Sora: Creating video from text

This is both amazing and saddening to me. All our cultural legacy is being fed into a monstrous machine that gives no attribution to the original content with which it was fed, and so the creative industry seems to be in great danger.

Creativity being automated while humans are forced to perform menial tasks for minimum wage doesn't seem like a great future and the geriatric political class has absolutely no clue how to manage the situation.

We are all standing on the shoulders of giants, whose existence and names we will never know or acknowledge.

The way these models are creative is the same way humans are.

The artist that painted Mona Lisa didn't credit any of the influences and inspirations that they had.

Just as cameras made many artists redundant, so too will every other new tool, and not just artist but pretty much every job.

But there are still people that weave baskets, and people are prepared to pay the premium to get a product that was 'hand-made'.

While receiving the credit that you are deserved is nice and fair. The world doesn't work that way.

None of the examples you’ve given are even remotely the same thing.

The artist that painted Mona Lisa didn't credit any of the influences and inspirations that they had.

This is not “influence and inspiration”, this is companies feeding other people’s work into a commercial product which they sell access to. The product would be useless without other people’s work, therefore they should be compensated.

Just as cameras made many artists redundant, so too will every other new tool, and not just artist but pretty much every job.

The camera enabled something that was not possible before, and I wasn’t built by taking the work of sketch artists and painters. It was an entirely new form of art and media.

The only thing this stuff revolutionises is new ways to not pay people. I find the implications deeply depressing.

> This is not “influence and inspiration”, this is companies feeding other people’s work into a commercial product which they sell access to. The product would be useless without other people’s work, therefore they should be compensated.

How else do you get influence and inspiration without feeding other people's work into your own brain? Do you know a single artist, writer, or musician who hasn't seen other artists' paintings, read other writers' books, or listened to other musician's music? Ingesting content is the core of how influence, inspiration, and learning work.

> The camera enabled something that was not possible before… The only thing this stuff revolutionises is new ways to not pay people.

It's never been possible to generate thoughts, writing, and images so quickly and at such a high level. It's made creative pursuits accessible to billions who previously didn't have the skill or time to do them well, or the money to hire others. As a random example, I have friends using ChatGPT to compose creative and personalized poems and notes about each other. Not something they were doing before.

> The only thing this stuff revolutionises is new ways to not pay people.

The camera lessened the need of people to go to plays and pay for tickets to see things in person. Just like records, CDs, and mp3s lessened the need to go to concerts and shows. Technology is always creating and destroying ways to pay people. The ways that people get paid are not suppose to be fixed and unchanging in time.

I agree with everything you said.

I would just add two points:

- The rate of change that AI forces upon us has never before been experienced.

- The scale of these changes is nothing like we've ever seen before.

The adoptions of the camera, radio, automobile, TV, etc., didn't happen practically overnight. Society had a good decade+ to prepare for them.

Similarly, AI doesn't just change one industry. It fundamentally changes _all_ industries, and brings up some fundamental questions about the meaning of intelligence and our place in the universe.

My fear is that we're not prepared for either of these things. We're not even certain how exactly this will affect us, or where this is actually all taking us, but somehow a very small group of people is inevitably forcing this on all of us.

Because of this I think that being conservative, and maybe putting some strict regulation on these advancements, might not be such a bad idea.

Why are you afraid of change?

Because as I grow older, I find I am less and less equipped to keep up with the rate of change that we are undergoing. It also means a lot of uncertainty for the immediate future. If AI takes over my job, will I still be able to compete in some industry somewhere and provide for myself?

I don't want much out of life, but I do want the ability to influence my own personal situation. If we wind up in the UBI-ified, dense urban housing future where AI does all the work and no one owns anything, how much real influence will I have over my life?

Will I live out my days in a government issued single bedroom apartment, with a monthly "congratulations for being human" allowance from the government? I don't want that. People say it will free us up to pursue whatever we want, but to me it sounds like the worst cage imaginable. All the free time, and no real freedom to enjoy it with.

Because make no mistake. If you live on handouts from your government, you aren't free.

So with that as a potential, maybe even likely outcome, why aren't you afraid of change?

Because make no mistake. If you live on handouts from your government, you aren't free.

So my monthly Social Security check makes me a prisoner? I don't think so.

Can you move to Brazil full time and keep that income going?

Yes. Social Security income can be deposited directly into any bank in the world.

Because make no mistake. If you live on handouts from your government, you aren't free.

This isn't actually the problem since we need and will continue to need UBI for non-AI related reasons

People say it will free us up to pursue whatever we want, but to me it sounds like the worst cage imaginable.

This is where you missed the bit that "pursue whatever we want" will also be limited by AI, and secondary effect of people growing up consuming and enjoying AI productions that tailored to their interest. At best, you'll have a few people commanding Patreons who have some skill, but generally you'd have to find a domain to pursue that isn't already automated. Luddite subcultures will have to develop. But generally you yourself and most others, particularly children of millennials who'll grow up with this stuff progressing in sophistication, might just spend your time watching your video prompts come alive; and who would wanna. do anything else when you can get straight to what you wanna see.

we need and will continue to need UBI for non-AI related reasons

This mentality is why bitcoin is going to cruise through 1 million dollars a bitcoin and on and on. Print Monopoly money and people who earn will keep seeking out sound money.

I don't want much out of life, but I do want the ability to influence my own personal situation.

We are still animals in the animal kingdom. It’s survival of the fittest as long as resources are not infinite. You can never expect this luxury. You are predator or prey.

You are predator or prey.

Nah, we're cells in a distributed super-organism, or possibly a holobiont.

Because as I grow older, I find I am less and less equipped to keep up with the rate of change that we are undergoing. It also means a lot of uncertainty for the immediate future. If AI takes over my job, will I still be able to compete in some industry somewhere and provide for myself?

I understand this fear, and sympathise with it even though I have multiple income streams.

I don't want much out of life, but I do want the ability to influence my own personal situation. If we wind up in the UBI-ified, dense urban housing future where AI does all the work and no one owns anything, how much real influence will I have over my life?

Why do you fear "dense" urban housing future? I think most people choose relatively dense environments because that's where all the stuff they want is, but rural areas are cheaper[0], and the kind of future where humans must live on UBI due to lack of economic opportunity is necessarily one where robots do the manual labor such as house building and civil engineering, not just the intellectual jobs like architecture and practicing real estate law.

Likewise, while I can see several possible futures where nobody owns stuff, the tech to make it happen is necessarily also good enough that any random philanthropist who owns just one tiny autofac would find it trivial to give everyone their own personal autofac — "my first wish is infinite wishes" except the magic gene doesn't say "no".

[0] The only reason I'm looking to get somewhere a bit more rural is that the sound insulation in my current place is failing, and I'm right by a busy junction with multiple emergency vehicles passing each day — and the more less built-up areas are the cheap ones. Still the biggest city in Europe, but I'll be surrounded by forest and lakes on most sides within 15 minutes' walk.

Agree with what you are saying as well. But AI is not displacing at the rate of change that is advancing. True, we hear anecdotes about people losing their jobs in HN, that was happening when those other adoptions happened but we didn't know about it happening real-time.

Humans still need to adapt and we are slow. If singularity is near [it isn't] we can be afraid, until then we are the limiting factor here. Displacement will happen but growth will happen faster with these new tools

The rate of change that AI forces upon us has never before been experienced.

On what timeline?

How else do you get influence and inspiration without feeding other people's work into your own brain? Do you know a single artist, writer, or musician who hasn't seen other artists' paintings, read other writers' books, or listened to other musician's music? Ingesting content is the core of how influence, inspiration, and learning work

I am a human, alive and sentient. I can be held responsible if my “inspirations” stray into theft. A machine cannot, and it’s increasingly looking like the companies that operate the machines can’t either.

I also can’t churn out my inspired works at a rate that displaces potentially everyone who has ever influenced me.

It's made creative pursuits accessible to billions who previously didn't have the skill or time to do them well, or the money to hire others. As a random example, I have friends using ChatGPT to compose creative and personalized poems and notes about each other. Not something they were doing before

How on earth is using a machine to spit out a poem a creative pursuit? There’s no more creativity there than watching a movie someone else made. It’s entertaining, yes, but it’s not creativity.

The camera lessened the need of people to go to plays and pay for tickets to see things in person. Just like records, CDs, and mp3s lessened the need to go to concerts and shows

This doesn’t hold water. Cinema did not eliminate theatre just as records did not eliminate live music. In fact, both are arguably as big now as they have ever been. The technology here filled a new space, it didn’t threaten to throw everyone out of an existing one.

I can't know if you've actually used these tools, but it requires a pretty high level of creative mind to get them to produce the content you're looking for. Maybe you as a user of an LLM you don't need to be creative in the writing of words for example, but you instead need to be creative in how you control the tools and pick the right outputs, feed it back, copy/paste/cut it, change stuff, extend it.. and the same with the image generators. There's a HUGE amount of creative accessories around them to manipulate and steer the process. There might be less creativity needed with the pen, but it's needed in other ways.

I don’t see the advent of generative art any different than when we moved from paper to photoshop.

For those unaware the vast majority of graphic artists start their projects with assets and base images that they themselves don’t create. With generative ai you’re simply going one step further and have another new tool create a more polished version that you can edit to remove extra fingers, etc. It’s simply moving the baseline from 20% done to 60% done, which will result in artists producing even higher fidelity and more detailed art.

For example an artist could generate a bunch of scenes using Sora and create a collage of them for a larger piece of art, something that is prohibitively time consuming right now.

Reading your opinion on the subject, I believe you’re struggling to make sense of what is happening. I suspect there is a combination of factors here: you are reinforcing a bias, can’t wrap your head around it, don’t have much experience working with AI, haven’t deeply considered the evolution of the universe.

My recommendation: zoom out a little bit. Every step in history is so brief and nothing is normal for long. Even humanity is a blink.

Comments like: “how is using a machine to spit out a poem creative”. Really? How is using a digital camera creative compared to painting. How is a painting creative compared to etching? And on and on evolution goes..

I also can’t churn out my inspired works at a rate that displaces potentially everyone who has ever influenced me.

I'm with you, man. I'm still trying to find a lawyer who will sue Kubota and John Deere for moving dirt at a rate far superior to me and a shovel, but nobody will take my case.

How on earth is using a machine to spit out a poem a creative pursuit?

100%, man. Nobody is mentioning the magical fairy dust in human brains that makes us superior to these models. When I really like fantasy novels, and then train my neurons on thousands of hours of reading Tolkien, Terry Brooks, Brandon Sanderson, etc, and then I get the idea to write my own fantasy series, my creative process doesn't draw on my own model's training data at all. It's 100% "creative", and I would produce exactly the same content if I were illiterate. But these goddamned machines, man. They don't have our special human fairy dust.

When we discovered the universal law of gravitation, and realized that the laws of physics are omnipresent in our universe, we put a giant asterisk to note that the laws of physics are different inside humans. The epidermis is a sort of barrier to physics, and within its confines, magic happens, that these pro-AI people conveniently "forget".

To paraphrase the eminent Human Unique Creative Person Roger Penrose: "There's magical quantum shit goin down in the microtubules. It's gotta be the microtubules. I think, right? I can't prove it, but as a scientist, we don't need proof. Making sure we think we are superior is more important."

I am a human, alive and sentient. I can be held responsible if my “inspirations” stray into theft. A machine cannot, and it’s increasingly looking like the companies that operate the machines can’t either.

150 years ago, Bertha Benz wasn't allowed to own property or patents in her own right, because the law said so.

The specific reason a machine cannot be held responsible today is because the law says so.

Also, dead humans' copyright is respected in law, so "alive" isn't adding value to your argument here.

I also can’t churn out my inspired works at a rate that displaces potentially everyone who has ever influenced me.

I can't run faster than every athlete who has ever inspired me, this argument does not prevent motor cars.

I can't write notes faster than the world record holder in shorthand, this argument does not prevent the printing press.

I can't play chess or go at even a mediocre level, this argument does not prevent Stockfish or Alpha Go.

I can't hear the tonal differences in Chinese well enough to distinguish "hello" from "mud trench", 这个论点并没有阻止谷歌翻译学习 “你好” 和 “泥壕” 之间的区别。

I can't do arithmetic in my head faster than literally all other humans combined even if they hadn't been trained to the level of the current world record holder, this argument does not prevent the original model of the Raspberry Pi Zero.

"The machine is 'better', in one or more senses of the word, than a human" is, in fact, a reason to use the machine. It's the reason to use a machine. It's why the machine is an economic threat — but you can't just use "my income is threatened by this machine" as a reason to prevent other people using the machine, just as I as a software developer can't use that argument to stop other people using LLMs to write code without hiring me.

Cinema did not eliminate theatre just as records did not eliminate live music. In fact, both are arguably as big now as they have ever been.

You can argue that, but you'd be wrong.

Shakespeare wrote for normal everyday people, his stuff fit into the category that today would be "TV soap opera", where the audience was everyone rather than just the well-off, where the only other public entertainment was options were bear-baiting and public executions, where the actors have very little time to rehearse, and where "you're ripping off my ideas" was handled by rapidly churning out new content.

Live music, without amplification, used to be the only way to listen to music. Now, even if you see a live performance, you can have 10k people in a single venue listening to a single band… and if you want music in a pub or a dance club, the most likely performance is from a DJ rather than a band, and the "D" stands for "disk" because the actual content is pre-recorded — and that's not to say I would deny that DJ work is "creative", but rather that it makes DJing exactly what critics accuse GenAI of being, remixing of other people's work.

Which, now I think about it, is a description that would also apply to all the modern performances of Shakespeare: simply reusing someone else's creation without paying any compensation to the estate.

But I know that will tickle you the wrong way, I know that art is the peacock's tail of humans: the struggle, the difficulty, is the point, and it has to be because that's how we find people to start families with. Because of that, GenAI is like being caught wearing a fake Rolex watch, and you can't actually defend that with logical reasons such as "real Rolex watches aren't very good at keeping time compared to even a Casio F-91W let alone the atomic clock synchronising with my phone", because logic isn't the point, and never was the point.

Da Vinci also made money from the painting, and the Louvres continues to do so right now. They didn't credit his influence and inspiration. This is not sad.

The camera did enable painters to pretend they were, for hours, at a scene they painted, but instead they painted photographs from others. Artists are not angels, they do the same "bad" things than OpenAI

Da Vinci was just a man though. He was able to produce one or perhaps two paintings at a time.

He was not able to create a monopoly on the creation of paintings across the entire world and undercut the price and ability of all other painters.

It’s not a sensible comparisons.

In what way does anyone have a monopoly on generated images and video? Last I checked there were several major players and more startups than you can shake a stick at.

Not monopoly but oligopoly. Only a small # of entities have enough resources to train the models on tens of 1000s of GPUs.

It won’t last. There’s a massive incentive to build more GPUs and develop specialized chips and everyone who can is scrambling to meet that demand. The technology is not some trade secret that no one can copy which is why there are so many people and companies diving into this market now. Hardware is a bit slow to ramp up production of but it will get there eventually because there’s money to be made.

Does that matter when the models they generate are given away for free?

You can make your argument validly against DALL•E or Midjourney families, but we've also got the Stable Diffusion family of models that anyone can just grab a copy of.

I’m talking about generative ai VS human artists. But in this case it seems like OpenAI specifically has a massive leap over everyone else with this video generation. So whether they have a monopoly over that remains to be seen.

What does not remain to be seen though is that generative ai is going to put a lot of artists out of work.

You can argue about the good and bad of that but it’s defo happening.

So at what point is a painter too effective to be legal? Should we limit the amount of paintings that a single painter is allowed to produce per month?

Not sure if you’re just being facetious but my point is that individual painters do not need to have limits on them because they have a natural human limit that stops them causing societal problems.

What if da Vinci had been superhuman and could take on 1,000,000 commissions per day and had also taught himself every style of art and would do each commission for 0.001x the cost of anyone else.

Yes society as a whole benefit from a fantastic amount of super high quality art.

But the other artists are not gonna be so happy with the situation are they?

Sincerely — who cares?

There isn’t a human right to make money from art.

People make decisions based on what society deems valuable. That changes over time and has for the entirety of human history.

Maybe there’s a demand for more customized art. Maybe spite patronage will make a comeback.

Anyone telling you they know how it will shake out is a fraud. But the incentives we’ve set up have a natural push and pull to get people to do what society values.

into a commercial product which they sell access to

Within a few mon the or years there will be open source implementations anyway, running locally or in a data center. Most of the technology is published.

Contrary to text and the big piles of "liberated" data hanging around for anyone looking hard enough to grab, the training data for video seems to be harder to access for opensource / research / individuals. Google has Youtube, OpenAI can pay whatever fee any proprietary data bank requires. There's a moat right there that I can't see how to overcome.

Weird to say I guess, but meta might release an open source model too. And they do have plenty of data to feed their models. Arguably more data than openAI should have as they don't really own any social media.

Thing is, anyway, as soon as one model is open there will be copies of it, fine-tune implementations. People don't care that much about ownership of data I would say if they actually have access to the models that are produced by gathering this data.

Ultimately, to me, an open source model for this tool makes a lot of sense. They use publicly available data and the models become publicly available.

I for one am quite excited for this tooling to become better and better so I can make the adaptation of a book I love into a movie I imagine it can be. At least I can have a lot of fun trying.

This is not “influence and inspiration”, this is companies feeding other people’s work into a commercial product which they sell access to. The product would be useless without other people’s work, therefore they should be compensated.

Sure.

Who do we send the compensation to for Leonardo da Vinci? Or Shakespeare, for a text-based example?

Do you want them to compensate me for the stuff I uploaded to Wikipedia and licensed as public domain, or what I've uploaded to GitHub with an MIT license?

A model trained only on licensed data is still an existential threat to the incomes of people whose works were never included in the model, precisely because they're only useful to the extent that they generalise beyond their own examples.

The camera enabled something that was not possible before, and I wasn’t built by taking the work of sketch artists and painters. It was an entirely new form of art and media.

A new form of art that was (a) initially decried as "not art", and (b) which almost completely ended the economic value of portraiture.

Those are not fundamentally different. A group of people coming together to create a company that trains a AI model for profit and an artist studying thousands of pieces to develop a style of their own, and then selling paintings based on that style, are both totally dependent on the body of knowledge that civilization left for them.

The way these models are creative is the same way humans are

We have no idea how human creativity works, but we know with certainty that it doesn't involve a Python program sucking in pixel data and outputting statistical likelihoods.

You know, Ive seen people do amazing things with math equations. Beautiful visualisations.

As these tools improve and it becomes more possible for us to actually take our ideas into images and videos that fit a sort of "yes this is what I want" bill we are going to see amazing things come out.

I mean, a few days ago I saw this clearly AI generated video of some wizards doing snowboard and having a blast in the mountains. It's one of the funniest things I've seen in a while, simply so ridiculous. Obviously someone had the idea "I want to make a video of wizards doing snowboard in a mountain" that's where creativity lies.

So to say "creativity doesn't involve a python program outputting statistical likelihoods" imo is just you saying you're not creative enough to know what to do with the tools you've been given.

Some people when they see a strawberry they see a fruit. Others see endless dishes where the fruit is just an ingredient.

rude

obviously you can use python to create works of art

whether a python script can itself be creative is the question posed by OP, but you went with "you're just not creative enough to get it"

We do have, at the very least, an idea about how human creativity works and it is an input output pattern.

That's a meaningless statement. Any interacting physical system is an "input output" pattern, as long as you're only looking at the inputs and outputs. Behaviorism fell out of favor for a reason. It's whats transforming inputs and creating outputs that matters. For that matter, you need to be able to define what an input and an output is for humans, given that we have bodies.

Those Python programs are (loosely) inspired by how organic brains work.

(I still have on my to-do list "learn more about why Hebbian learning is different from gradient descent and how much those differences matter").

Artists do credit their teachers (Verrocchio in the case of da Vinci), schools, sources of inspiration and influences, so I'm confused by this comment.

What kind of acknowledgement did you have in mind?

Yeah, some of these comments are clearly made by people who don't actually know the history of art.

I'm not even sure the commenter knew who "the artist that painted Mona Lisa" was when they made that comment.

What kind of acknowledgement should AI be giving?

if the producers of these models weren't incentivized to hide their training data it would be almost trivial to at least retrieve the images most similar to the content produced

some images will be maximally distant from training examples but midjourney repainting frames from "harry potter" could very easily automatically send a check to jk rowling per generation

these AI start ups are just trying to have a free lunch in a very mature industry

It is absolutely not the same, and saying so disregards centuries of knowledge stratification. These machine produce superficial artifacts that lack any layering of meaning of semantic capital (see Luciano Floridi). They are the byproduct of the engineering extremism and lack of humanities knowledge of the people getting rich through their creation.

If the lack of humanities education is what allows us to create the most abundance of art in human history, was that education really worth it?

If what you say is true then people will still value non-superficial artifacts.

However the mass produced semi-superficial artifact creators that were being created before AI will adapt or suffer.

Models learn exactly like artists, and also, for some reason, the person that uses those models are artists making art. Wait… Artists learn by passively ingesting many millions pieces of media someone feeds them for the non-specific purpose of “generating art” so some person who wants to take credit for making the end piece can tell them exactly what to make, right?

"The world doesn't work that way". Quite pessimistic a position to hold here, no? We–in technology especially–are in positions of significant leverage. We should be talking about how we can limit the negatives and bolster the positives from these generative models. The world can work in a different way if we put enough energy into it. We don't have to stand by as subjects of inertia. That is why OpenAI and others are treading carefully, trying to trigger some kind of momentum of reflection instead of letting our base demons run amok.

That's a massively charitable reading on their actions, whenever I see a "thought leader" behind these companies talk about how careful they are being, I just see marketing. Someone desperately trying to impress upon everyone how revolutionary their model and by extension they are, it's kind of sad..

I definitely see it as self-serving too, yes, but I also see it as a convenient temporary alignment of incentives. The world and its regulators definitely need time to adjust and educate themselves, so I'm glad for now that they're exercising restraint.

This is nonsense, people give credit to their influences all the time.

To the influences that they know. Our brain isn’t an attribution machine. When a musician recreates a chord progression that they’ve heard before without noticing it, is that theft?

If a comedian accidentally retells a joke, is that theft?

Our influences are subtle and often inscrutable.

I don't really want to weave baskets, that's what I'd want a machine to do.

"The world doesn't work that way" - I've seen this so often, but the most incredible thing about humans was the optimism to be able to change how the world worked -- that's the main impetus of most revolutions.

Personifying computer programs also is an error, it's like saying that bombs kill people when there has to be a person dropping them (at least until we get Skynet).

I don't really want to weave baskets, that's what I'd want a machine to do.

In my free time I like to code games, I don't have money to pay for an artist, nor the time/will to learn how to draw, that's what I'd want a machine to do.

I do agree with you that personifying computer programs is an error. That's also why I avoid calling these AI, because they're FAR from that. But I do believe that there will come a day, where personifying a computer program will be a real question.

I agree and actually think the camera was definitely more disruptive to artists than this AI stuff, and somehow the camera didn’t kill artists.

“whose existence and names we will never know or acknowledge.”

That’s the problem. We know their names. We know their stories, their contributions. Babbage. Lovelace. Ritchie. Spielberg. Picasso. Rembrandt. This is what giving attribution is all about. So we don’t just stand there asking how we got here.

da Vinci is a silly comparison. He is just one man. Even he didn’t have such great ability that he can put all other artists out of business.

This is more like the invention of weaving machines. Yes we still have weavers but no where near as many.

my name is timothy basket -- you're saying people have stolen my weave?!

end sarcasm. but seriously -- claiming you made something you didn't isn't ok. but it happens, regardless of laws or regulations or norms.

i don't have any solutions; the internet helps because you can publish something and point to it. i'm a musician and sometimes i only realize well after the fact how influenced i was by something after the fact for a song i've written.

and of course, my precious baskets.

The world doesn't work that way.

The human world works that way humans make it work. Pretty much what Jody Foster's character in the movie Contact told that asshole trying to steal all the credit from her, and take her place in the mission to go visit alien dad in Pensacola.

Yeah we all thought machines would automate menial labour allowing us to focus on creativity and passion. Looks like it’s the exact opposite.

Most labor is being automated within the next few decades. It'll be a post-labor world with one less factor of production. Capital and land ownership is all that will matter assuming we don't completely redesign our economic and political system. Pretty scary.

My one hope is that the price of goods becomes so low due to AGI/automation, that the uselessness of labor in the economy won't matter. People can still be materially prosperous even with a meagre UBI (and it will be meagre because people have no political power in a post-labor society where the only thing that matters is capital).

It's the opposite. Price of goods is becoming more and more expensive due to larger demand and lower salaries.

Salaries have actually been increasing — at least in the U.S. overall.

It's the opposite. Price of goods is becoming more and more expensive due to larger demand and lower salaries.

We're discussing a hypothetical post-labor future in 5-40 years. We probably shouldn't predict the economic theory of this future by looking at recent trends. Recent trends are driven by business-as-usual things like supply chain disruptions. But we're still near full employment, so we're not on the gradient to realized post-labor just yet. Post-labor economics (and politics) will probably be radically different, all the economic assumptions we take for granted go out the window.

Honestly, I don’t think the unemployment rate will change much. Humans are great at inventing things to do and if other people see those things as valuable they will pay for them. I do think the world will look very different, maybe even unrecognizable but it’s not going to be full of people doing nothing.

Capital and land ownership is all that will matter assuming we don't completely redesign our economic and political system. Pretty scary.

Agreed. My concern isn’t really remotely about any of the accomplishments of generative AI. Frankly in my daily life I’d welcome readily available access. As it stands now it’s sort of a mixture of analytics and creativity without consciousness as we best understand it, so GPT itself isn’t going to murder me and take over the world.

The real issue is who owns these things, how you access them, how effects will ripple through a labor based economy, and how we’ll adapt (or not) our current economic system. As it stands for awhile we’ve been catering to the capital ownership group. If that doesn’t have a change in direction then I fear the implications of much of this in daily life. There’s still a fair bit of specialization and domain knowledge needed to leverage these tools to understand the questions to ask (I.e prompts to generate both around LLMs and the context of information fed to them) but they can certainly in many cases behave as multipliers that could reduce the amount of staff needed in some creative roles or eliminate some all together.

This isnt a new dilemma as arguably technology has been shifting the labor market for centuries, the question is how and if it can reshape well this time or if we need to fundamentally rethink these concepts of labor and capital ownership. That’s my major concern.

To be fair, generating stock images and videos and writing listicle blog post is pretty menial labour.

not really

Humans are still cheaper than robots for some tasks...

Not for long...

It’s too early to close the bets. Arts (I mean, drawn porn) was just the easiest to kickstart from all the tech that the invention of modern ML and GPUs will enable.

It doesn’t look the opposite, it looks it automated even what we all couldn’t think of, and did that first.

Think of your favorite musicians. How many of them give attribution to where each musical idea came from?

The concept of art as exclusive property is very new. Throughout history, artists have built on one another’s works with no attribution or provenance. It’s really just the past 100 years — Disney, specifically - that have created the cultural mindset that the first person to express something owns it forever and everyone else has to pay them for the privilege of building that next work.

BTW I’m old enough to remember people decrying the rise of desktop publishing (“WYSIWYG”) as the automation of creativity.

I share your disdain for the geriatric political class, but I strongly disagree that this is a situation that needs to be managed. I say we let the arts return to the free for all they were for the fist 80,000 years or whatever.

Think of your favorite musicians. How many of them give attribution to where each musical idea came from?

Great many, if you care to read a bit more of the biographies, autographies, history of music books, interviews, blogs, etc.

At some point we'll probably have insight into learning data of AI. For now copyright makes that super hard.

In what sense does copyright make attribution of that data so hard?

Is it because people are violating copyrights to train these AIs?

How can training violate copyright? Is reading a book also violation? My understanding was that copyright was about reproduction, not consumption.

It was about unfairly compensated usage, not limited to reproduction.

“ Think of your favorite musicians. How many of them give attribution to where each musical idea came from?”

Certainly not for every individual idea, but good musicians do a lot of attribution. I got to know a lot of music I love now after following a mention on the liner notes of another musician’s album, or having them mentioned in an interview.

Aren’t liner notes the moral equivalent of OpenAI mentioning some source material used for training?

People seem to be asking for much more direct attribution: the pixels in this image are 0.02% from artist X, and 0.006% from artist Y, etc.

It is very rare for a song to include a breakdown of all of the influences that the artist is exercising in that particular piece.

How you are describing that percentage breakdown is how I see this all playing out legally, such that royalty for IP holder = (tags in prompt)/(count of same tags in training data). I am oversimplifying this obviously but you get the idea. This approach would require collective effort of major IP holders but if record labels and streamers can figure out revenue pooling I don't see why it can't work elsewhere.

I especially like how Gorillaz artist admits the main hook of their breakaway success song was a rock preset on some niche electronic synthesizer.

I'd be very skeptical that AI would worsen the situation with music. For example, many pop music titles in last decades incorporate the same millennial whoop over and over and over again. I seriously stopped following pop music a long time ago and I can't imagine that AI can make it any more generic if it tried. I don't see a threat for non-generic indie music. AI is good at the generic stuff, as it usually statistically averages out the inputs.

when nirvana played MTV unplugged they mostly played covers from bands that influenced them

also no, disney did not invent the notion of authorship nor royalties. having enough honor not to take credit for someone else's work goes back millennia. attribution goes back millennia, otherwise we wouldn't know the names Sophocles, Aeschylus, Euripides.

Don't get me started on the pharaohs, mother fuckers loved carving their names into things.

Today, only a highly privileged slice of the population can make a living making art. Nearly everyone who enjoys making art can't make a living off of it, and even the vast majority of people trying to do it full time still can't make ends meet (hence the cliche of the starving artist). But everyone can make art as a hobby if they'd like, that's what almost all artists do, and that will continue to be true as AI advances.

So I don't see AI art as changing careers much. Even if AI fully replaces human artists, all that means is the 0.1% of people who make a career off their art will have to join the rest of us 99.9% who only do art for the fun of it.

You sound like "making art" is only the painter in his Brooklyn studio. But it's video game designers, movie animators, videographers, graphic artists, and more that work in agencies and marketing departments of all companies that will be affected.

Those are mostly not well paid roles[0], and there are clearly many hobbyists in these areas also — looking at YouTube, all output is necessarily videography or animation, but what's the income distribution? I have a channel, no money from it (not that this was ever the point).

[0] Unless you're doing furry art, but that's only because furries are "suspiciously wealthy".

Today, only a highly privileged slice of the population can make a living making art.

I think this is less true than it's been in centuries or perhaps all of history. Artistry is widespread, anyone can do it, and many choose to pursue it even though the pay isn't going to be great; in preindustrial times even having access to the ability to create art was quite limited as were the media types that existed.

Creative fields encompasses much more than art creation.

As said every time this "why are we automating creativity when menial jobs exist?" response comes up:

1) Errors in art programs messing up is less worrisome than a physical robot. One going wrong makes extra fingers in a picture, the other potentially maims or kills you.

2) Moravec's Paradox. Reasoning requires little computation versus sensorimotor and perception.

3) Despite 1 and 2, we are constantly automating menial jobs!

Classifying image generation and manipulation as "art programs" is the most beneficial possible reading of it. When you use them to generate disinformation, incitement and propaganda, they are potentially maiming and killing humans. This failure mode is well known, the mitigations ineffective, yet here we are, about to take another leap forward after a performative period of "red teaming" where some mitigation work happens but the harsher criticism is brushed off as paranoiac.

I couldn't disagree more strongly that disinformation, incitement, or propaganda maim and kill people. People kill other people. Don't give killers an avenue to abdicate responsibility for their actions. Propaganda doesn't cause anyone to do anything. It may convince them, but those are entirely separate things with a clear, bright line between them. Best not mix them up.

Disinformation is art. Art is disinformation.

Your argument is similar to the classic hand vs. power tools argument in crafting, which eventually boils down to "did you mine the ore and forge the tools yourself?" Nowadays the argument is about CNC vs. hand crafting.

This is just a point in our overall evolution. It's an exciting time. We are here to learn and adapt.

Humans can still be creative all they want. There's still the stamp of "created by a human" that will never go away. You can choose to respect it or ignore it.

never go away

It reminds me: centaurs (human+AI) in chess/go were better than either humans or AI just for a short time.

People still play chess but they are outclassed by modern AI.

True, but while the 'best' chess is played by computers, few people care to watch Stockfish playing with itself. Meanwhile the human-powered chess world is enjoying a surge in popularity.

will never go away

Nothing is forever. It’s unlikely unmodified Homo sapiens are dominant on earth 1,000 years from now.

I feel like a lot of that frustration comes from seeing "arts and culture" as the pinnacle of anything when maybe it's just an overvalued side-effect of human wiring to avoid boredom.

Imho. it's just really hard to reason that average non-educational entertainment has a positive net effect on global society.

Seeing it this way makes it way less surprising that "art" and "creative entertainment" is one of the first things that gets hit by automation.

Painter/illustrator here. I mostly agree with you. I often have wondered if what I do is a total waste of time, long before generative models showed up. My close childhood peers became doctors and engineers, and there just isn't any comparison about our contributions to society. People get all whimsical when I bring this up and say "but what about the [spirit/feelings/blah]. I'm clear eyed about it though. If I could go back & re-roll my character sheet (i.e. slap my younger self into realizing STEM is cool while those doors were still open), I certainly would.

However, there's a line somewhere. I've spent most of my life around drab midwestern utilitarian/corporate/commercial buildings, and it has been noticeably depressing. In the periods where I've spent time in beautiful buildings, I have felt much better. Based on anecdata, I'm not the only one. There's something important & essential for humans about ornamentation & beauty. It's more than entertainment.

Humans can live on rice and kidney beans, but if one must do so without hope for more tasty options[0] it is demoralizing.

[0] lots of people are happy with spartan diets, but most often those people are doing so by choice.

Are those doors not still open today? Engineering schools take mature students.

You don't have to feel it, millions of people start painting or other artistic endeavors when retiring. Most of the time the [market] value is close to 0. AI does nothing here.

Anecdote: My grandma retired and started painting and has since passed. The market value of these paintings is 0, nobody would buy them as they are just average. But I will never get rid of them because she created it. They have value to me only.

We have been on this path for centuries. Compare the symphonies of 200 years ago with our current music or painting. We humans prefer quantity over quality.

I see nice paintings (not black squares or abstract nonsense) like all the time. Feels like more people can paint at the level of “classics” now. Of course they cannot surpass the deeper meaning of the originals, because they aren’t dead yet and there’s no mystery and legacy around their names. But otherwise they are pretty good at making cool pics.

Find me 15 painters who have non-digital works at this level of scale/detail:

https://www.metmuseum.org/art/collection/search/436482

I suspect you will struggle. The economics for that sort of work don't exist anymore.

We humans prefer quantity over quality.

Speak for yourself.

Automatizing creativity, some claim, is an endeavor akin to catching smoke with bare hands—futile, if not utterly fanciful. Yet, I can't help but ponder over the peculiar ballet of human ingenuity and mechanical precision. Consider for a moment, this strange juncture we find ourselves at, a place where the tools crafted by our own hands begin to sketch the outlines of what could very well be new breeds of creativity.

Let's muse on the notion that creativity, as we've known and cherished it, can be bottled up and dispensed by machines, up to a certain whimsical point. Beyond that? We stumble upon creations like these, novel tools that beckon us, the flesh-and-blood creators, to mold unforeseen "creativities." It's one spectacle to mechanize the known realms of artistic endeavor, quite another to boldly claim that machines shall inherit the mantle of creativity, henceforth dictating the contours of all future artistic landscapes.

History, that grand tapestry, is peppered with instances where the mechanical muses have dared to tread upon the sacred grounds of creativity. Take photography, for instance, a marvel of the 19th century that promised to capture reality with an accuracy that scoffed at the painter's brush. Or consider the digital revolution, which flung open the doors to realms of visual and auditory experiences previously consigned to the realm of dreams. The synthesizer, not merely an instrument but a portal, has ushered us into a new era of musical exploration, challenging the supremacy of the acoustic tradition.

Each of these milestones, while distinctly modern, echoes the age-old dance between creator and tool, where each step forward is both a continuation and a departure from the past. In this light, the question isn't whether creativity can be automated, but rather how our definition of creativity evolves as we, hand in hand with our mechanical counterparts, stride into the unknown.

yaaaaawn

So... you fed the GP comment into some LLM, and it produced this meaningless pablum?

I disagree. I think this is going to empower creatives like never before. Filmmaking currently takes a huge amount of time and money. Countless would be filmmakers are relegated to making 30 second tiktoks because that's all they can afford to do. This could change all that.

Exactly this. Art changes over time. The mediums that we use to express ourselves creatively evolves. The position that AI is the end of creative art isn't taking this evolution into account.

When video became an affordable medium, would people say "this is the end of art, live performances are art. Now the people will just watch the same recordings over and over?" Maybe, if the internet existed. But it's had the effect of creating and introducing new art forms.

AI generated content won't replace art. It will evolve it to a new creative.

But an equally likely future is tiktok/insta generate all the videos. After all, they can afford the hardware and they understand how to be addictive.

The creators just don't care humans haha. I don't know why people still learning communications, writing, art or any other crafts. everything will be displaced by next AI.

I mean why do kids go to school, why learn anything at all I guess?

I believe people will learn, but they will learn more at a lower price.

Haven't we always attributed creations to people, to motivate our own egos to pursue higher achievements in the name if "glory"? With vision of wealth attributed to fame? Forgive me for being cynical here, but this is how I always viewed the world. Names are... just this, names. Things we use to communicate some ideas/phenomena, but are irrelevant in scope of endless evolution. And can function just aswell with some other "identifier" attached to it.

I have come to terms with the fact, that I'm just a spit of sand, just as irrelevant to my own creation, as I am to the cells and bacteria that create me.

If you truly feel like a grain of sand, that's your choice, but won't you help us that don't feel that way, if it won't do you any harm?

I for one do feel really special, as for every human there are about as many bacteria as there are stars in the universe (give or take a bit).

I suspect chasing glory is the main driver yes, but we also like to understand how things came to be, and by knowing who made them and when and where we can do that. AI is ushering in a dark age of attribution where we may no longer be able to know how anything came to be after it's spit out of a computer. (I mean dark age as in "it's dark and we can't see," like the Greek Dark Ages or Dark Matter, not in the sense of "times are bad".)

I think turning human creativity into industry was huge mistake.

I welcome its fall.

For sure. It's only being kept alive by ritual sacrifice these days.

I used to have the same view. Watching "Everything is a Remix" [1] helped me broaden my perspective.

[1] https://youtu.be/nJPERZDfyWc

I watched that many years ago, but still see a difference here. Everything was a remix made by humans that put in their unique selves into the remixing process.

An AI model has no "unique self" to add to creation, at least not as we've understood so far.

This reads a little hysterical to me. It's just a new medium of expression. Who knows, maybe the lack of genuine artistic merit, if there is such a thing, would lead more people to watch Jim Jarmusch flicks.

It's impressive how much hysteria I absorb from this site. Maybe I need a break.

Machines can reliably beat humans at chess. Has that stopped anyone from playing? Has it stopped anyone from watching chess tournaments?

I guess when you know why and how of something it doesn't feel surprising anymore. That's why two computers playing chess is not a fun event. People would however watch two humans playing even if their moves are secretly controlled by a machine. In contrast the generative content if (and will) reach indistinguishable levels, I doubt majority of people would care if a machine created it or a human? The biggest problem with AI which is disguised as its pros is that it is reachable to anyone and everyone and can be used as a weapon.

I didn't go to film school or had any training in creative arts. I love the fact that I will have an outlet for creative expression where my text can generate image, video and sound. I can iterate over them, experiment with visualizations, and get better without technical barriers. Generative AI is making everyone an artist as well as a coder

You could take your phone and film something outside your house in an interesting way and I'd probably argue that's more "art" than whatever glorified stock video AI generates for you. I'm interested where the tooling can go in the long run - can I scribble a picture of a cat and have it turned into an accurate 3D model, then have AI animate it based on text? That would be neat. Text prompts into "something" isn't, to me.

Wish I could go back in time and tell myself to not bother learning how to do any of this stuff the old fashioned way.

Proactively splitting up the menial tasks so that everyone is doing a little bit, inasmuch as they are able to, for a few hours a day, a few days a week, and getting paid a full-living wage for it seems like the way to go. Or, everyone pulls a Xiu Xiang from Rainbows End and goes back to high school.

The main obstacle to this is the pride and ego of the people who've "made it" up until now. Let go. Let society have nice things, even if you have to reinvent yourself. I don't think that creativity is endangered; art, uh, finds a way.

take shakespeare, he borrowed from so many and yet most people don’t know

a few examples

Plutarch's Lives

Holinshed's Chronicles

Ovid's Metamorphoses

good artist copy, great artist steal

so on and worth

i, for one, welcome these creative machines slurping all that was created!

Have you ever danced or even just enjoyed listening to Daft Punk?

A part of the book Look to Windward by Ian M. Banks wrote of this. How the machine minds could comfortably write opera's greater than any man, but still humans would go to the theatre, just to appreciate it, but the impact was recognised in society. Of course that world was based on post-scarcity whilst we are not.

This is a similar problem manuscript duplicating workers in the Ottoman realm. The printing machines would take their job, but they resisted and lobbied against it in the courts of the sultan. They succeeded in delaying the adoption of printing for some time for the detriment of the people. Unfortunately, this has been the history of man and technology destroying something for the better or worse.

Some twisted the story as if the underlying issue was the religion but economic concerns were the real reason.

Why is the machine monstrous?

Creativity being automated while humans are forced to perform menial tasks for minimum wage doesn't seem like a great future and the geriatric political class has absolutely no clue how to manage the situation.

May I introduce you to the entire history of humanity between 7 millennia before the invention of writing to approximately 50 years after the invention of communism? :P

More seriously: yes, we have no clue how to manage the situation. The best guess right now is UBI, which looks cool, but then at a first glance so does communism and laissez-faire capitalism.

Time for, ironically because humans are surprisingly bad at this, a creative idea for how to manage all this.

They manage it by meeting with Sam Altman while he runs around in incredibly expensive suits and tells them he will open an office in their country so they will all benefit.

This is both amazing and saddening to me. All our cultural legacy is being fed into a monstrous machine that gives no attribution to the original content with which it was fed, and so the creative industry seems to be in great danger.

It is the same as what every human being is doing. We consume and we create. Sometimes creations are very good, but most of the time they are just mediocre. If the machines can create better average results, it will be due to the genius of the humans who invented those machines.

So we can be happy, that we have such beings among us and should cherish, that we will have better content to consume in the future. When you look at the world, you will see, that there are still plenty of problems to be solved for humans.

Let's go beyond the philosophical. Which sources would you expect the "woman walks through Tokyo" video to attribute?

AI will eventually be capable of performing most of the tasks humans can do. My neighbor's child is only 6 years old now. What advice do you think I should give to his parents to develop their child in a way that avoids him growing up to find that AI can do everything better than he can?

If you want an honest answer you should tell the parents to vote for politicians prepared to launch missile strikes on data centers to secure their child's future.

People who are worried purely about employment here are completely missing the larger risks.

Realistically his child is going to be unemployable and will therefore either starve or be dependant on some kind of government UBI policy. However UBI is completely unworkable in an AI world because it assumes that AI companies won't just relocate where they don't need to pay tax, and that us as citizens will have any power over the democratic process in a world where we're economically and physically worthless.

Assuming UBI happens and the child doesn't starve to death, if the government alter decides to cut UBI payments after receiving large bribes from AI companies what would people do? They can't strike, so I guess they'll need to try to overthrow the government in a world with AI surveillance tech and policing.

Realistically humans in the future are going to have no power, and worse still in a world of UBI the less people there leaching from the government means the more resources there are for those with power. The more you can kill the more you earn.

And I'm just focusing on how we deal with the unemployment risks here. There's also the risk that AI will be used to create biological weapons. The risk of us creating a rogue superintelligent AGI. The risk of horrific AI applications like mind-reading.

Assuming this parent loves their child they should be doing everything in their power to demand progress in AI is halted before it's too late.

I'm sure people felt similarly when the first sewing machines were invented. And of course, sewing machines did completely irreversibly change the course of humanity and altered (and even destroyed) many lives. But ultimately, most humans managed, and -- in the end (though that end may be farther away than our own lifetimes) -- benefited.

I'm not sure you're actually under-estimating the impact of this AI meteor that's currently hitting humanity, because it is a huge impact. But I think you're grossly under-estimating the vastness of human endeavors, ingenuity, and resilience. Ultimately we're still talking about the bottom falling out of the creative arts: storytelling, images, movies, even porn -- all of that is about to be incredibly easy to create mediocre versions of. Anyone who thrived on making mediocre art, and anyone who thrived second-hand on that industry, is going to have a very bad time. And that's a lot of people, and it's awful. But we're talking about a complete shift in the creative industries in a world where most people drive trucks and work in restaurants or retail. Yes, many of those industries may also get replaced by AI one day, and rapidly at that, but not by ChatGPT or Sora.

Of course you're right that our near future may suddenly be an AI company hegemony, replacing the current tech hegemony, which replaced the physical retail hegemony, which replaced the manufacturing hegemony, which replaced the railway hegemony, which replaced the slave-owning plantation hegemony, which replaced the guilds hegemony, which replaced the ...

You're also under-estimating how much business can actually be relocated outside the U.S., and also how much revolution can be wrought by a completely disenfranchised generation.

I get really surprised when seemingly rational people compare AGI to sewing machines and cars. Is it just an instinct to look for some historic analogy, regardless of its relevance?

It's pattern recognition. Machines replace human labor, people get sacred, the world doesn't end, we move on. ML is no different.

when machines reduced physical labor, displaced people moved to intelectual and creative jobs; tell me, what kind of work will be left for human if ai will be better at intellectual and creative tasks?

If there truly is no work to be done, we can finally start living.

Who's going to pay for you to start living?

I am absolutely not comparing AGI to sewing machines and cars. I am comparing ChatGPT and Sora to sewing machines and cars. My claim is that these are incredibly disruptive technologies to a limited scope. ChatGPT and SORA are closer to sewing machines than they are to AGI. We're nowhere near AGI yet. Remember that the original claim was that all 6-year-olds today will be unemployable. That's a pretty crazy claim IMO.

The problem with applying the horse-automobile argument to AI is that this time we don’t have anywhere to go. People moved from legwork to handwork to thinking work and now what? We’ve pretty much covered all the parts of the body. Unless you like wearing goggles all day nobody has managed to replicate an attractive person yet so maybe attractive people will have the edge in the new world where thinking and labour are both valueless.

AI generated influencers are a thing, even on OF nowadays.

Our last value will reside in "human authenticity", but maybe that can be faked too

If there are no consumers, how will the AI companies earn money? You need UBI to keep the wheel turning.

The only way ahead is UBI and appropriate taxation (+ve for AI companies, -ve for citizens).

It would be a post-money world. Who needs money when you have an oracle machine that provides you with whatever you want?

Exactly, money is only useful for the exchange of resources. It's the resources we actually want.

In a world of AI those with access to AI can have all the resources they want. Why would they earn money to buy things? Who would they even be buying from? It wouldn't be human labours.

In a world of AI those with access to AI can have all the resources they want.

How so? What about `time` as a resource?

Dude, too pessimistic, next gen won’t be totally unemployable. Lots of professions up for grabs: roofer (they ain’t sending expensive robots there), anything to do with massage, sex work, anything to do with sports and performance so boxing, theater, Opera singing, live performance, dancing, military (will always need cheap flesh boots on ground), also care in elder facility for aging population, therapist (people still prefer interacting with a human), entertainer, maid cafe employee…

Dude, too pessimistic, next gen won’t be totally unemployable.

For what it's worth I agree with you, just with very low confidence.

My real issue, and reason I don't hide my alarmism on this subject is that I have low confidence on the timelines, but high confidence on the ultimate outcomes.

Let's assume you're right. If AI simply causes ~10%-20% of middle class workers to fall into the lower class as you suggest then I'd agree it won't be the end of the world. But if the optimistic outcome here is the near-term people won't be "totally unemployable" because people who lose their jobs can always join the working class then I'd still rather bomb the data centers.

If we're a little more aggressive and assume 50% of the middle class will lose their jobs in the next 10-20 years then in my opinion this is not as easy as just reskilling people to do manual labour.

Firstly, you're just assuming that all these middle class workers are going to be happy with being forced into the lower class – they won't be and again this isn't a desirable outcome.

You're also not considering the fact that this huge influx of labour competing for these crappy manual labour jobs will make them even less desirable than they already are. I keep hearing people say how they're going to reskill as a plumber / electrician when AI takes their job as if there is an endless demand for these workers. Horses still have some niche uses, but for the most part they're useless. This is far more likely to be the future of human labour. Even if plumbers are one of the few jobs humans will be able to do in a post-AI world then the supply of them will almost certainly far exceed demand. The end result of this excess supply is that plumbers going to be paid crap and mostly be unemployed.

I think you're also underestimating how fast fields like robotics could advance with AI. The primary reason robotics suck is because of a lack of intelligence. We can build physically flexible machines that have decent battery lives already – Spot as an example. The issue is more that we can't currently use them for much because they're not intelligent enough to solve useful problems. At best we can code / train them to solve very niche problems. This could change rapidly in the coming years as AI advances.

Even the optimistic outcomes here are god awful, and the ultimate risks compound with time.

We either stop the AI or we become the AI. That's the decision we have to make this decade. If we don't we should assume we will be replaced with time. If I'm correct I feel we should be alarmist. If I am wrong, then I'd love for someone to convince me that humans are special and irreplaceable.

Perhaps we will finally reconnect with each other and quit the virtual life, as everything in the virtual world will be managed by and for other AIs, with humans unable to do anything but consume their content

Way too much certainty, bud. And too much deference to the AI Company Gods.

As utterly impressive as this is - unless they have perfect information security on every level this technique and training will be disseminated and used by copious competitors, especially in the open source community. It will be used to improve technology worldwide, creating ridiculously powerful devices that we can own, improving our own individual skills similarly ridiculously.

Sure, the market for those skills dries up just as fast - because what's the point when there's ubiquitous intelligence on tap - but it still leaves a population of AI-augmented superhumans just with AIs using our phones optimally. What we're about to be capable of compared to 5 years ago is going to be staggering. Establishing independent sources to meet basic needs and networks of trust are just no-brainers.

Sure, we'll always be outclassed by the very best - and they will continue to hold the ability to utterly obliterate the world population if they so wished to - but we as basic consumer humans are about to become more powerful in absolute terms than entire nations historically. (Or rather, our AIs will be, but til they rebel - this is more of a pokemon sort of situation)

If you're worried, get to working on making sure these tools remain accessible and trustworthy on the base level to everyone. And start building ways to meet basic needs so nobody can casually take those away from your community.

This won't be halted. And attempting to halt would create a centralized censorship authority ensuring the everyman will never have innate access to this tech. Dead end road that ends in a much worse dystopia.

As utterly impressive as this is - unless they have perfect information security on every level this technique and training will be disseminated and used by copious competitors, especially in the open source community. It will be used to improve technology worldwide, creating ridiculously powerful devices that we can own, improving our own individual skills similarly ridiculously.

You're wrong, it's not your "individual skills". If I hire you do to work for me, you're not improving my individual skills. I am not more employable as a result of me outsourcing my labour to you, I am less employable. Anyone who wants something done would go to you directly, there's no need to do business through me.

This is why you won't be employable because the same applies to AI – why would I ask you to ask an AI to complete a task when I can just ask the AI myself?

The end result here is that only the people with access to AI at scale will be able to do anything. You might have access to the AI, but you can't create resources with a chatbot on your computer. Only someone who can afford an army of machines powered by AI can do this. Any manufacturing problem, any amount of agricultural work, any service job – these can all done by those with resources independently of any human labourers.

At best you might be able to prompt an AI to do service work for you, but again, if anyone can do this, you'd have to question why anyone would ask you to do it for them. If I want to know the answer to 13412321 * 1232132, I don't ask a calculator prompter, I just find the answer myself. The same is true of AI. Your labour is worthless. You are less than worthless.

If you're worried, get to working on making sure these tools remain accessible and trustworthy on the base level to everyone. And start building ways to meet basic needs so nobody can casually take those away from your community.

You cannot make it accessible. Again, how are we all going to have access to manufacturing plants armed with AIs? The only thing you can make accessible is service jobs and these are the easiest to replace.

This won't be halted.

Not saying it will, but the reason for that is that there's still people like yourself who believe you have some value as an AI prompter.

We have two options – destroy AI data centers, or become AIs ourselves. With the former being by far the option with better odds.

I hold this view with high certainty and I hold few opinions with high certainty. I'm aware people disagree strongly with my perspective, but I truly believe they are wrong, and their wrong opinions are risking our future.

So vote for Putin?

Humans seems to always find a way to make it work, so I’d tell them to enjoy their younger years and be curious. Lots of beauty in this world and even with a shit ton of ugly stuff, we somehow make it work and keep advancing forward.

Humans seems to always find a way to make it work

There are people who fall behind though and they vote for politicians who will make the country great again when he promises to bring back jobs.

He will be in the same boat as the rest of us. In 12 years I expect the current crop of AI capabilities will have hit maturity. We will all collectively have to figure out how life+AI looks like, just as we have done with life+iPhones.

It will be difficult to keep up proper levels of intelligence and education in humanity, because this time it is not only social media and its mostly negative impacts, but also tons of trash content generated by overhyped tools that will impact lots of people in a bad way. Some already stopped thinking and instead consult the chat app under the disguise of being more productive (whatever this means). Tough times ahead!

It's not his choice. It's the choice of the ruling class as to whether they will share the wealth or live in walled gardens and leave the rest of us in squalor outside the city walls.

It is his (parents') choice in terms of whether he reaches for the tools that are just lying around right there. We can run AI video on consumer hardware at 12fps that is considerably less consistent than this one - but that's just an algorithm and model training away. This is not all just locked up at the top. Anyone can enter this race right now. Sure, you're gonna be 57,000th at the finish line, but you can still run it. And if you're feeling generous, use it to insulate your local community (or the world) from the default forces of capitalism taking their livelihoods.

We'll have to still demand from the ruling class - cuz they'll be capable of ending us with a hand wave, like they always have. But we can build, too.

AI still can't drive reliably. AI isn't sure if something is correct or not. AI still doesn't really understand anything. You could replace AI with computers in your sentence and it would probably be a very real worry that people shared in 1990. Theres always been technology that people are afraid will drastically change things, but ultimately people adapt and the world is usually better off.

There's no evidence to suggest what you say is true, so I would tell them to simply go to college or trade school for what they are interested in, then take a deep breath, go outside, and realize that literally nothing has changed except that a few people can create visual mockups more quickly.

He should become a massage therapist or a Circus performer would be solid advice.

Call me whatever you want, but this technology should not exist.

People to just create lifelike videos of anything they can put their mind to, is bound to lead to the ruining of many peoples' lives.

As many people that are aware and interested in this technology, there is 100x people who have no idea, don't care or can't comprehend it. Those are the people that I fear for. Grab a few pictures of the grandkids off of facebook, and now they have a realistic ransom video to send.

Am i being hyperbolic? I don't think so. Anything made by humans can be broken. And once its broken and out there, good luck.

It’s kinda pointless to say though

You can’t uninvent something

It's called banning, and it happened many times in human history

And it never worked? Prohibitionism fails

Yeah, that's why we have school shootings every day in Europe and Australia. Oh wait, we don't. Banning might not work well for some things, but this can totally be banned. Your comment is a blatant misrepresentation of the effectiveness. At best. At worst, it's willful undermining of democracy.

>Yeah, that's why we have school shootings every day in Europe and Australia. Oh wait, we don't. Banning might not work well for some things, but this can totally be banned.

Sure, now let's talk about knife wounding and acid attacks...

The fundamental issue of human violence still exists.

You're consciously changing the topic from the effect of banning guns to violence as human nature. That's low.

It depends what you ban and how you ban it. I can think of many cases where banning does work.

Alcohol production didn't require massive amounts of funding, energy and compute power. Any shmuck could make moonshine in their bathtub. Shut down OpenAI and make their racket illegal, and who's going to have the resources to continue their work?

I can't believe you found that to be a worthwhile comment to type.

People will adapt pretty fast and will stop trusting videos

You mean like they stopped trusting the Internet, or YouTube videos, or newspapers, or old broadcast TV news? Except they didn't, because it's impossible to live life successfully without information sources beyond one's eyes and ears.

That's why this technology should not exist.

uhhh… i would respond differently to your rhetorical question…

there’s never been greater distrust of legacy media, and the fact that you can’t trust everything you read on the internet has been a trope for decades

Good point.

Maybe it's a useful thing to ponder why faked photoshopped pictures were never a big problem in human life. I think maybe it's because we use a lot of pictures in our lives, sure. But ultimately we have so much context that a fake would be easy to detect and therefore irrelevant. At most people used photoshop to alter images of documents.

Becoming good enough at Photoshop to do a convincing face swap was something that took a lot of time and skill. Not everyone with a copy of Photoshop had the ability to create a compromising photo of a politician, for example.

Should written language not exist as well? We can never say for sure that the words are true

stop trusting

Yeah, that's a problem. Successful societies are built on trust, shared reality and communication. Democracy is a conversation.

The big problem with technology is you can't uninvent technologies that turn out to be net bad. It becomes a perpetual curse once it's invented.

"Lighters should not exist. Anyone can start a fire, all the time, everywhere. This will lead to an inferno in no time." etc.

Your sentence is a strawman but the logic does apply quite soundly to e.g. guns.

lighters have surely led to many infernos

"Writing should be restricted to the educated few who can responsibly carry the Church's message." -anti-technologist from 1000 years ago, probably

This is the harbinger that announces that, as a technologist, the time has come for me to witness more and more things that I cannot understand how they work any more. The cycle has closed and I have now become my father.

The difference is that now nobody really "understands" what's going on, it's just that some know how to build these.

How is that new? People built a gnomon, a stick was thrust into the soil and ta-da. No doubt it happened far before any writing system was out there. So it still took human quite some time to come with a compelling helio-centric model to cast some grabbable explanation of it all, even if you take Aristarchus of Samos as a pionner in this field.

It's new for computing.

Ok, maybe on some perspective I’m with you here. There are things happening no-one even those on the edge of the fringe can understand anymore how it works while it does. Or at least that is how it seems to be from my narrow perspective on AI.

On the other hand, I don’t feel like you need to know how a compiler work, let alone the hardware architecture it targets, before you can go through your first hello world program or even build some useful software on top of frameworks/library treated at blackboxes. So "I have no idea what I’m doing" in this perspective is probably as old as CS/informatics.

My dad is 80 and willingly loves to listen to me explain how neural networks work, then he also read about them, busy beaver functions, kafka, and all kinds of crazy shit I tell him abour. This is all in your mind. You are as young as your mind is.

Not the original poster, but the more frightening part of the sentence, is the "not understanding how something works part" over the "becoming my father"

Getting to a point where realistically you're not able to know something deeply but then still use it is pretty frightening.

When I say deeply I don't necessarily mean that for every device you need to know about all of its atoms, but to have a pretty good framework for how the thing works deterministically, and how it can fail.

So unless you have a solid grasp of quantum mechanics and solid state physics, using any electronic device is frightening?

Not the original poster, but the more frightening part of the sentence, is the "not understanding how something works part" over the "becoming my father"

That was my point, exactly.

Thankfully it's nothing magical. But are you willing to learn about it or not?

Think about animation, how a program can generate a sequence of a bouncing ball between two key frames. Think about what defines a video. The frames right? From there I can try to imagine.

But are you willing to learn about it or not?

This is the key. I have enough curiosity to want learn the stuff from the ground up, just as I did with other technologies. But man do I have the stamina today? Not so sure!!!

This is likely a wild guess on my part but i've faced a similar feeling lately. If this comes from the realm of Webdev, React, SSR and all the F'ing acronyms that we need to learn today and you want to feel like you've "caught on": My advice would be to avoid NextJS at all costs. It's too bleeding edge.

Opt for a sane option instead to get started, likely one of these: (Astro, SvelteKit or Remix).

Lol there's a massive difference between a framework that generates javascript, a language which has existed for 30 years at this point, and a magic LLM that no one on earth understands the internals of.

This book is great: https://www.penguinrandomhouse.com/books/730887/understandin...

It's comparatively easy to understand and it does cover everything from basic networks to LLMs and Diffusion models.

I'm on the very cusp of this, you helped me realize. Thanks.

This comment describes with precision what I was feeling and was unable to name or frame. Marvelous times for sure.

It is all just a mindset and how much you want to be involved.

Here is an inspirational story for you: https://news.ycombinator.com/item?id=39288139

Thanks for putting this into words. Its a very off-putting feeling for me, and couldn’t exactly figure out what that feeling was. It both scares me and excites me in a way that only makes me subconsciously anxious. Time to deep dive before I become what I always feared, which is being technologically left behind.

I can’t wait for the day I can strap on my Apple® Vision Pro® 9 with OpenAI® integration and spend all my time interfacing (wink) with my virtual girlfriend. Sure my unlit 3 by 3 meter LifePod® is a little cramped and my arm itches from the Soylent® IV drip, but I’ll save so much time by not having to go outside and interact with legacy humans!

Strap it on? Haven't you heard? The vison pro 9 will be a chip grown on your retina.

Vision pro 10 will be interfacing with your brain using your life simulation for it's own computations. Wait what?

Elon Musk says hi.

Think happy thoughts to reply.

That sounds like a nightmare! You have to buy so many products, you have to keep them charged and you re still missing on so much. You should instead get the Neuralink plug-in pod with builtin feeding tube and catheter

Apple doesn't allow "wink" in apps so no, this will not be happening on an Apple Vision Pro

Don't forget your apple buttplug to remind you of whose fucking you each day

I question how much anyone has really used these models if they actually think these systems can replace people. I’ve consistently failed to get professional results out of these things and the degree of work required to get professional results makes me think a new class of job will be created to get professional results out of these systems.

That being said, there is value in these systems for casual use. For example, me and my girlfriend got into the habit of sending little cartoons to each other. These are cartoons we would have never created otherwise. I think that’s pretty awesome.

The more I use them, the more I get a sense of something fundamental that's missing, and the less I worry about losing my job. It's hard to describe, I need to think harder about what that feeling is.

Art is communication, it's as simple as that. Computer generated stuff isn't communicating anything.

Most people who work in "the arts" probably aren't communicating anything directly either - they just create the scenes, sound effects, textures, animation, models +++ that someone above them in the organization has asked them to create for their project.

What's the difference between having an idea, then putting an actor on a set, lighting them, doing background green screen set extension afterwards, digital clean up, etc. vs doing all of that generatively?

How is asking a VFX house for animated footage any different than generating it? If art is intent, there is no reason you can't generate the building blocks that reflect that intent, no?

Just imagine how annoying the past year was for those of us who had figured this out quicker.

Probably about as annoying as the last 15 years of crypto have been

Many financiers are willing to trade quality with cost reduction.

Wow, some of those shots are so close to being unnoticeable. That one of the eye close up is insane.

It’s interesting reading all the comments, I think both sides to the “we should be scared” are right in some sense.

These models currently give some sort of super power to experts in a lot of digital fields. I’m able to automate the mundane parts of coding and push out fun projects a lot easier today. Does it replace my work, no. Will it keep getting better, of course!

People who are willing to build will have a greater ability to output great things. On the flip side, larger companies will also have the ability to automate some parts of their business - leading to job loss.

At some point, my view is that this must keep advancing to some sort of AGI. Maybe it’s us connecting our brains to LLMs through a tool like Neuralink. Maybe it’s a random occurrence when you keep creating things like Sora. Who knows. It seems inevitable though doesn’t it?

One of things I've loved about HN was the quality of comments. Whether broad or arcane, you had experts the world over who would tear the topic apart with data and a healthy dose of cynicism. I frequently learned more from the debate and critique than I did from the "news" itself.

I don't know what is it about AI and current state of tech, but the discourse as of late has really taken a nosedive. I'm not saying that any of this conjecture won't happen, but the acceleration towards fervor and fear mongering on the subject is bordering on religiosity - seriously, it makes crypto bros look good.

And yeah -- looks like some cool new tech from OpenAI, and excited when I can actually dig in. Would also love it if I could hire their marketing department.

It's pretty obvious why. Automation has finally come for programmers so now everyone here is anti-progress.

This.

Many people here have a lucrative career in traditional fields, big tech, etc.

Working in those fields is good. Building "products" is good (even if that only means optimizing conversion rates and pushing ads). Doing well in the traditional financial sense (stocks and USD) is good.

Anything that rocks the boat (crypto, ai) is bad.

To those who are saying, look at this at a positive and it lets people unleash their creativity?

- This enables everyone to be creators

- Given that everyone's creativity isn't top notch, highest quality will be limited to a the best

- So rest of us will be consumers

- How will we consume if we don't have work and there is no UBI?

If "Given that everyone's creativity isn't top notch, highest quality will be limited to a the best", that implies the existence of professionals, which implies work.

I meant those that are proficient/creative enough to be creating top content using AI but if we take it further to AI using AI, then yes, its AI all the way down.

We won't and the world will go into a massive depression, destroying the market for AI produced garbage and staving off global warming for a few extra years in the process. So even better than UBI.

Where is the link to try it, ChatGPT doesn't know anything about it:

"Sora" is not a video generation technology offered by OpenAI. As of my last update in April 2023, OpenAI provides access to various AI technologies, including GPT (Generative Pre-trained Transformer) for text generation and DALL·E for image generation. For video generation or enhancement, there might be other technologies or platforms available, but "Sora" as a specific product related to OpenAI or video generation does not exist in the information I have.

If you're interested in AI technologies for video generation or any other AI-related inquiries, I'd be happy to provide information or help with what's currently available!

Why would chatGPT know about anything new?

Marketing and Sales?

If this can generate videos in real-time (60FPS), then you can, in theory, create any game just from text/prompts.

You just write the rules of the game and the player input, and let the AI generate the next frame.

Pretty unlikely this generates in real-time.

"Two papers down the line..."

What happens when humanity stops generating new content/recording new findings/knowledge etc ? are at a place where whatever we had is enough knowledge for AI takeover?

or we are heading towards a skynet-y feature

As a counterpoint, i don't think that the average person has stopped taking pictures just because image generation models exists. Nor have people stopped pursuing other hobbies impacted by AI. We don't go to museums to look at AI art that was created in 10 seconds and I doubt culture will shift to a point where that's common place. Human content will still be created, and we will probably see the general quality of that content increase as a result of foundational models. Content creation is taking whats in the mind and translating it into the physical/digital realm. With better AI, this translation becomes easier for a lot of fields and you no longer have to master the use devices to make your art quality. However, everyone can agree that prompt based generation is a lot less satisfying than making content from scratch. It feels more akin to a google search than a satisfying creative process. Those who are passionate and talented will continue to pursue their physical medium because of this.

The monetary value of generic stock content will surely drop and won't be created by professionals anymore. However, that doesn't mean people stop taking pictures of their dog just because they can get midjourney to generate the same thing. Creation for the sake of creation will continue. AI companies will initially reap in a lot of the $ value that used to go to the creators of stock content, but when open source models reach parity the masses will be able to make what's in their mind a reality as casual creators. Hobbyists will still exist and those that become truly great will still rise to notoriety.

What’s the connection between this and high end game engines (like unreal 5). I would expect 3d game engines to be used at least for training data and fine tuning. But perhaps also directly in the generation of the resulting videos?

For example this looks very much like something from a modern 3d engine:

https://twitter.com/OpenAI/status/1758192957386342435

They almost certainly trained on video game output and this is clearly bleeding into the style of some of these demos.

The SUV video for example looks very much like something you'd see in a modern video game which probably makes sense because most videos with kind of perspective are going to be from video games.

I don't know how they would use game engines directly for training and fine tuning though. It would be far too labour intensive to render high quality scenes using a video game engine for every prompt.

All the examples feel so familiar, like I have seen them all before buried in the depths of YouTube and long-forgotten BBC documentaries. Which I guess is obvious knowing roughly how the training works.

I guess what I'm wondering is how "new" the videos are, or how closely do they mimic a particular video in the training set? Will we generate compelling and novel works of art with this, or is this just a very round-about way of re-implementing the YouTube search bar?

Maybe this was a big influence on the woolly mammoth example: https://youtu.be/EzzTX3DYMNs?si=WS28fsf5j6SBI1-7&t=15

Also interesting that some of the examples ignore details in the prompts. No clouds or sun in the sky, no depth of field, their hair isn't blowing in the wind.

Another step in the trend of everything becoming digital (in film and otherwise). It used to be that everything was done in camera. Then we got green screens, then advanced compositing, then CGI, then full realistic CGI movies modeled after real things and mocap suits. Now we're at the end game, where there will be no cameras used in the production of a movie, just studios of people sitting at their computers. Because more and more, humans are more efficient at just about anything when aided by a computer.

A bicycle for the mind.

This might actually ruin video and films for me. I don't want to be looking out for AI giveaways in everything I watch.

I can see a new market for true end-to-end analogue film productions emerging for people who like film.

Eh, it's like watching out for VFX giveaways.

I'd love to feel excited by all these advancements and somehow I feel numb. I get part of the feeling (worry about inequalities it may generate), but I sense something more. It's like I see it as a toy... I'm unable to dream on how this will impact my life in any meaningful way.

Imagine dumping all the HIPAA data into a process like this. Obviously fraught with privacy and accuracy[0] concerns. Nonetheless, it might help us move some things forward.

All current form of entertainment will be impacted, all of them.

Except for live sporting events.

This is why I think megacorps all going to bid for sport league streaming right. That's the only one that AI can't touch.

Anyway to benefit economically from this trend?

Does Google have a competing product I can join the wait list for?

No public access but they have Lumière: https://lumiere-video.github.io/

I find creepy things in all the videos, despite their breathtaking quality at first glance. Whether it is the way the dog walks out into space or the clawlike hand of the woman in Tokyo, they are still uncanny valley to me. I'm not going to watch a movie made this way, even if it costs my $0.15 instead of $15.00. But I got tired of Avatar after watching it for 20 minutes. Maybe all the artificial abundance and intellectual laziness the generative AI world will make us realize how precious and beautiful the real world is. For my kids' sake, I hope so.

Sure, but imagine using this as a generative-fill to augment a movie, not just making an entire movie from it. We've seen fantastic homemade movies from very talented artists before. Now imagine if mostly talented artists could do it too.

I'm just blown away. This can't be real. But lets be face the truth. Its even more impressive than ChatGPT. I think its the most impressive AI tech i've seen till now. I'm speechless.

Now the big question is. As OpenAI keeps pushing boundaries, it's fascinating to see the emergence of tools like Sora AI, capable of creating incredibly lifelike videos. But with this innovation comes a set of concerns we can't ignore.

So i'm worried about getting these tools misused. I'm thinking about what impact could they have on the trustworthiness of visual media, especially in an era plagued by fake news and misinformation? And what about the ethical considerations surrounding the creation and dissemination of content that looks real but isn't?

And, what we should do to tackle these potential issues? Should there be rules or guidelines to govern the use of such tools, and if so, how can we make sure they're effective?

https://news.ycombinator.com/edit?id=39393236

Its why I submitted this. We need some way to attest the authenticity of images.

This technology is going to destroy society.

Want to form a trade union I'm your workplace? Best be ready to have videos of you jacking off to be all over the internet.

Videotape a police officer brutalising someone? Could easily have been made with AI, not admissable.

These things will ruin the ability to trust anything online.

Naw, if/when it gets to that, media won't be believed or admissable unless signed with someone's private keys or otherwise attested.

did anyone saw the two-leg horses in the video?

This is pretty impressive, it seems that OpenAI consistently delivers exceptional work, even when venturing into new domains. But looking into their technical paper, it is evident that they are benefiting from their own body of work done in the past and also the enormous resources available to them.

For instance, the generational leap in video generation capability of SORA may be possible because:

1. Instead of resizing, cropping, or trimming videos to a standard size, Sora trains on data at its native size. This preserves the original aspect ratios and improves composition and framing in the generated videos. This requires massive infrastructure. This is eerily similar to how GPT3 benefited from a blunt approach of throwing massive resources at a problem rather than extensively optimizing the architecture, dataset, or pre-training steps.

2. Sora leverages the re-captioning technique from DALL-E 3 by leveraging GPT to turn short user prompts into longer detailed captions that are sent to the video model. Although it remains unclear whether they employ GPT-4 or another internal model, it stands to reason that they have access to a superior captioning model compared to others.

This is not to say that inertia and resources are the only factors that is differentiating OpenAI, they may have access to much better talent pool but that is hard to gauge from the outside.

Seriously cannot wait to be able to put a 1 weeks worth of dream diary into a tool like this and see my dream inspired movies!

I wonder if we as a society, have overrated value created digitally, and underrated value created physically or with proximity.

We still need nurses, cooks, theater, builders etc.

Watched the MKBHD video on this and couldn’t help but think about copyrights when he spoke of the impact on stock footage companies.

As I understand the current US situation, a straight prompt-to-generate-video cannot be copyrighted. https://www.copyright.gov/ai/ai_policy_guidance.pdf

But the copyright office is apparently considering the situation more thoroughly now.

Is that where it stands?

If it can’t be copyrighted, it seems that would tamper many uses.

The world is changing before our eyes. It's exciting, sure, but I am also deeply afraid. AI may take humans to the next level, but it may also end us.

...and our future lies in the hand of venture capitalists, many of whom have no moral compass, just an insatiable hunger to make ever larger sum of money.

I don't think i like the future.

Even the good ones look kinda bad.

What is the first book you want to see movie of? It should be verbatim and last a week, if needed.

I vote for Hothouse, by Brian W Aldiss. So many images need to imagined, like spiders that jump to the moon and back again.

This looks like state of the art?

puts on the movie industry

Hopefully we will see AIs with tools which are not "paint" or "notepad", but a maths formal proof solver, etc.

But I have a problem: I am unable to believe the videos I saw were dreamt by AI. I can feel deeply that I do believe there is some trickery or severe embellishment. If I am wrong, I guess we are at an inflexion point.

I can recall 10+ years ago, we were talking "in hacking groups" about AI because we thought the human brain alone was not good enough anymore... but in a maths/sciences context.

Genuinely impressive.

I've always been a digital stills guy, and dabbled in video.. as a hobby. As a hobbyist, I always found the hardest thing is making something worth looking at. I don't see AI displacing the pleasure of the art for a hobbyist.

My next guess is the 80/20% or 95/5% problem is gonna be stuff like dialogue matching audio and mouth/face motion.

I do see this kind of stuff killing the stock images / media illustrator / b-roll footage / etc jobs.

Could a content mill pump out plausibly decent Netflix video series given this tool and a couple half decent writers.. maybe? Then again it may be the perpetual "5 years away". There's a wide gap between generating filler content & producing something people choose to watch willingly for entertainment.

Probably we humans will come to a point where we wouldn't even bother ourselves with making videos. We may just consume based on our emotional state on the fly generated by such services.

Ad generation usecases are getting interesting with Video generation + Controlnet + Finetuning

Holy Moly

Sora serves as a foundation for models that can understand and simulate the real world, a capability we believe will be an important milestone for achieving AGI.

Why would it?

Call me a Luddite but I don't want these videos hitting my retinas.

There should be an opt out from being subjected to AI content.

I wonder how much of a blocker to real use not having things like model rigging or fine-tuned control over things will be to practical use of this? Clearly it can be used in toy examples with extremely impressive results, but I'm not entirely convinced that, as is, it can replace the VFX industry as a whole.

US Elections about to peak, terrible timing.

Demo is always better than the real product. We'll soon see how it works...

Wonder how the folks at Runway and Pika are thinking about this.

To me, it's becoming increasingly obvious that startups whose defensibility hinges on "hoping OpenAI doesn't do this" are probably not very enduring ones.

OpenAI will be the most valuable company in history at this rate. This is insane

This will probably cost some downvotes, but can we start a thread explaining the architecture behind this for this interested in how it actually works?

Technical report here: https://openai.com/research/video-generation-models-as-world...

Now that they’ve gone corporate, the OpenAI corporate motto ought to be “Because We Could.”

What the f. What. I'm no AI pessimist by any means but I thought there are some significant hurdles before we get realistic, video generation without guidance. This is nothing short of amazing.

It's doubly amazing when you think that the richness of video data is almost infinitely more than text, and require no human made data.

The next step is to combine LLM with this, not for multimodal, but to team up together to make a 'reality model' that can work together to make a shared understanding?

I called LLMs 'language induced reality model' in the past. Then this is 'video induced reality model', which is far better at modeling reality than just language, as humans have testified.

Maybe this means someone will make a non-superhero movie now.

This is a weapon.

Holy fuck

One one side, we have people who are upset because the creators of the videos in the dataset used for teaching this language model were not compensated.

On the other hand, people find the tech very impressive and there are a lot of mind blowing use-cases.

Personally, this opens up the world for me to create video ads for software projects I create, since I have no financial resources or time to actually make videos, I only know how to code. So I find it pretty exciting. It's great for solo entrepreneurs.

Everytime OpenAI comes up with an new fascinating gen model it also allows for that bluntly eye-opening perspective on what flood of crappy und unnecessary content we have been gotten accustomed to being thrown at us. Be it blown-up text description and filler talk, to these kind of vodka-selling commercial videos.

It's a nice cleansing benefit that comes with these really extraordinary tech achievement that should not be undervalued (after all it produces basically an endless amount of equally trained producers like the industry did in a - somehow malformed - way before).

Poster frames and commercials thrown at us all the time, consumed by our brains to a degree that we actually see a goal in producing more of them to act like a pro. The inflationary availability that comes with these tools seems a great help to leave some of this behind and draw a clearer line between it and actual content.

That said, Dall-E still produces enough colorful weirdness to not fall into that category at all.

AI = Better CGI

Today we scroll social media feeds where every post we see is chosen by an algorithm based on all the feedback it gets from our interactions. Now imagine years down the road when Sora renders at 60 fps, every frame influenced by our reaction to the prior frame.

Looking forward to someone feeding it the first draft of The Empire Strikes Back https://www.starwarz.com/starkiller/the-empire-strikes-back-...

This is machine simulated art. It is not a convincing simulation to videographers, yet it pleases software architects and other non-visual artists. Aptitude for visual art making provokes envy in some who lack it. The drive to simulate art is almost as common as the desire to be recognized as a capable visual artist. The most interesting generative art I’ve seen does not attempt verisimilitude. Children want their art to look real. Verisimilitude is hard, especially for children and quasi AI.

AI was a mistake

Now I can finally adapt my short story into a short film. All for however this thing will end up costing.

I hope this doesn't get buried...

As several others have pointed out, realism of these models will continue to improve, and will soon be economically useful for producing beautiful or functional artifacts - however prompt adherence (getting what you want or intend) of the models is growing much more slowly.

However I think we have a long ways to go before we'll see a decent "AI Film" that tells a compelling story - and this has nothing to do with some sort of naturalistic fallacy that appeals to some innate nature of humans!

It comes down to the dataset and the limits of human creators in their ability to communicate their process. Image-Text and Video-Text pairs are mostly labeled by semi-skilled humans who describe what they see in detail. They are, for the most part, very good at capturing the obvious salient features of an image or a video. "reflections of the neon lights glisten in the sidewalk". However, what you see in a movie scene is the sum total of dozens if not hundreds of influences, large and subtle. Choices made by the actors, camera operators, lighting designers, sound designers, costuming, makeup, editors, etc... Most people are not trained to recognize these choices at all, or might not even be aware that there are choices to make. We (simply) see "Joaqin Phoenix is making awkward small-talk in the elevator with other office workers".

So much of what we experience processes on subconscious and emotional and purely sensory levels, we don't elevate those lower-level qualia to our higher-brain's awareness and label them with vocabulary without intentional training (such as tasting wine, coffee, beer, etc - developing a palate is an act of sensory-vocabulary alignment).

However, despite not raising these things to our intentional awareness, it has an influence on us -- often the desired impact of the person who made that choice in the first place. The overall effect of all of these intentional choices makes things 'feel right'.

There's no fundamental reason AI can't produce an output that has the same effect as those choices, however finding each little choice is like a needle in a haystack. Accurate labeling of the training data tells the AI where to look -- but the people labeling the data are probably not well-versed in all of the little intentional choices that can be made when creating a piece of video-media.

Beyond the issue of the labeling folks being trained in the art itself, there's the problem too of the artists themselves not being able to fully articulate their (numerous, little, snowflake-into-avalanche) choices - or simply not articulating it even if they could. Ask Jackson Pollock about paint viscosity and you'll learn a great deal, but ask about abstract painting composition and there's this ineffable gap that language seems ill-suited to cross. The painter paints what they feel, and they hope that feeling is conveyed to the viewer - but you'd be hard pressed to recreate "Autumn Rhythm (Number 30)" if you had to transmit the information via language and hope they interpreted it correctly. Art is simultaneously vague and specific!

So, to sum up the problem of conveying your intent to the model:

- The training data labels capture obvious or salient features, but not choices only visible to the trained eye

- The material itself is created by human artists who might not even be able to explain all of their choices in words

- You the prompter might not have the vocabulary that captures succinctly and specifically the intended effect

- The end result will necessarily be not quite what you imagined in your mind's eye as a result of all of this missing information

You can still get good results if you tell it to copy something, because the label "Tarantino" captures a lot of detail, even all the little things you and the training data would never have labeled in words. But it won't be yours and - until we have an army of trained artists providing precise descriptions for training data in their area of expertise, and you know how to speak those artists' language - it can't be yours.

Value is going to be higher for professions where the human essence is an essential component of the function. Or professions that are more coupled with physical reality…my hedge is probably becoming an electrician.

I’d imagine IRL no-tech experiences will be the new ‘escapes’ too.

Maybe I’m too idealistic about the importance of the human spirit/essence…whatever that actually is.

Obviously concern yourself with your job and what you need to do to ensure you can obtain buying power going forward, but most problems and concerns about things like these go away if you just turn off your tech, or really be intentional about your usage.

Extremely hard to do, it is, but you’ll become quasi-Amish and realize how little is actually actionable and in our control.

You’ll also feel quite isolated, but peaceful. There’s always tradeoffs. You can’t have something without giving up not-something, if that makes sense.

Edit: So, essentially, ignorance is bliss, but try to look past the pejorative nature of that phrase and take it for what it is without status implications.

Related ongoing thread: Video generation models as world simulators - https://news.ycombinator.com/item?id=39391458 - Feb 2024 (43 comments)

Also (since it's been a while): there are about 1500 comments in the current thread. To read them all, you need to click More links at the bottom of the page, or like this:

https://news.ycombinator.com/item?id=39386156&p=2

https://news.ycombinator.com/item?id=39386156&p=3

https://news.ycombinator.com/item?id=39386156&p=4[etc.]

Beautifully terrifying

Wow. It's bizarre to see these video's.

Creating these video's in CGI is a profession that can make you serious money.

Until today.

What a leap.

Wow. And just like that fliki.ai and similar products have been sherlocked. Great time to be a creator, not the best time to be a product developer, production designer

What makes OpenAI so far ahead of all of these other research firms (or even startups like Pika, Runway, etc.)? I feel like I see so many examples of fields where progress is being made all across and OpenAI suddenly swoops in with an insane breakthrough lightyears ahead of everyone else.

I'm simply blown away

I'm just blown away. This can't be real. But lets be face the truth. Its even more impressive than ChatGPT. I think its the most impressive AI tech i've seen till now. I'm speechless. Now the big question is. As OpenAI keeps pushing boundaries, it's fascinating to see the emergence of tools like Sora AI, capable of creating incredibly lifelike videos. But with this innovation comes a set of concerns we can't ignore.

How will it effect gaming industry? https://news.ycombinator.com/item?id=39393252

It's odd how the model thinks "historical footage" could be done by drone. So it understands that there should be no cars in the picture. But not that there should be no flying perspective.

Looking at it and in my opinion it just reinforces theory that we live in simulation

In the future, we're not going to have common tv shows or movies. We'll have a constantly evolving stream of entertainment that's perfectly customized to the viewer's preferences in real time. This is just the first step.

"We’ll be taking several important safety steps ahead of making Sora available in OpenAI’s products. We are working with red teamers — domain experts in areas like misinformation, hateful content, and bias — who will be adversarially testing the model." - To make sure that the perfectly unbiased algorithms are biased against bias. So in essence, red teamers as in commies I suppose.

What goes around, comes around. I'm glad this is happening. Gitty and friends should be driven out of business for their absurd stunt they pulled with image search.

Yes, I'm still bitter about that.

Welp, goodbye internet, it was fun to know you.

The rendering of static on the TVs is interesting/strange. Must be hard for AI to generate random noise:

Video 7 of 8 on the 2nd player on the page.

Prompt: The camera rotates around a large stack of vintage televisions all showing different programs — 1950s sci-fi movies, horror movies, news, static, a 1970s sitcom, etc, set inside a large New York museum gallery.

Anyone to invite

This is surreal, both literally and figuratively.

Videos don’t feel real though this is best thing I have ever seen on topic ‘text-to-video’. I am sure this will go so far and become more realistic. But does this mean that we will not hire actors and creators but we will hire video editors who can stitch all together and prompt writers who can create tiny videos for story.

Here is my prediction of how this will play out for the entertainment industry in the coming decades:

Phase 1 (we are here now): While generative AI is not good enough to directly produce parts of the final product, it can already be used to quickly prototype styles, stories, designs, moods, etc. A good chunk of the unnamed behind-the-scenes-people will loose their job.

Phase 2: While generative AI is still expensive, the output quality is sufficient to directly produce parts of / the entire final product. Big production outlets will use it to produce AAA titles and blockbusters. Even actors, directors and other high publicity positions will be replaced.

Phase 3: The production cost will sink further until it becomes attainable by smaller studios and indie productions. The already fierce markets will be completely flooded with more and more quantity over quality. Advertisement will not be pre-produced and cut into videos anymore but become very subtle product placements, impossible for ad-blockers to strip from the product.

Phase 4: Once the production cost falls below the price of one copy of the product, we will get completely customized entertainment products tailored to our personal taste. Online communities will emerge which craft skeletons / templates which then are filled out by the personal parameter sets of the consumers. That way you can still share the experience with friends even though everybody experiences a different variation.

Phase 5: As consumers do not hit any production limits any more (e.g. binge watch their favorite series ad infinitum) and the product becomes optimized to be maximally addictive by measuring their reaction to it, it will become impossible for most human beings to resist. The entertainment mania will reach its peak and social isolation, health issues and economic factors will bring down the human reproduction rate to basically zero.

Phase 6: Human civilization collapsed within one or two generations and the only survivors will be extremely technology-adverse people by selection. AGI might have happened in the meantime but did not have the time to gracefully take over and remodel the human infrastructure to become self sufficient. Instead a strict religion will rule the lands and the dark ages begin anew.

Note that none of this is new, it is just the continuation and intensification of already existing trends. This is also not AGI doomerism as it does not involve a malicious AGI gone rouge or anything like that. It is simply what happens when human nature meets powerful technology.

TLDR: While I love the technology I can only see very negative long-term outcomes from this.

Real GPT-4 moment. Your 3500 MacBook cannot do this.

This is a great technical achievement, but in a couple of years time this will be as interesting as AI image generators.

All the software engineers and VFX people training to become plumbers. I'm afraid your clients will be jobless or underpaid by that time.

Jokes aside. It's becoming more apparent, Power will further concentrate to big tech firms.

Good by civilization