AI-generated music just around the corner: How do you feel about it?

[chocobitz825]

Maybe. Or they will break the current system, but won't be able to establish a coherent market for AI produced content, sort of like what happened with NFTs, except they will destroy some of the most important industries and professions that current society uses to reproduce itself (education, culture and entertainment, journalism, medicine, law, etc.) in the process. I see that as a definite possible outcome, and perhaps even the most likely. In any case I would say the future on all of this is most cloudy. Yes, money tends to win in the end, and will almost certain do so in this case too, but the way it ends up winning is not generally the way anyone can predict, at least in part because it is hard to say who now has the decisive coin that will be leveraged into the ultimate hoard.

I think if we're optimistic we've assumed that since the mass commercialization of art, it was always for the sake of art, but I think the true focus of the industry hasn't changed. Create content for profit. Whatever value we feel art and culture have in the world, it's most likely secondary to the desire to consume, and be a part of a large social experience.

AI will change everything. Not for the better in our industries, but it will provide for people. This is a fight we can't win if we're thinking AI or nothing. The best we can hope for is that AI is of some use to us in our professional fields.

 

[AnhrithmonGelasma]

"The new ‘neuroforecasting’ approach uses a machine learning model applied to neural responses to predict hit songs with 97% accuracy

...

“By applying machine learning to neurophysiologic data, we could almost perfectly identify hit songs,” said Paul Zak, a professor at Claremont Graduate University and senior author of the study published in Frontiers in Artificial Intelligence.

“That the neural activity of 33 people can predict if millions of others listened to new songs is quite amazing. Nothing close to this accuracy has ever been shown before.”"

Hit or Miss? AI and Brain Waves Tune into Future Hit Songs with 97% Accuracy - Neuroscience News

Researchers developed a machine learning model that uses neural responses to predict the success of songs with 97% accuracy.

neurosciencenews.com

From the study:

"... we analyzed data only from songs with which participants were unfamiliar.

For unfamiliar songs, self-reported liking was statistically identical for hits and flops ...

The key contribution of the present study is to demonstrate that neurophysiologic measures accurately identify hit songs while self-reported “liking” is unpredictive."

Accurately predicting hit songs using neurophysiology and machine learning

Identifying hit songs is notoriously difficult. Traditionally, song elements have been measured from large databases to identify the lyrical aspects of hits. We took a different methodological approach, measuring neurophysiologic responses to a set of songs provided by a streaming music service...

www.frontiersin.org

Granted, self-reported unfamiliarity may not be accurate. The study needs to be replicated with a larger sample size and songs that haven't been released to the public.

 

[Vladinemir]

That's a wonderfully utopian way of thinking, but it's not how the world works. AI is a business opportunity. In board rooms all over the world, companies are strategizing how to increase profits using AI. These strategies will involve delegating all kinds of jobs to AI. Whether those jobs are menial like doing taxes, or creative like composing music, is irrelevant.

I wonder what will happen if AI develops so quickly it will cause job loss on a large scale. It doesn't look like there will be equally large number of new jobs. Who will then use products and services provided by AI? Number of customers might drop significantly or the price will need to be reduced which could yield even lesser profit. Could lucrativeness actually cause the end of many companies? Too superficial thinking I guess, but maybe someone with knowledge in this area could share their thoughts.

 

[chocobitz825]

I wonder what will happen if AI develops so quickly it will cause job loss on a large scale. It doesn't look like there will be equally large number of new jobs. Who will then use products and services provided by AI? Number of customers might drop significantly or the price will need to be reduced which could yield even lesser profit. Could lucrativeness actually cause the end of many companies? Too superficial thinking I guess, but maybe someone with knowledge in this area could share their thoughts.

Probably the number of people trying to self-employ will increase. How much they can make is yet to be seen…but when considering how much money makes the rounds in virtual economies like twitch and tik tok…a whole new age of entertainment and employment might be ahead of us.

 

[Sophus]

I wonder what will happen if AI develops so quickly it will cause job loss on a large scale. It doesn't look like there will be equally large number of new jobs. Who will then use products and services provided by AI? Number of customers might drop significantly or the price will need to be reduced which could yield even lesser profit. Could lucrativeness actually cause the end of many companies? Too superficial thinking I guess, but maybe someone with knowledge in this area could share their thoughts.

If AI gets that good, then basically everyone can create their own company and offer services based on AI.

Don't know how to create a company? Ask the AI! How to win customers? Ask the AI! How to solve the problems of the customers? Ask the AI! And so on.

You don't get any composing jobs? Create your own AI movie and compose music for it and sell the movie.

 

[Alchemedia]

Netflix Advertises $900K AI Job Ignores WGA's & SAG's Strike

Why is Netflix trying to replace writers and actors without paying them?

nofilmschool.com

 

[chocobitz825]

Netflix Advertises $900K AI Job Ignores WGA's & SAG's Strike

Why is Netflix trying to replace writers and actors without paying them?

nofilmschool.com

This move will likely lead them to a model of profitability. If not Netflix, some other company trying the same. There’s no winning here.

 

[MartinH.]

"The new ‘neuroforecasting’ approach uses a machine learning model applied to neural responses to predict hit songs with 97% accuracy

...

“By applying machine learning to neurophysiologic data, we could almost perfectly identify hit songs,” said Paul Zak, a professor at Claremont Graduate University and senior author of the study published in Frontiers in Artificial Intelligence.

“That the neural activity of 33 people can predict if millions of others listened to new songs is quite amazing. Nothing close to this accuracy has ever been shown before.”"

Hit or Miss? AI and Brain Waves Tune into Future Hit Songs with 97% Accuracy - Neuroscience News

Researchers developed a machine learning model that uses neural responses to predict the success of songs with 97% accuracy.

neurosciencenews.com

From the study:

"... we analyzed data only from songs with which participants were unfamiliar.

For unfamiliar songs, self-reported liking was statistically identical for hits and flops ...

The key contribution of the present study is to demonstrate that neurophysiologic measures accurately identify hit songs while self-reported “liking” is unpredictive."

Accurately predicting hit songs using neurophysiology and machine learning

Identifying hit songs is notoriously difficult. Traditionally, song elements have been measured from large databases to identify the lyrical aspects of hits. We took a different methodological approach, measuring neurophysiologic responses to a set of songs provided by a streaming music service...

www.frontiersin.org

Granted, self-reported unfamiliarity may not be accurate. The study needs to be replicated with a larger sample size and songs that haven't been released to the public.

I think companies like apple that work on VR headsets with very detailed eyetracking are going to mine the hell out of the data they collect there to create scary accurate predictive models of what certain groups of people will like or even more generally how they will react to it. I think it will lead to a whole different level of algorithmic manipulation than we already have on sites like youtube etc.. They'll be able to create AI generated content based on what they already know certain people will engage strongly with and then will feed it to them directly.

 

[chocobitz825]

I think companies like apple that work on VR headsets with very detailed eyetracking are going to mine the hell out of the data they collect there to create scary accurate predictive models of what certain groups of people will like or even more generally how they will react to it. I think it will lead to a whole different level of algorithmic manipulation than we already have on sites like youtube etc.. They'll be able to create AI generated content based on what they already know certain people will engage strongly with and then will feed it to them directly.

Well at least it won’t be boring. Sensory overload all day long.

 

[Mr Greg G]

Netflix Advertises $900K AI Job Ignores WGA's & SAG's Strike

Why is Netflix trying to replace writers and actors without paying them?

nofilmschool.com

It's just sad. When it will come to this point, I will just stop watching new movies. What's amazing in a piece of art, and especially one you like, is that it was carefully thought out by a human brain. You remove this, it's just pointless. It's like watching a robot performing gazillion notes on a piano: it's not impressive.

 

[Epic Wonder]

I do not think any humans would qualify for that job at Netflix. Only AI would be able to gain the deep knowledge required to execute this job. In essence, humans need not apply.

 

[Stringtree]

I find myself reading AI "journalism" on sites that used to feature real writers, and the bottom just drops out after a few lines. I find myself in the uncanny valley, and I feel cold and sick. Far from home.

It has some techniques of real writing, like painting a backstory and easing the reader into the narrative, but is like Stepford Wives after a short few sentences.

My teeth grit together as I realize I've been taken by a writing model burping up bullshit, spewing flowery, nice phrases that mean nothing. Writing that isn't in any particular voice or character. Writing that isn't human.

Hard newswriting traditionally possesses a particular tone. Dry, factual, unaffected, dispassionate. In outstanding cases, the telling of a story can elide into the novelesque, but there'd better be good reasons for that.

Like stories about mistakes in history, long ago. Or uniquely outrageous stories that deserve a little editorializing. An article that doesn't purport to supplant the "just the facts" perspective of straight journalism.

How's your BS detector? Most people, with or without formal training, have more than adequate skills in this area.

 

[chocobitz825]

I find myself reading AI "journalism" on sites that used to feature real writers, and the bottom just drops out after a few lines. I find myself in the uncanny valley, and I feel cold and sick. Far from home.

It has some techniques of real writing, like painting a backstory and easing the reader into the narrative, but is like Stepford Wives after a short few sentences.

My teeth grit together as I realize I've been taken by a writing model burping up bullshit, spewing flowery, nice phrases that mean nothing. Writing that isn't in any particular voice or character. Writing that isn't human.

Hard newswriting traditionally possesses a particular tone. Dry, factual, unaffected, dispassionate. In outstanding cases, the telling of a story can elide into the novelesque, but there'd better be good reasons for that.

Like stories about mistakes in history, long ago. Or uniquely outrageous stories that deserve a little editorializing. An article that doesn't purport to supplant the "just the facts" perspective of straight journalism.

How's your BS detector? Most people, with or without formal training, have more than adequate skills in this area.

I feel ya, AI-written articles can totally seem like they're just spinning out sentences without a real human touch. But, honestly, have you checked out some of the stuff churned out by people lately? With everyone scrambling to pump out content left, right, and center, a lot of it has ended up being just as generic, if not more.

So, yeah, AI writing can be a bit of a bummer sometimes, but let's not pretend that all human writing is hitting it out of the park these days. We're all trying to navigate this content flood, you know? We'll just have to see where it takes us.

 

[LostintheBardo]

I find myself reading AI "journalism" on sites that used to feature real writers, and the bottom just drops out after a few lines. I find myself in the uncanny valley, and I feel cold and sick. Far from home.

It has some techniques of real writing, like painting a backstory and easing the reader into the narrative, but is like Stepford Wives after a short few sentences.

My teeth grit together as I realize I've been taken by a writing model burping up bullshit, spewing flowery, nice phrases that mean nothing. Writing that isn't in any particular voice or character. Writing that isn't human.

Hard newswriting traditionally possesses a particular tone. Dry, factual, unaffected, dispassionate. In outstanding cases, the telling of a story can elide into the novelesque, but there'd better be good reasons for that.

Like stories about mistakes in history, long ago. Or uniquely outrageous stories that deserve a little editorializing. An article that doesn't purport to supplant the "just the facts" perspective of straight journalism.

How's your BS detector? Most people, with or without formal training, have more than adequate skills in this area.

Anyone know if any blind testing has been done on this?

 

[Stringtree]

Yes.

Bad writing has always been there for us to read. But it's been, until recently, pre-ranked within a hierarchy of seriousness with which we are to take it, cues to its authenticity and housed within publishing venues in which dreck writing couldn't survive more than a few seconds without being discovered.

Purveyors will now allow anything, even machine-written prose, inside that envelope of surety. Without question, for a quick buck, but without realizing the practice is eroding their standing as curators, they will allow more confusing content into a swamp of acceptable possibilities.

The end result is that we will no longer be able to rely upon a shared reality, based on careful management of the trajectory of the good, and rejection of the lazy and the bad.

It's happened anyway, I'm just not very happy with it. Maybe the curators ought to have been fired long ago.

 

[Stringtree]

Anyone know if any blind testing has been done on this?

The problem is, friend, I don't want to accuse anyone of this sort of thing. Maybe they just suck. I was mostly calling out to other people who love reading to ask if they'd experienced this kind of emptiness in reading.

There's a tone. There's an emptiness. I've been reading a long time, and I think there's an encroaching stupidness in writing. But I'm not going to accuse a person of using these tools and shirking their writing.

Anyone else hear the dull whistling of AI prose? Poetry? Songwriting? There's always been stupid, but this new regime is terrifying. There's nobody there.

 

[chocobitz825]

I feel ya, AI-written articles can totally seem like they're just spinning out sentences without a real human touch. But, honestly, have you checked out some of the stuff churned out by people lately? With everyone scrambling to pump out content left, right, and center, a lot of it has ended up being just as generic, if not more.

So, yeah, AI writing can be a bit of a bummer sometimes, but let's not pretend that all human writing is hitting it out of the park these days. We're all trying to navigate this content flood, you know? We'll just have to see where it takes us.

this was AI btw. *shrug* fun experiments.

 

[AnhrithmonGelasma]

this was AI btw. *shrug* fun experiments.

"Artificial intelligence can seem more human than actual humans on social media, study finds

... “The most surprising discovery was that participants often perceived information produced by AI as more likely to come from a human, more often than information produced by an actual person. This suggests that AI can convince you of being a real person more than a real person can convince you of being a real person ... ”"

Artificial intelligence can seem more human than actual humans on social media, study finds

A new study suggests that OpenAI's GPT-3 can both inform and disinform more effectively than real people on social media. The research, published in Science Advances, also highlights the challenges of identifying synthetic (AI-generated) information, as GPT-3 can mimic human writing so well that...

www.psypost.org

 

[chocobitz825]

"Artificial intelligence can seem more human than actual humans on social media, study finds

... “The most surprising discovery was that participants often perceived information produced by AI as more likely to come from a human, more often than information produced by an actual person. This suggests that AI can convince you of being a real person more than a real person can convince you of being a real person ... ”"

Artificial intelligence can seem more human than actual humans on social media, study finds

A new study suggests that OpenAI's GPT-3 can both inform and disinform more effectively than real people on social media. The research, published in Science Advances, also highlights the challenges of identifying synthetic (AI-generated) information, as GPT-3 can mimic human writing so well that...

www.psypost.org

I would believe it. It’s the most practical place I’ve seen AI used. With a need to constantly post, and the contents of the posts not needing to be particularly deep, AI makes serviceable friendly posts that fit marketing needs. It’s the kind of labor I don’t want to do, and hiring someone else to do feels like a waste of their better skills as well. The misinformation problem is really something we’ll struggle with, AI or otherwise. I would find misinformation to be the problem, rather than the method of creation and distribution. Perhaps a reliable fact checking AI is necessary.

Just a note on the earlier post, its first draft was extremely bland and obviously AI. It was crafted better when I told it to be more casual and less perfect.

The prompt for that response was:
"write a reply to the following text that states that even if AI does feel generic, I would argue that recently with the need to constantly make content, most human-written articles have been equally bland and generic"


Anyways, I enjoy this existential crisis. It's led me to believe that our ability to express ourselves is what most people think is what being human is about. People think our ability to convey our feelings should be sacred and no machine could mimic it. I think AI could sufficiently craft things that would be perceived as artistic. I've come to the conclusion that more than our ability expression emotions, our ability to feel emotions is the true human experience. What triggers that emotional response is irrelevant. Whether you're moved by a poem, an AI art piece, a random cat pic, or a sunset..the true experience is feeling something from that experience.

There's a song by the artist Sleeping At Last, called "Saturn" with a beautiful line by a human artist expressing a beautiful concept, but the concept itself doesn't need humans to validate the experience.

"I'd give anything to hear
You say it one more time,
That the universe was made
Just to be seen by my eyes.

With shortness of breath, I'll explain the infinite.
How rare and beautiful it truly is that we exist."

 

[dpasdernick]

Machines were supposed to figure out how to cut my lawn so I would have more time to write music. Instead, it figured out how to write music so now I have more time to cut my lawn. Clever little beggars.