I have to admit I have not been impressed with AI music composition tools so far, but then I noticed this track shared on another forum…with a discussion on how these tools might make us composers redundant in the next 5-10 years. Should we be afraid, and are you?
Note: This entire composition was generated by an AI tool for music composition.
Oh man, this is a topic I could rant about for hours 
To try and keep things as short as I possibly can: fear it or not, it’s happening, so I’d much rather see composers on the RIGHT side of history, and embrace it EARLY.
Historically the music industry has been SO SLOW to adopt new technologies, to their own detriment. If in the '90s the industry had just ACCEPTED and EMBRACED mp3s and the internet, maybe they could have monetised downloadable mp3s right from the start, NORMALISING paying for downloads, rather than just pointing their finger and whinging for 10 years that music was being sold 
But the same thing is happening with AI…
I’ve seen composers in other forums saying “WE SHOULD UNIONISE AND BAN AI FROM MUSIC”
The idea of AI someday totally replacing composers is pretty scary, but I think also quite farfetched in the long run.
I’ve also seen people saying that “what if AI accidentally composes a John Williams piece - who’s going to get sued?!”
The concept of suing AI for ripping off other composers I also find questionable. I mean, what’s the difference between AI ripping off John Williams and a composer ripping off John Williams?
And finally, I immediately think of TWO times the industry buried its head in the sand against new technology: Napster, and Netflix.
They tried to litigate, shut down, whatever else, against Napster INSTEAD of trying to work out how to ADOPT or INTEGRATE it. And look where it ended up? Years of problems, years of suing, years of people getting music online the only way they could: illegally. Instead of setting up “official” digital download channels. It took FOUR YEARS before you could download music legally after Napster. I expect the industry would be very different had they embraced it.
And Netflix. I worked at Blockbusters when Netflix came out. And you know how Blockbusters dealt with this massive, new technology? They ignored it. Had Blockbuster embraced tv/movie streaming, maybe they wouldn’t have gone under.
So yeah, technology can appear scary, or that it’s “taking over”, but it’s going to happen whether we like it or not. We just have to work out how to best embrace it and make it work FOR us.
Good points overall, but I don’t really fully agree on the analogies of Napster and MP3 compared to AI software. I would argue this is completely different.
I do believe that we can indeed make this work for us in the short term. Long term, I have to say I believe AI will take over practically everything, creating a society like in the movie WALL-E. And yes, that I am in fact scared of personally.
Now talking about the optimistic view point: Can you (or anyone here) recommend good AI tools as plugins you can use inside your DAW? 
I’m not really sure what there is to fear? Self expression will always be superior in music. That means that it won’t matter if there’s AI that can do the job too. I mean we have AI for everything now and it’s increased work rather than destroyed it. Let’s just move with it and not see it as a threat.
If you don’t adapt then you get left behind. I mean, if Bach was around now he would do terribly… unless he learnt EDM 
Interesting topic. I’m not too sure about AI completely composing a piece. I agree with @Geoffers that you won’t completely remove the human, artistic element, but AI does have it’s use here. My piece for the Jurassic Park contest, while composed by myself, is performed by AI through using NotePerformer with Dorico Pro notation software.
I’ve said it before, while the technology isn’t perfect, I find it performs and sounds more realistic than me spending hours trying to automate parameters in a DAW with MIDI.
Hello there,
from a software perspective of view, i have to admit, that a software - how perfect it may be - , only can simulate. A software can and will never replace a human beeing, because a software cannot feel. There are no emotions, no expectations, no fear, and no hate. So the choice of everything will only be a random pick of from a bunch of information AI learned before.
You will never be able to send a reel to a computer and it comes back with a final score. For AI an mountain view would be scored similar to a love scene. A real reel composer could “feel” the difference, and give emotions to that scene.
I want to say something to that Napster and MP3 topic.
Remember the 80s and 90s when the first PCs and Homecomputers came out? Everyone was playing “copied” games. But the result was, that these games like boring “Pong” was played millions of times and it became famous. Now, these guys who played stolen versions of games, are the inventors of Steam and other Platforms, where you a) have to register and b) have to pay to win.
Always keep in mind, that you finally should “want” your things to be stolen. Thats the only way you can grow.
AI will be present in the next couple of years, and i think AI music will be placed on comercials or small low budget projects, but it will never replace real composers, a VST can never replace a real player.
Keep that in mind.
Michael
Based on this song you shared by AI, I think there should be some fear from working composers and future of how music will be selected by programs in general. its a great track that doesn’t sound like its from AI. Who came up with this concept and technology, a human at the end of the day, wanting to gain power, money and some control over a market- same as with Amazon, and other AI things taking over different work sectors- stores closing, companies bankrupt, everything being bought online- just like these companies were replaced so can other professions. I guess its up to us folks to choose whether to take this seriously and fight this competitive take over of an industry of creative minds by continuing to be innovative and outdoing AI with our own new ways of making and distributing music.
Well Michael, I wouldn’t say never. I’m a Trekkie, so I have to think that someday we’ll have Mr. Data-like androids who can call up and blend different styles of other composers and musicians to create their own style 
Currently no, but someday. For me, again I think if NotePerformer had slightly better sounding samples, like Spitfire, and Steinberg can perfect Dorico’s playback to more accurately read the composer’s performance markings on the score and build in some expression RR’s, it could be sooo realistic.
I will not say, that it is not possible - no. If you see the new Flight Simulator (really worth having a look at) you see what is possible with graphics, so it will be with music too.
But it will not replace humans, because machines will never be able to “decide”. They can only do somthing what is more or less better than the other. This is what i want to say. So every Director or Producer, wants to talk to somebody, what he likes or dislikes on a score. Possible could be, that AI gives you a basement of a score, and composers do the finishing touch. That could be a possible way …
That’s actually kind of a cool idea. It’s kind of similar to how Sonokinetic’s libraries work; they’re phrased-based and you can overlay and mix different phrases giving you a base to work with that you can then compose over. Not AI but kind of the idea.
lets build a small AI in a second
we take a key of C major, so the computer will only pick white keys. to keep it simple, we take full, half and quarter notes. This ist he setting of our AI. What will happen next. Usually a melody starts on the Base note. So C will be our first note. Hm – which length? usually a quarter note. Which note is next. Let me compare 2 Million Melodys. Probably a quater D. Okay - next note… Probably a half E. Oh wait, you said it shoul be heroic, okay, lets pic a half G. – First Bar Completed. Does it sound good? +1 for this melody – saved into the database. 60 Percent oft he composers would take now a full note of F, but in this case, we randomly pick the second best value of E. Does it sound good? -1 for this idea, saved into the database.
This is how software works. Every melody has to be judged by real humans. Computers can only provide Information but not value them.
AI works with really huge databases and with possibilities, but it can never be creative on its own, it can only be an Simulation.
Computers make decisions all the time - that’s literally what makes them computers, as opposed to plain storage devices. However, there’s no “intuition,” emotions or anything like that, so all they can do is approximate it through various combinations of logic, mathematical models, collected data, and randomization. If we can’t accurately explain the problem, the closest thing to a solution that computers can provide at this point, is an extrapolation based on data that we provide - and with insufficient data, that pretty much amounts to random garbage, so that’s not of much use without human supervision.
Given the great difficulties we as humans have already when it comes to creating a coherent, engaging, and enjoyable piece, especially in more intimate and emotional genres, let alone conveying the specific emotions intended, and that we don’t even have useful scientific definitions of what human emotions actually are, I think we’re still decades from seeing computers create anything of significance all on their own.
Realistically, what I’m hoping to see in my lifetime is an AI Composer Assistant that can actually create professional level arrangements and orchestrations, with proper articulations, good expression automation and all. Even if you’d still have to edit some parts to taste, it would save a lot of time, and allow for a lot more experimentation on the actual composition level. However, we don’t even have the virtual instruments necessary to properly render performances without endless manual tweaking and problem solving, so we still have some way to go, even for that…
If you think about it, human brains are just bio-computers and work relatively the same way, through logic, collected data and mathematical models. That’s how composition works essentially; recalling data learned and mathematical models (music theory), a bit of randomization (the artistic process). The key is the emotions. Can a mathematical model be created to describe those?
If you look at composing methods, there are already different models, so to speak that determine how to write for a specific emotion (slow/fast tempo, solo instruments, minor chord/major chords etc.) The key is in the individual human interpretation of all of it.
I wouldn’t say Logic is the right word. Perhaps instead of logic it’s actually probability.
Logic implies they make a judgement on what they think… where computers work primarily on variables making their logic much different to ours.
Totally agree that perhaps they’ll use phrase libraries. In fact I expect that phrase libraries will ultimately be the way the industry goes in the future. Right now people are against them as we like to control variables while also keeping our skill intact. I think this will soon dissipate due to how good phrase libraries will become.
The natural evolution of the library will slowly include more and more pre baked nuance in their sound I think.
I get what you mean, but my university degree was philosophy, so I did have to study logic so I do think of it in a much different way than most, that is deductive logic which is a more mathematical way of thinking and solving problems. It’s the basis of the Scientific Method and follows a formula such as, “If X and not Y then Z—X and not Y, therefore Z.” Or “X and Y, therefore not Z.”
Aristotle originally created the basis for this system, (his famous example is, "All men are mortal. Socrates is a man. Therefore Socrates is mortal.) and it has evolved to become a very computational and objective system. I’m not computer scientist, but I’m pretty sure probabilities and a form of deductive logic is how computers “think.”
I’ve had the lucky opportunity to see Honda’s Asimo humanoid AI robot in person twice. While it is of course programmed and responds to specific voice commands, it does have the ability to “learn”
Well anyway, I forgot where I was going with this 
but I don’t think there’s anything to fear about AI composer tools. Perhaps TV/game/commercial studios will use this technology because it’d be cheaper than hiring a real composer, but that may mean that us human composers would go back to doing more traditional concert music.
Sorry I’ve not been posting much both work and life have been a bit crazy recently
I think we’ll see more AI tools cropping up - we’re already getting AI assisted /machine learning mixing tools such as izotope’s ozone/neutron. Or things like soothe. .Algonaut’s Atlas is interesting as well as it uses AI to organise and present suggestions for drum samples.
I’ve played around with Orb composer with mixed results. I can see there being more AI assisted plugins cropping up in the future. musi co looks interesting https://www.musi-co.com/ generative music that reacts to various inputs - movement, code, gestures and other inputs.
Even spotify playlists use AI to generate suggestions or even running playlists that sync to your running pace if your device has an accelerator.
For games we’re already kind of getting to a hybrid point where clips or snippets can be blended on the fly depending on whats happening. I think in reality this is exciting rather than a total threat. It could give us new tools to add harmonic or rhythmic suggestions. Perhaps enable us to take paths we wouldn’t normally go.
I can see things like sample engines incorporating AI to change articulations and samples on the fly depending on whats being played etc.
Kind of why I prefer using notation software to DAW/MIDI if I can avoid it; I HATE having to put in all the key switches and stuffs to set the articulations. I just want to write and have the musicians perform knowing what articulations to use etc. That’s where AI comes in handy. 
Perhaps it is logic then 
thanks for the clarification matt 
(bows in reverence)
Any sufficiently advanced technology is indistinguishable from magic.
Even as a senior developer, I’m not entirely sure sometimes… But generally speaking, it’s all perfectly logical if you know how it works! I’m quite sure that applies to the human brain as well, but we’re still pretty far from fully understanding it.
Hahaha! Well, probabilities and logic. Similar bits and pieces. Yes, i love to put my Spock ears on when it comes to logic!