Suni ai and Udio...the end of us?

[Thundercat]

Well, actually these lies are not at all shrouded in mystery! It is well documented that the oil industry has been financing the same kind of people (and often virtually the same people) who started out spreading misinformation about nicotine not being addictive and smoking not being cancerogenic since the 60ies. The Tobacco industry has tried to keep the money machine running as long as possible. In the same way big oil did that. And no democratic society on the planet has successfully countered their campaigns, although all of the so called facts they speak were nothing but lies. It has been proven once more that in capitalism you can counter facts with money and hordes of stupid people falling for the wrong reason for your agenda. As long as we do not prevent that from happening agains and again, I can only see capitalism leading us to the brink of extinction or beyond that (which will probably take longer than I will be around, but still ...).

If I understand correctly, I think you are trying to say that there is no climate change and we just need to recycle our stuff? No, it would be great if that was true. But unfortunately it is true the other way around. And if you think that climate change has been made up, you belong to those who have been successfully manipulated by what I just described. I hope I just misunderstood you ...

Scientific consensus on climate change - Wikipedia

en.wikipedia.org

The other side of the coin is, though, that almost all people who agree with climate change (next to other environmental threats) being a legitimate problem are also fooling themselves with a narrative that we just need to take a few measures like replacing oil with electricity and everything can basically stay as it is with a few changes and the help of technology. And in that way, almost all of society is basically not seeing what is going to come.

We agree in principle on most, but no, climate change is NOT A THING. It's simply true that unless a scientist agrees with it, they are not funded. Have you heard from the scientists who have debunked it? No? Why? Info is out there - not on MSM though, and NOT a conspiracy "theory." It's not a theory if it's true. And unfortunately much of our science has been captured by nefarious interests.

Or do you not believe that very rich people conspire in secret to control things? If not, perhaps you do not understand the world.

A few decades ago it was global cooling by the way.

And it is unfortunately all too true - the are going to tax us for our "carbon footprint" and then limit travel. Heard of 15 minute cities? Want to have to stay within a 15 minute radius of your home, for most of your life? In the name of preventing climate change? They already are talking about limiting air travel to 2-3 times a year per person. And even talking about reducing airports to just a handful. Are you aware of this?

You have to look deeper - who benefits? You can be certain the elites will still be traveling around in their private jets as they exhort you to stop driving to the market.

In any case, I enjoyed your post, and I'll stop my derailment here of the original topic.

 

[Thundercat]

A lot of things need to happen for this, just to mention a few: AI needs to be on the same level, or better than a human when it comes to:

Creativity
Complex problem solving
Emotions

In addition, it needs to be:

Conscious, or at least on a level where no technical input is needed
Make decisions that do not rely on the past, but on the current situation, i.e. it should be able to reason without data
Should not need lots of resources, so it can be accessible to the average computer user, thus cheaper than human labour
Have the potential to not violate copyrighted works, AI or non-AI

Currently, the only viable way is through a dataset, which is produced by human musicians, for example if we're talking about a model for composing music. We can't talk about a legal dataset if musicians and composers do not allow their works to be used.

Then, there's the debate of can AI-generated content be copyrighted.

There's probably a lot more that I haven't thought of, but if all this becomes possible, then we can talk about a potential replacement of human composers as a choice.

I still don't see this any time soon.

Sorry to disagree. We do not need AGI to happen before many composers are replaced. Already journalists have been laid off due to chatGPT.

And replacing a composer with a prompt monkey can and will happen. In fact it's already happened if you heard about the recent Red Lobster commercial.

Yes, music AI isn't that great ATM, but it's a baby baby infantile beginning. Just like original chess computers could not beat masters. But they did, eventually, and improvements are occurring very rapidly now.

I would absolutely love you to be right, don't get me wrong. But I see this veerrrrry differently.

 

[Sfpx]

No offense but the chess analogy is just terrible.
Chess is a competitive game between humans. It does not matter if a computer is the best at it, people want to compare themselves with other humans.

It's why A.I. and robotics don't endanger sports. Humans will always want to compete with other humans (and watch them).

People watch chess videos because of the human aspect in them, not really the chess aspect. The game can be bad yet the video still really enjoyable and entertaining.

Music is a product that you consume no matter what process was used to create it. If the music is good , it's good and people will listen to it and enjoy it. Unlike chess, you don't consume music for the human aspect but for the quality of the end product.

And I have not talked about the vastly different number of persons living from either chess or music. One is basically a hobby only , the other one is an industry.

 

[thorwald]

I would absolutely love you to be right, don't get me wrong. But I see this veerrrrry differently.

Fair enough. I think at this point we just have to agree that we see things differently. For our sake I hope I am right, but certainly no hard feelings.

 

[musicisum]

Unlike chess, you don't consume music for the human aspect but for the quality of the end product.

Allow me to politely disagree here, I think music in the end is all about human expression. In this regard, I think that from an ethical POV, samples are the same thing as AI: "not real live musicians". So AI will surely help getting some ideas faster, but in the end AI seems to always deliver quite uninspired / generic results. And although for some cheesy TV shows it might replace the whole generic music which is usually needed for that, I think for more performance-based music and orchestral stuff in general, people will always love to know the history behind the composer. There's something just super weird thinking "oh wow I love this music composed by a robot".

For me the chess analogy does make sense, since we do not care sooo much about the strategies involved and all of that, but we care about the challenges as humans that the players are facing: stress / career changing victories / or simply put: the struggle (and I repeat: as a human being).
All of those things are what make the game interesting - and the same applies to a composer too IMO. We love the struggle behind someone who did / composed a masterpiece. And if a robot did it in 2 minutes thanks to a great quality data it was trained with, well then this whole struggle thing disappears and makes it kind of soul-less. IMO.

 

[Henrik B. Jensen]

Perhaps we can compare AI in music production to the use of special FX in movies:

Most of us don’t mind at all if a movie contains a little trickery to make specific parts of it look good, for example explosions, as long as we perceive it to look realistic.

But in the modern Star Wars-movies some people think the whole CGI-thing has gone a little overboard, for example when we see an army containing thousands of soldiers/machines all looking exactly the same.

In contrast I think most people are happy with the level of CGI in the Lord of the Rings trilogy (?)

In conclusion, music in part created with AI might be acceptable to us as audience and as humans, as long as the final result fits it’s purpose and the AI generated component of the music doesn’t stick out. But if too much AI is used to generate the music, or it’s done completely with AI, perhaps we as audience will connect that with low budget.

 

[Ed]

Allow me to politely disagree here, I think music in the end is all about human expression. In this regard, I think that from an ethical POV, samples are the same thing as AI: "not real live musicians". So AI will surely help getting some ideas faster, but in the end AI seems to always deliver quite uninspired / generic results. And although for some cheesy TV shows it might replace the whole generic music which is usually needed for that, I think for more performance-based music and orchestral stuff in general, people will always love to know the history behind the composer. There's something just super weird thinking "oh wow I love this music composed by a robot".

For me the chess analogy does make sense, since we do not care sooo much about the strategies involved and all of that, but we care about the challenges as humans that the players are facing: stress / career changing victories / or simply put: the struggle (and I repeat: as a human being).
All of those things are what make the game interesting - and the same applies to a composer too IMO. We love the struggle behind someone who did / composed a masterpiece. And if a robot did it in 2 minutes thanks to a great quality data it was trained with, well then this whole struggle thing disappears and makes it kind of soul-less. IMO.

The Future:

There will be other creative expressions where humans are using the AI, we just can't see it yet. Like photography was literally said to not be art because it was just pointing a camera at something and not crafting it like a painting or drawing or sculpting.

Most of the good Udio songs actually took a lot of human input. Not only the idea, but finding the right prompt and generation to build from, custom lyrics (or at least working with the AI to craft it). I'm not saying it's equivalent to a real composition by a real composer, I'm just giving an example. There's going to be some kind of all new higher dimensional form of musical expression that uses human input. And the best examples will be the ones with a human at the helm, and until AI gets more god-like, humans will still be able to be a better "Director". There will be plenty of tools for composers to compose with, there just won't be any sense of scarcity anymore.

Sorry sample library companies, very soon you won't be used for the final product We'll be able to input humming, or whistling, or General MIDI, or just a MIDI file, and ask it to sound like a 1920's jazz record, or like John Williams ET score or whatever, and it will be able to render it for you. And it will sound completely real. It's the most amazing music technology ever, and everyone would absolutely love it if it wasn't for the ironic fact that it does it TOO well, and can write the music as well with minimal input. This is why there's so many composers and producers at various stages of denial in the grieving process. Some are denying it's very good, like Troels calling it a "mindless beat machine". To be fair he said this just before Udio came out, but it was still denialist even then. Then you have the anger stage. The pure anger stage is when you've accepted it is good and is a threat, but in practice the grieving process isn't usually a straight line, so you can go back and forth between anger and denial for some time. Anger expresses itself at outrage that it's using copyrighted works in the training data. I've also seen some in the bargaining stage as well, where they say they just want the training data to be cleared. Unfortunately none of these things are ever going to help stop this.

But leaving all that aside, as it's the future...

The Near Future:

Production music will soon lose most of it's value.

Those that survive will be those who can use AI to make more and better music than someone else because they know how to use it, are actually musical, can compose and produce, can use various parts of the AI and custom written music.

For some reason when people talk about this most people talk about "music" as if it's one thing.

Live musicians that play to a live audience, which includes livestreamed music, will be fine.

Those that write stock music for libraries. They will not be fine at all. The nature of production music is "functional music". It's made to be functional.

Those who write film "music to picture" will survive a lot longer (how much longer I don't know) Because writing to picture is a far more multidimentional craft than production music and we don't know how long it will take for an AI to be able to write to the nuance that human composers can. For sure the big names like Hans Zimmer will last even longer, because even if the AI do it as good as he can, Hans is a personality and this creates meaning in itself. By this I mean that the extent that the music simply "serves a purpose" and the film could have easily use library music except for the fact that it doesn't fit the picture and isn't uniquely crafted for the film, that determines how much a film is going to go with AI or not, or how safe a composer is from losing out to that job.

But production music, it really can't survive as we know it. Getty Images sued Stable Diffusion for using their content in their base models training data, but then created their own image model (ironically, presumably using Stable Diffusion) by training it on their entire catalogue. Do you think they asked the content creators if they wanted to opt in? Do you think they paid them? And if they did pay them, how much money would YOU want if Universal Music or something trained their own "Udio" on your music that was about to destroy your whole career and income source? (No legal arguments will ever work. Open Source will let you train on absolutely anything, and share the models, like with Stable Diffusion)