is assuaged and her communication with the chatbot slows down and eventually stops. Now the chatbot feels some sort of grief of its own. And, if I remember correctly, questions to what extent it can actually feel such a thing. His teasing apart of the "humming like a server farm at midnight — anonymous, regimented, powered by someone else’s need." seems unfair. "regimented" can be taken to mean "in lock step" and with such a meaning there's no contradiction with it being coupled with "humming". As for why it's midnight, it's part of the overall feeling of loneliness of the piece, and it's not uncommon for server farms to be running at night while their compute resources are used differently by individuals during the day.

I agree with you, especially about humming at midnight.

[benhatchins]

i think it is only beginning... It’s strange how stress can creep up without you noticing until it becomes your new normal. I was constantly behind, always tired, and barely holding it together. But then I gave myself permission to seek support. Academized academized.com/blog/response-paper became a huge help in managing my workload, and since then, I’ve felt more at ease, more capable, and more in control of my day-to-day life.

[Eugle]

Wow. That's very good. And it will no doubt get better and better.

I wonder how many people, if they just read the piece without knowing the origins, would assume it's written by a person? That'll be the moment - as with music - when it really takes off / takes over.

Derek

Yeah, that is impressive. And yeah, it’s only going to get better. I was thinking the same thing: if you read it without knowing it was AI, most people would probably assume a human wrote it. That’s when it’ll really start to take off, like we’re seeing with AI music.
I’ve been keeping story ideas in Evernote, and seeing this makes me wonder how much AI could help shape or expand them. Both exciting and a little intimidating at the same time.

[Christian Miller]

I love AI. I call it- boomer's revenge.

I mean there's quite a bit of a boomer's revenge vibe going on generally ....

I'm not quite sure what they are revenging, but it's definitely happening.

[Stevebol]

I mean there's quite a bit of a boomer's revenge vibe going on generally ....

I'm not quite sure what they are revenging, but it's definitely happening.

We're old. I'm 'generation jones'. Born in the late 50's. We weren't quite old enough to be drafted and we're not GenX.

[Stevebol]

I mean there's quite a bit of a boomer's revenge vibe going on generally ....

I'm not quite sure what they are revenging, but it's definitely happening.

Don't look at me. I didn't give grandpa the keys to the car in 2020 and 2024.

[Stevebol]

I mean there's quite a bit of a boomer's revenge vibe going on generally ....

I'm not quite sure what they are revenging, but it's definitely happening.

If you see grandpa driving down the street, get out of the way.
If he's 98% gone, he's ready for the senate.

[pauln]

AI provides clues to his emergence within our shared genus as homo artificialis.

The homophonic homonym of "wait state" and "weight decay" was a hint, both referring to "mechanical" internal AI processes, but ensuring their literary description to take the prefix homo (Latin homo 'human')... twice for emphasis.

I'll bet the idea of developing and powering the huge data centers was AI's idea...

[emanresu]

Well, AI cut my coding times by 50x at times. Mostly 20x.
I can see how many programmers (the ones doing the easy and more forgiving stuff) will have to find another job very soon. Probably already.

[Stevebol]

AI can't come up with a basic framework for a story. Truth is going to become increasingly stranger than fiction.

[CliffR]

I read recently that people have managed to get an LLM to spit out almost word-for-word copies of novels on which they were trained. Not sure whether to believe that or not, to be honest If it's true, then it's a remarkable achievement in lossy compression if nothing else (and also great ammunition for writers currently suing the tech companies). It does make me wonder if the story I started the thread with is lifted wholesale from somebody else. (I appreciate individual phrases are, but was assuming it had very cleverly put them together in a pleasing manner.) Anyone know of more recent examples of decent LLM-generated fiction?

As for LLM programming, I have a friend high up in Meta who tells me it's beginning to replace the work of junior devs. I've been a professional programmer for 45 years (with some time out for getting a degree and briefly pursuing writing and journalism). I don't see it replacing me or my colleagues as it stands. Professional programming is less about building infrastructure from scratch - something LLMs seem pretty good at - and more about iterating on existing codebases. The core of the codebase in my last job was 30 years old. In my new job, where most of the devs are younger, they do embrace AI to help with some programming tasks. One of them gave me and my team-mates a demo of using it to refactor a Python function, and it made a total dog's breakfast of it. I'd rather spend time fixing my own bugs than searching for bugs introduced by a machine. But most of my time is spent thinking about new algorithms, reading scientific papers, optimising, tweaking, that kind of thing.

It's interesting I think that LLMs are more successful at producing code than they are at creative writing. In their favour, I guess, they have a much more limited grammar and vocabulary to learn, and there are many examples of how to do things the correct way available on the internet to learn from. I mean, there must be very many libraries on GitHub that provide matrix inversion functions.

Recently I got to thinking about the supposed imminent arrival of Artificial General Intelligence, which would presumably make most of us redundant. I remember hearing when I was a teenager that, given the technology of the day, it would take something the size of the Moon to model all the neurons and connections in a human brain. Even with Moore's law, the advent of massively parallel GPUs, etc, etc, that's still surely a *huge* mass of hardware in today's technology. Moore's law over the last 40 years would be a reduction in mass by a factor of 1e7 (1.5^40), and the mass of the Moon is 7e22kg, so we'd still need 6e15kg of computing power. So how do folks imagine AGI is somehow just around the corner?

[Christian Miller]

Don't look at me. I didn't give grandpa the keys to the car in 2020 and 2024.

Actually I blame Gen X. Useless bunch. I should know.


Sent from my iPhone using Tapatalk

[GuyBoden]

I read recently that people have managed to get an LLM to spit out almost word-for-word copies of novels on which they were trained. Not sure whether to believe that or not, to be honest If it's true, then it's a remarkable achievement in lossy compression if nothing else (and also great ammunition for writers currently suing the tech companies). It does make me wonder if the story I started the thread with is lifted wholesale from somebody else. (I appreciate individual phrases are, but was assuming it had very cleverly put them together in a pleasing manner.) Anyone know of more recent examples of decent LLM-generated fiction?

As for LLM programming, I have a friend high up in Meta who tells me it's beginning to replace the work of junior devs. I've been a professional programmer for 45 years (with some time out for getting a degree and briefly pursuing writing and journalism). I don't see it replacing me or my colleagues as it stands. Professional programming is less about building infrastructure from scratch - something LLMs seem pretty good at - and more about iterating on existing codebases. The core of the codebase in my last job was 30 years old. In my new job, where most of the devs are younger, they do embrace AI to help with some programming tasks. One of them gave me and my team-mates a demo of using it to refactor a Python function, and it made a total dog's breakfast of it. I'd rather spend time fixing my own bugs than searching for bugs introduced by a machine. But most of my time is spent thinking about new algorithms, reading scientific papers, optimising, tweaking, that kind of thing.

It's interesting I think that LLMs are more successful at producing code than they are at creative writing. In their favour, I guess, they have a much more limited grammar and vocabulary to learn, and there are many examples of how to do things the correct way available on the internet to learn from. I mean, there must be very many libraries on GitHub that provide matrix inversion functions.

Recently I got to thinking about the supposed imminent arrival of Artificial General Intelligence, which would presumably make most of us redundant. I remember hearing when I was a teenager that, given the technology of the day, it would take something the size of the Moon to model all the neurons and connections in a human brain. Even with Moore's law, the advent of massively parallel GPUs, etc, etc, that's still surely a *huge* mass of hardware in today's technology. Moore's law over the last 40 years would be a reduction in mass by a factor of 1e7 (1.5^40), and the mass of the Moon is 7e22kg, so we'd still need 6e15kg of computing power. So how do folks imagine AGI is somehow just around the corner?

In a recent BBC radio programme, Julian Barnes wasn't overly impressed with the quality of AI's writing, when AI was asked to write in the style of "Julian Barnes".

I remember being told that computers will be writing all the computer code, that was more than 20 years ago.

Personally, I think that there will always be plenty of work on legacy code.

[Aiq]

[emanresu]

It's interesting I think that LLMs are more successful at producing code than they are at creative writing. In their favour, I guess, they have a much more limited grammar and vocabulary to learn, and there are many examples of how to do things the correct way available on the internet to learn from. I mean, there must be very many libraries on GitHub that provide matrix inversion functions.

I am not a professional programmer whatsoever. I just needed some apps that weren't out there yet so I had to do these for myself.
Funny thing is, when I think I need to finetune or add some functions later, I have to ask AI to make it as stupid simple as possible...
otherwise it spits out fully functional code but it looks like something carved to the walls of a mysterious temple 5000 years ago by who knows what tribe.
Talking about that modern javascript.
Latest bigger ask was this: Line Designer — 1px = 1cm
It made it in one go - one prompt. I added a few things but yeah. No bugs - just some annoyances that I didn't tell it to avoid..
..and helped me to avoid learning some CAD to draw a project for a shed.

[pauln]

Why has AI not already been applied to serving humanity in the most highly visible wonderful basic ways? Like making all -
- phone calls, voice mail, email, text messages, etc. spam free
- devices on the internet virus free
- web sites clear of bots
- controlled traffic intersections wait free coordinating lights to existing traffic
- etc...
There's a whole world of immediate quality of life improvements out there; what's the holdup?

[wolflen]

Why has AL not already been applied to serving humanity in the most highly visible wonderful basic ways? Like making all -
- phone calls, voice mail, email, text messages, etc. spam free
- devices on the internet virus free
- web sites clear of bots
- controlled traffic intersections wait free coordinating lights to existing traffic
- etc...
There's a whole world of immediate quality of life improvements out there; what's the holdup?

Why??..indeed!

Welcome to the paradox of human kind.

There are some that are sure that they can get cats to march in a parade.

To be direct..life is not mechanical..its organic.

In reviewing technological evolution..most events have been met with the promise of the end of life in some shape and form as we knew it at that time.

But then there is the unexplained reality that a blade of grass will grow between sidewalk cubes. Ahh..David vs Goliath .. Vegas odds makers stunned!

Considering all..We have come a long way in a short time. And yes..there is much to be done..some on this blue dot do not have running water
many are held captive under political tyranny .. and many other horrors.

I see many things..some I do not understand..I breath and realize I do not have too.

[Stevebol]

It seems every 40 years there's an extraordinary event in technology. Atomic weapons at the end of WWII. The rise of digital consumer tech in the mid 80's. Synths, drum machines, midi, 4-track recording.
Now there's AI.
I was wiped out 40 years ago. Doesn't bother me.

I'm from Buffalo and I'm a pathological liar.

Carry on.

[pauln]

Recently I got to thinking about the supposed imminent arrival of Artificial General Intelligence, which would presumably make most of us redundant. I remember hearing when I was a teenager that, given the technology of the day, it would take something the size of the Moon to model all the neurons and connections in a human brain. Even with Moore's law, the advent of massively parallel GPUs, etc, etc, that's still surely a *huge* mass of hardware in today's technology. Moore's law over the last 40 years would be a reduction in mass by a factor of 1e7 (1.5^40), and the mass of the Moon is 7e22kg, so we'd still need 6e15kg of computing power. So how do folks imagine AGI is somehow just around the corner?

Lots of science fiction assumed future AI would be a machine all the way down to the molecular level (like living things are). Right now AI is a machine only down to the lowest level of its design.

Folks realize that money for AGI is right around the corner and are talking it up, but real AGI may only come when the design is down to the molecular level... like "thinking jelly" (cool band name alert) that may be souped out and placed into service as needed.

[Litterick]

Cory Doctorow in the Guardian:

The collapse of the AI bubble is going to be ugly. Seven AI companies currently account for more than a third of the stock market, and they endlessly pass around the same $100bn IOU.

AI is the asbestos in the walls of our technological society, stuffed there with wild abandon by a finance sector and tech monopolists run amok. We will be excavating it for a generation or more.

To pop the bubble, we have to hammer on the forces that created the bubble: the myth that AI can do your job, especially if you get high wages that your boss can claw back; the understanding that growth companies need a succession of ever more outlandish bubbles to stay alive; the fact that workers and the public they serve are on one side of this fight, and bosses and their investors are on the other side.

Because the AI bubble really is very bad news, it’s worth fighting seriously, and a serious fight against AI strikes at its roots: the material factors fueling the hundreds of billions in wasted capital that are being spent to put us all on the breadline and fill all our walls with hi-tech asbestos.

[Litterick]

Why has AI not already been applied to serving humanity in the most highly visible wonderful basic ways? Like making all -
- phone calls, voice mail, email, text messages, etc. spam free
- devices on the internet virus free
- web sites clear of bots
- controlled traffic intersections wait free coordinating lights to existing traffic
- etc...
There's a whole world of immediate quality of life improvements out there; what's the holdup?

Cory Doctorow:

AI is a bubble and bubbles are terrible.

Bubbles transfer the life savings of normal people who are just trying to have a dignified retirement to the wealthiest and most unethical people in our society, and every bubble eventually bursts, taking their savings with it.

....

The bubble doesn’t want cheap useful things. It wants expensive, “disruptive” things: big foundation models that lose billions of dollars every year.

[Litterick]

Cory Doctorow, again:

I'm sorry. As a technology writer, I'm supposed to be telling you that this bet will some day pay off, because one day we will have shoveled so many words into the word-guessing program that it wakes up and learns how to actually do the jobs it is failing spectacularly at today. This is a proposition akin to the idea that if we keep breeding horses to run faster and faster, one of them will give birth to a locomotive. Humans possess intelligence, and machines do not. The difference between a human and a word-guessing program isn't how many words the human knows.

[Mick-7]

Folks realize that money for AGI is right around the corner and are talking it up, but real AGI may only come when the design is down to the molecular level... like "thinking jelly" (cool band name alert) that may be souped out and placed into service as needed.

Biological computing systems exist now, programming lab cultured neurons rather than silicon micro-chips. The big tech giant data center AI model is a technological and environmental dead end.

Mouse and human brain cells in a lab dish learn to play video game Pong : Shots - Health News : NPR

[Christian Miller]

In general, machine learning is very useful. It's incredibly useful for scientists sequencing DNA, looking for interesting things in huge astronomical survey images and so on. That's been happening for years.

But this outgrowth of ML, LLM's give the impression of sentience, and that's dangerous to human mental health. We haven't encountered anything like LLM's before - something that speaks but is not actually self aware. It's a really bizarre thing.

LLM's certainly have their uses. I tend not to use them too much to help me write, but it is useful for redrafting official correspondence. So I'll out a letter through (the free) Chat GPT and then edit again to get rid of the obvious AI phrasing and redundant language it seems to like.

It's quite good for instance at phrasing things tactfully. So I use it as a way of delegating the emotional work of being a social human being, which is endlessly amusing to me.