How will smart software (artificial intelligence) influence the perception and creation of music?

[telecode]

Do you think AI will write all of our music in 20 years?

I think AI tools will be used in making music in the future. Just like we use tools like drum machines and synths libraries instead of real drummers and orchestras. I dont know what those tools will be.

Its possible AI will be used to make the generic and filler type stuff. If you think about it, making music and making songs is a very mathematical process that usually follows a template (aka defined rules). But I dont think it will ever make the truly amazing ground breaking stuff, and there will always be a need and there will always be amazing ground breaking musicians and artists.

[Forgotten]

All of it? I don't think it will even be writing a small fraction of it.

[xoxos]

AI/ML is basically an open-ended formula trained on data provided by humans. trending, people are using AI for every task they can think of. because people are interested in AI, they would prefer to hear about an AI solution than a programmed solution.

when your AI is generating music, chances are, whoever implements it (google magenta), won't extrapolate much. ML data sorting fits to a set, it is interpolative. things that sound like what you've already heard.. a poor understanding of music, but one shared by many contemporaries.

remember the first tron? it was weird lookin, abstract. the way i remember it, c. 1980, all these weird fiction movies like alien, thing, star wars, conan, clash of teh titans, happened, then after the mid-80s fantasy fiction dropped off the face of the planet for a looong time (maybe because everything would be compared to the big yielders). i had a look at the tron remake, looks like every bit of abstraction has been squeezed out of it. photo realism, nothing abstract, unusual, novel.

abstraction has been like that since.. it has to resemble a preexisting abstraction. the first star wars movies had a gimmick, each film had new environment, new vehicles. now, to be star wars, it has to have old environment, old vehicles.

i think AI has a tremendous potential to continue to bury culture under the mediocrity of market dominance over actual substance, same as we've seen all our lives. will we even be able to tell teh difference between real and artificial dogpoo. it doesn't matter because your life depends on getting your product on the shelf and moved regardless of your ability to consider its ultimate merit and detriment.

[xoxos]

Anyone who truly appreciates music knows AI music is dead, both actually and figuratively.

music is a special place where death and life coexist

we been tryin to tell yall for years
how you need to open your ears
take a trip down to teh dirty sout
find out what this woof woof about

https://xoxos.bandcamp.com/album/dope-beats-for-suckers
*smiles and takes off hat*

[telecode]

AI/ML is basically an open-ended formula trained on data provided by humans. trending, people are using AI for every task they can think of. because people are interested in AI, they would prefer to hear about an AI solution than a programmed solution.

when your AI is generating music, chances are, whoever implements it (google magenta), won't extrapolate much. ML data sorting fits to a set, it is interpolative. things that sound like what you've already heard.. a poor understanding of music, but one shared by many contemporaries.

remember the first tron? it was weird lookin, abstract. the way i remember it, c. 1980, all these weird fiction movies like alien, thing, star wars, conan, clash of teh titans, happened, then after the mid-80s fantasy fiction dropped off the face of the planet for a looong time (maybe because everything would be compared to the big yielders). i had a look at the tron remake, looks like every bit of abstraction has been squeezed out of it. photo realism, nothing abstract, unusual, novel.

abstraction has been like that since.. it has to resemble a preexisting abstraction. the first star wars movies had a gimmick, each film had new environment, new vehicles. now, to be star wars, it has to have old environment, old vehicles.

i think AI has a tremendous potential to continue to bury culture under the mediocrity of market dominance over actual substance, same as we've seen all our lives. will we even be able to tell teh difference between real and artificial dogpoo. it doesn't matter because your life depends on getting your product on the shelf and moved regardless of your ability to consider its ultimate merit and detriment.

I guess I'll chime in here as I have a keen interest in this stuff. (I will also risk getting flamed for having an unpopular view in a community of "creative" music maker types!)

The buzz around AI, ML and deep learning mostly revolves around getting fast expensive computers to analyze and find patterns is large amounts of data. And yes, computers are much better and much faster at analyzing and finding patterns in large amounts of data.

Now to bring it around to music and creation of music and creativity. There are basically three types of creativity. (stolen from Boden!)
- combinational. where you combine different creative ideas that have already been done in the past.
- exploratory. where you generate a new creative ideas based on the creative ideas you have combined from above.
- transformative . where you push boundaries of what is means to be a creative idea. aka you are making stuff that sounds like nothing ever done before.

For anyone that has been around a long time and/or is well versed and exposed to lots of different music and artists, you will notice that a lot of music out there usually sounds like a combination of other stuff that has been done in the past or it sounds like a pastiche or "hommage" to eras long gone (aka the current retro 80s sounding stuff and the popularity of Quentin Tarantino "art"!)

The bad news is most of us music makers make the first two! Very few make the last one and the ones that do make transformative creative music are going to be the very few really well known creative niche ones that have been around a very long time.. like Phillip Glass et la.. and everyone else is just gonna get dismissed as making "real weird sounding mutta fuggin shit!".. (unless it ever becomes a big hit and is proven to make lots of $$$$)

So.. that beings us to the amount of music content out there for really fast and powerful computers to be able to analyze and find patterns in. There is a lot more generic and mediocre sounding music out there being made by people and very little tranformative music. ML thrives on lots of content and data and hence will be able to very efficiently analyze the mediocre and generic sounding stuff and will be able to in the not too distant future very easily replicate it. This is bad news for music makers who make a living from making generic mediocre stuff which is most music makers.

(Yeah, i know the speil.. but "your" music is the best thing since sliced bread and it comes from the "heart" and is "unique" and sounds like nothing else ever made.)

[vurt]

are you that robot that lives in toronto?

[mgw38]

The buzz around AI, ML and deep learning mostly revolves around getting fast expensive computers to analyze and find patterns is large amounts of data. And yes, computers are much better and much faster at analyzing and finding patterns in large amounts of data.

Two corrections. Expensive computers are no longer needed for Deep Learning. Without DL your phone could not perform face recognition.

Not long ago Nvidia came out with the Jetson Nano which has a GPU based 128 core processing unit capable of running DL applications in a 99$ Rasperry PI sized package.

And second, DL is not about finding patters. Well, you can use it to find patterns, but at its core it works at a much more abstract level.

[xoxos]

imo the? appreciable quality of generative music is to contrast the acculturated discretion (*getting over dumb shit is a personal journey). in the 90s, electronic musicians were always described in music magazines as "gaining inspiration from (street) sounds around them" .. CG music hasbeen part of that environ, like metal guitarists inspired by pacman. i've seen a bird imitate a dial-up modem.. i've seen pitching change culturally with technologies and trends.

anyone talking about what is not possible isn't talking.

maybe teh whole "from teh heart" thing is like, 80s, synths are not real instruments, add a decade or so, "synths are real instruments if played form teh heart". my idea about experiential subjectivity was influenced in the 80s by the KLF, "play two notes, out come a million years of pain and lust". maybe a whole bunch of us are just suckers who have unrealistic perspectives on what music, and recording, is.

which is my real issue, i can't separate what things mean to music from what they mean to everything else enough to BS a music career. life is going to change, not just music. when that happens, i don't want to be sitting there trying to get my 12" collection in the right order.

[telecode]

imo the? appreciable quality of generative music is to contrast the acculturated discretion (*getting over dumb shit is a personal journey). in the 90s, electronic musicians were always described in music magazines as "gaining inspiration from (street) sounds around them" .. CG music hasbeen part of that environ, like metal guitarists inspired by pacman. i've seen a bird imitate a dial-up modem.. i've seen pitching change culturally with technologies and trends.

anyone talking about what is not possible isn't talking.

maybe teh whole "from teh heart" thing is like, 80s, synths are not real instruments, add a decade or so, "synths are real instruments if played form teh heart". my idea about experiential subjectivity was influenced in the 80s by the KLF, "play two notes, out come a million years of pain and lust". maybe a whole bunch of us are just suckers who have unrealistic perspectives on what music, and recording, is.

which is my real issue, i can't separate what things mean to music from what they mean to everything else enough to BS a music career. life is going to change, not just music. when that happens, i don't want to be sitting there trying to get my 12" collection in the right order.

I envision (and once again, i am just some random guy on the internet!) AI/ML in music will be used as a tool in your toolbox and will enable you to create more interesting things. It will help you achieve another level if you are creative. or it will just spew out generic stuff if you are not really that creative. Its a tool.

It's sort of like the synths or the 808s when they were released and compared to the present. Back in the 70s, Larry Fast did these amazing wizard type things with synths that was a technology that was not available to everyone and it was amazing and impressive. Today, what Larry Fast was able to do my 14 year old can do on his laptop with VSTs and its not that impressive anymore. Back in the late 70s and early 80s the beats and stuff that early dance music makers and hiphopers were doing with Linn Drum machines and 808 was cool and interesting and impressive. Today its not that impressive as any shmo can make beats on an iPhone. They are all technology and they are all tools in a toolbox. What you do with those tools and your ability to translate your creativity into some sort of "value" to society (aka your listeners) is what matters. Programming an AI or neural networks to be able to translate creative output into high value for society is going to be a lot harder (maybe impossible!) compared to programming it to look for patters in Bach and EDM music and spew out generic Bach and EDM.

[Jafo]

Honestly, I don't see the problem. Technology exits to free humans; if something is simplistic enough to be done by a computer, maybe it should be done by a computer. Pleasant is pleasant. *shrug* Maybe we should listen to more complex music?

[Aloysius]

I think AI could make generic shite for Films, TV etc ... but I think humans will still want to make music too.

[Dunbar]

I imagine most people will become editors and curators of sounds. The collections they present will showcase their good taste rather than skill, knowledge, or talent.

The DJ will inherit the Earth.

[Googly Smythe]

...Technology exits to free humans; ...

If only.

(I know the word should be "exists".)

[telecode]

I imagine most people will become editors and curators of sounds. The collections they present will showcase their good taste rather than skill, knowledge, or talent.

The DJ will inherit the Earth.

My understanding is that's what the AI scientists expect. But as always, they are technologists and not artists so I take that all with a grain of salt.

I think it will be yet another tool in a tool box just.like any other tool. It's what you do with the tools that determine it's true artistic value to society. But don't discredit the tools just because they are new.tools you are unfamiliar with.

[ARPie]

I think that tools can be only as good as the one who is operating them. "Good" in terms of valuable means that we are expecting "new" ideas that we like. But what we like is changing in times.

Imagine someone would have presented glitch hop music in the times of 68... It would have been considered as not being good music. It's a little bit like Marty Mc Fly as a time traveler playing his guitar on stage.

So an AI allways needs to fulfill our expectations. Until some "new" idea gets in the range to be acceptable as a good idea.

I will have an eye on the future processes and tools for creating music, but basically in my opinion, I think, I will be not very impressed by what a tool like AI can help. As the human is the limiting factor at the decission what is "good" or not, it will allways stay in the range between boaring expectable results and unusable noise. The range in between is to large for all different tastes of music.

In the end it leads to a point where I am fooled by marketing buzzwords (and spent a lot of money) to use AI to invent something what I never have invented. The truth for me is that I have just selected from the AI what I like.

In my personal conclusion, it is more fun to be creative (make music myself, trying out new methods that I like/invent, be proud of my ideas, making errors and to devellop my skills in inventing) as to tell an AI, what result is good or not good until it fulfills my expectation. If I would like to use the decission process, I only need to find and listen to "good" music from others or mix/combine existing material. Here an AI could help better, but this will not be very "creative" .

I know, there are a bunch of people, which call themself "creative musicians" by clicking together preproduced loops and fragments. My decission is: I don't like that method. I love natural stupidity instead of artificial intelligence!