Capital-M Music

Essays

On Artificial Intelligence and Human Music

by Richard Bruner

Introduction June 2024: This essay was originally written in my Life Tips guide at the end of the Philosophy of Music Theory section, but as that guide has gotten excessively large and this set of “tips” only ever loosely fit in there, I’ve decided to pull them out on their own. These “tips” actually make a fairly coherent essay in their own right as a unit, though I have maintained the numbered system I used in the Life Tips guide. The final point in the first set (No. 5 in this document) also talks about what I’m now calling “Human Music” (because apparently that’s a phrase we’re going to need going forward…).

Then I’ve also duplicated Life Tips General Tips No. 10 below that, which deals with why we write papers for college classes and why we should keep writing our own papers and not have Generative AI text systems like ChatGPT write the rough drafts for us. That was the first of the “mini-essays” I put in that guide, and I thought it would be fun to write a full blown essay with analysis of quotes from others about writing essays (a “meta-essay” if you will!). Since then I’ve written several small essays in the Life Tips guide. I might pull some of the others out later too, but in this document I’m only including my writing on Generative AI.

Then I also wrote a little post on Facebook a few weeks ago related to AI separately from all of this writing, so I’ve included that at the bottom of this document as well. That was about my reaction to Apple’s iPad ad “Crush!”, which isn’t officially about AI but many people interpreted that in the context of Gen AI coming in and potentially (and actually) taking creative jobs away from people. Note that I use the phrase “Gen AI” mostly in this document, which is just short for Generative Artificial Intelligence.

The technology is evolving rapidly right now, and I may have a different take on some of these ideas in a year or two or three, but this is where I find myself as of May 2024. I think most of my points probably will hold up pretty well over time, because they aren’t mostly about current failings in technology but rather about the philosophical implications of tools that claim to do a lot of the “work” of art and creativity for you (or even claim to replace human artists entirely, especially in a commercial context), but we’ll see as time goes on. I will no doubt keep adding and updating some of these points in the future.

If you have any comments for me on any of this, you can reach me at my contact page here, or you can email me directly: richardbrunermusic “at” gmail “dot” com.

Update August 2024: I’ve added a section below at the bottom with links to articles (and a couple books) I’ve read over the past year or so about some of these topics. Some I reference in my essays, most I don’t but I still found them interesting. I also wrote another mini-essay to go along with the links. (Sidenote, but I wish Apple News had a way for me to save personal notes and highlights in articles like you can do in kindle books, and that there was a way to organize my saved articles so I could save some to reference later and some just to read later - that’s one reason I’m making this list below, so I can find articles I might reference later myself or point other people to.)

  1. Is AI Art Really Art? Here’s a potentially controversial point! Generative Artificial Intelligence (Generative AI or Gen AI) has come a long way in the past few years. It can now create passable “original” art in many different fields of art. But it isn’t clear to me that this is really art, or at least it isn’t clear to me that any human can lay claim to being an artist by using Generative AI tools.

    I saw something on Facebook recently from someone who said they hoped that in the future it would be seen as a kind of “ableism” to say that someone using a Generative AI tool to create something wasn’t an artist. I strongly disagree with this idea. I think there needs to be some nuance here, but to put it in extreme terms to start with:

    Someone typing a text prompt into a music Gen AI tool and saying “Make me a 3:22 hip-hop track with a high-energy beat in the style of ___ (Drake, etc.)” is literally doing the same job that music supervisor clients for my production music library company do. They send us briefs with some text, and even some examples of tracks they like, and then we look in our library and see if we have something to match that and send them several possible tracks. Hopefully they then pick one and license it from us, but at no point does anyone seriously think that they are composers because they prompted us to send them some (pre-composed) tracks.

    This is no different from typing in a text prompt into a Gen AI music system, and having the system create you 15 possible tracks matching your prompt (your brief), and then you pick one and go with it (or you send it back and say “I liked this one the best but it’s still not quite right, and here’s why…” We’ve had clients do similar things too). It may be true that no one else in the world has quite the same track that you do, but you no more composed it than our library clients did. Maybe a better way to think about this is that you were a client commissioning a track from a composer (the AI system) - you get a unique track but some(thing) else composed it.

    So where should the nuance come in? I’ve established to my satisfaction that full Gen AI that makes tracks of music from text prompts doesn’t make the prompter an artist (or at least doesn’t make them a composer), but I don’t think that rules out entirely using AI tools to assist in composing. To look at a version of this that isn’t AI to start with, I have issues with people using pre-existing loops and stringing them together and calling it composing, but I have less of a problem with doing that with percussive elements of a bigger track, for example. In many styles of popular music, the drum grooves are pretty similar, and you can either make your own loops for them, or use pre-existing loops. In many cases the result will be pretty much the same. You might put together two loops, or add a little element here or there to make it a little different, but I don’t think you should say you are not a composer for doing that as long as you don’t take someone’s pre-existing loop, use only that, and then say you composed it (doing that, by the way, would be considered copyright infringement, similar to plagiarism in writing prose). If you add other elements on top of that, then you can still say you composed something. Even if you only added several pre-existing loops together in a way that they weren’t before, you might still call that composing of a sort. The more melodic you get, the harder it is for me to accept you as a composer if you are using pre-existing melodies. At best you might be able to call that arranging depending on how you do it.

    This might also apply to newer systems such as “AI” chord generators. I still think it’s worth studying harmony and other music theory courses yourself and becoming familiar with how chord progressions work and what some of the common progressions are in whatever genre you are learning, but if you do something interesting with the chord progression yourself then you can probably still call yourself a composer. I don’t think that using the “Lament Bass” chord progression should necessarily mark you as “not a composer” just because you didn’t come up with the progression from scratch. It might be a cliché, but it can still be effective. Using AI to generate progressions that you then work with similarly could still be viable. Somewhere between there and full blown Gen AI prompt systems is where I draw the line. If you don’t need to have any formal training (whether from a teacher, or books, videos, etc), or even any time spent messing with making music yourself using an instrument or recording yourself in some form, then I’m not convinced that you can call yourself a composer, or say that you are exhibiting your own musical self-expression.

    One other thing that irks me on a purely commercial basis is that it’s not clear to me how to stop people from using Gen AI systems like this, taking the tracks as output, registering them with BMI or ASCAP and collecting writers royalties for those tracks without ever putting in any work towards actually becoming a composer. Of course, some people try to do that already - I’ve heard of production companies insisting on taking part of the royalties as a condition for licensing a track from a library or artist (which isn’t supposed to be allowed, but again, it’s hard to stop it if people are willing to give it to them). You will often see pop songs in particular that have 5+ writers listed, including people who probably didn’t have any actual writing role in the project. People used to talk about someone “adding the tambourine part” to get writer credit, but nowadays there’s often not even the real pretense that someone wrote something. But AI allows this at scale, and while you still have to get usage to generate royalties, the mere fact that there might not be anyone actually involved in really writing a track that gets paid for it rubs me the wrong way even more than all of those other versions of this. Someone at the production company can use these systems, and not only not have to pay a composer or library for the track, but actually pocket our royalties instead, which doesn’t seem very fair to me!

  2. (AI Continued): Dr. John Mortensen has a video out recently where he says:

    “Music is not really a commodity that we want to create through labor-saving devices. Now, some people do just want to generate garbage through AI and put it in the background of a video and not lift a finger, but most people who are really interested in music don’t want too much labor-saving technology. They want to do the work themselves. They want to understand and do the work by hand rather than having it taken away from them…

    What does it mean to be a human, and to be alive in the world and working? How do we find meaning in this? It’s not by surrendering the learning and the work of music to labor saving technology, it’s by engaging in it ourselves and getting dirt under our fingernails and breaking a sweat. It’s actually really fun and it gives meaning…It’s a way of being human, and a way of being in the world.” (5:48 in the video, cued in link)

    I agree with everything he says in this quote, but I’m not sure it was the musicians who were asking for the labor-saving technology. He gives a hint himself here - it might be the people who paid us previously who might want to not pay us anymore, especially if they could take some of the money that we were collecting before… I hope that music remains viable, and I certainly think it will continue in some form as a human pursuit, but it might just get ever more difficult to make a living with it. It’s never been easy, but they certainly aren’t trying to make it easier! I actually don’t think that money is the main reason to do music, and as I mentioned in the introduction to [Capital-M] Music, I will continue to make music no matter what happens because it’s simply who I am. But in our hyper-capitalist society, it’s easier to convince many people that they should continue to support the arts if it seems like you could make money with it, and it’s certainly nicer to get paid for the hours of work that you do than to not get paid for that! You can’t put as many hours in if they are unpaid, because you still need money from somewhere to keep a roof over your head and keep food on the table, which are also nice things for human survival :-).

    I just recently saw an ad on Facebook for a synthesizer in Reason Studio that generates bass lines (screenshot from April 21, 2024), and the ad copy said “Reason Studios analyzed hundreds of real-world basslines to reverse engineer how they work so well and how they can work for your next track - so you don’t have to” (emphasis mine). They must have missed the point that I actually want to - that wasn’t a problem to be solved! I think there probably are people who think that sounds great, but that’s actually an immediate reason for me not to get such a product, as I mostly want to make my own music, at least when I’m using computer synthesizers. I don’t need it to sound like something no one’s ever heard before, but I don’t want to use tools specifically designed to make me sound like everyone else, or tools designed to do the “work” of composing for me. This is part of what I like about composing, but that’s literally the way that these Gen AI tools work - they analyze existing material, statistically average it, then generate output based on that, possibly tweaked based on some additional parameters you give them. But when I make music, I want it to be the music I wanted to make, not the music that a system thought might be a useful average of everything else.

    Some people in the commercial music world do have a very mercenary attitude about it, doing whatever they need to do to maximize profits at whatever cost to their artistic (and sometimes personal) integrity. Other people are “artists” to such an extent that the idea that someone even could make money with music seems foreign to them, and they do want their music to sound as unique as possible. I’m in the middle between those - I am fine with doing creative work for profit (I do that myself), but I want anything I do in that regard to be my own work, even if it’s based in some cases on my own internal version of averaging some things that already exist. I would actually think of this as becoming familiar with the style of music in question and learning what makes it work, so you can use characteristic elements of the style in your own projects, and push the boundaries a bit in various ways to make it your own. That can be subtle or dramatic depending on the project you are doing and your personal taste, but you have to know the norms for the style yourself before you can do that. Another way to make it your own would be to combine elements from multiple genres that you are familiar with, which again, to do that you’d need to know them yourself. This is one reason to think about Multi-Genreism as I outline in the main Capital-M Music paper. Also see the final paragraph of Life Tips Composition Tip No. 28 for an analysis of one group I think did that well. I’d like to think that I bring enough of myself to any projects I do and they don’t just sound like everything else, but at least I feel like I came up with it in any particular project and I didn’t have a computer generate output that I then tweaked lightly and passed off as my own “original” work. There is too little artistic integrity in such an approach for me, which is something of the point of this “tip”. See also General Tips Tip No. 10 (also copied below) for more on this idea in the world of writing papers for school, something I’m doing again, where some similar ideas apply.

    Some people go so far as to talk about AI auto-generating content for you as a consumer based on either what you tell it you want now, or what it thinks you want (for example from a service like Spotify), but I don’t know that I want that experience as a consumer of media, let alone as a creator of it. I think it would seem cool for a minute, maybe even a week, but once the novelty wears off there’s nothing of interest in music that is tailored precisely to what you like or want all the time. The interest in music comes from the way that an artist subverts your expectations. That can go too far, and music that doesn’t satisfy your expectations sufficiently doesn’t establish any patterns which they can then “artfully” break, but if you could always predict with certainty what the next thing would be, the music would be boring (see Composition Tip No. 24 for further discussion of this point). There is a potential commercial use for music like that, sometimes referred to (often derisively) as “sonic wallpaper”, but it’s fairly limited and is not really that interesting either to make or to listen to in most contexts. It takes the concept of “functional music” to the extreme, and while you might be able to make some money with it in the short term, you are probably going to drive yourself nuts after writing it for just a short time and want to go back to something more satisfying. This is one difference between the commercial and art music worlds, but even in the commercial space it’s nice to have something more intellectually and artistically satisfying. This also is an area that AI probably will take over in the near-ish future, and I’m not sure that we should be too sorry to see that go, unless you were making a living writing music in that manner.

  3. Platonic vs. Aristotelian Intelligence (Re: Gen AI): I just picked up a copy of Dennis Yi Tenen’s book Literary Theory for Robots, and in this book one thing he discusses is two forms of intelligence, which he terms “Platonic” and “Aristotelian” Intelligence (Tenen 34). Platonic intelligence is “inward” facing - this is the kind where an individual knows what they are doing specifically, and how to do things themselves. Aristotelian intelligence is “universal”, and all that matters is the effect on the world for this kind of intelligence, not how something is achieved, so tools can themselves be said to have a kind of intelligence, and collective knowledge exists in this space as well. The kind of intelligence I think about for someone to call themselves a composer is in many ways more Platonic by this definition.

    It’s Aristotelian in the sense that you learn about what has come before, which is a kind of collective knowledge. Music theory is based on analyzing lots of pieces from a specific period or place to see what composers were doing and find commonalities between how different composers worked, especially the theory of the “Common Practice” period in classical music history or other similar periods in other genres. The Classic Era theory concept of the “Schema” as proposed by Dr. Robert Gjerdingen in his book “Music in the Galant Style” is one such concept, which was developed through the teaching tool of Partimento in Naples during the 17th and 18th centuries (and spread from there to the rest of Europe). In this system, they taught people figured bass lines and voice leading patterns to use to write, improvise and perform music, and they very consciously used them in much the way that we do in modern popular music styles with common chord progressions. In this sense, the teaching of music uses Aristotelian principles of collective knowledge (but also Platonic principles of having an individual specifically learn this information themselves - no matter how many other people before you have learned not to use parallel 5ths or octaves, you have to learn it yourself as well).

    The tools we have available today can also help you achieve things sooner and more fully than you otherwise might be able to do them, another Aristotelian principle. But I think it should matter that the individual credited with creating something actually had some idea what they were doing for the raw mechanics of creating that thing. As Tenen says in Literary Theory for Robots, “The instrumental goal of calculation or composition stands secondary to the intrinsic value of personal growth, achieved in the struggle of learning. The hand carries the load of value through lived experience. And experience cannot be automated” (Tenen 47). I have less of a problem with people using tools to help with something if they could have done that thing without the tool. Obviously, it might take them longer or be more tedious to do that thing without the tool, and they might even have to look up something they’ve forgotten because of how long it has been since they last did it by hand, but for me the problem starts when tools replace learning the skills themselves.

    I mentioned above that I would in principle be ok with people using AI chord generators to come up with chord progressions, though I still recommended that people take harmony classes and learn theory. I’m going to amend what I said to insist more that people really should take the classes (or read the books or watch the videos - however you choose to learn the material is up to you, but you should learn the material yourself somehow). Once you can do it yourself, then potentially using such a tool can make it faster to complete work, or even inspire you in ways you wouldn’t have been inspired otherwise (“AI” random line or sound generators, for example). But you will be in a better position to judge the quality of the output, and to tweak it to better suit your needs (you don’t have to use it as given) if you know something about how the concepts work yourself; and like I’ve said elsewhere in this set of guides, you should be in control of your tools, the tools shouldn’t be “controlling” you (see Musicianship in the main paper for more on this topic).

    This also applies to things like notation playback / sample libraries as I’ve discussed under Life Tips Composition Tips Nos. 1+2 - they are useful as long as you already have some idea of what the music should sound like in real life, but if you try to learn what it would sound like from those tools, they will mislead you if you aren’t aware how to mentally compensate for their output, or how to tweak the playback to make it sound more realistic than the default. And there are many kinds of music you could write where the playback wouldn’t really work at all - you have to know what you want and have some idea yourself what it will sound like, or you can’t write music like that (and you probably wouldn’t even think to write music like that). This applies particularly to many modern “extended techniques”, but there are even some relatively common techniques that playback doesn’t handle very well.

    For that matter, this is why we still had to take math tests without the calculator for parts of the test when I was in high school and earlier (not sure if they still do that today, but they should!). This is to make sure that you have some knowledge of what the machine is doing, so you can judge the results to be sure they make sense. Maybe you don’t need to consult log tables anymore, or use slide rules as you once did, but you should still know what the calculator is doing.

    I also got into flight simulation fairly seriously with the release of Microsoft Flight Simulator 2020 during the pandemic, and started watching a lot of real-life flight videos on youtube as well as flight sim videos. One of the points that they stress frequently in the real life flight videos is that autopilot exists to make your life as a pilot easier, it does not exist to replace your manual skills as a pilot. They say you should always know why the autopilot is behaving the way that it is, and it should be doing exactly what you would do by hand in that scenario. If you ever have any doubts, even for a second, as to why the autopilot is behaving the way that it is, you should immediately shut it off and fly by hand, and you should never fly with an autopilot in conditions that you wouldn’t be comfortable hand-flying the plane. In this world, life and death are literally in the balance (potentially for multiple people), so it’s a fairly extreme example; but I think the concept of tools making your life easier and not replacing your underlying skill set probably applies to most fields, and certainly applies to music, even if we don’t have to risk death generally to make music (thankfully!).

    Also, it’s helpful not to be completely reliant on any technological tool to the degree that’s practical - you should still be able to operate at least a little bit if you don’t have your gear with you. As I mentioned in the Capital-M Music paper (see Musicianship), I can still make music if the power goes out! I can even make music of a sort if I don’t have my instruments with me - I can sing, and there are other ways to do body music 🙂. It would take a lot to stop me from making some kind of music, though I’ll grant that it’s hard to play violin concertos without a violin.

    Remember - you are the musician, your instruments are tools to help you make music (see also General Tips No. 4). To quote the great Tony Stark: “You can take away my house, all my tricks and toys; one thing you can’t take away - I am Iron Man!” (Iron Man 3, ending).

    [I will add here that even though I say that your instruments are tools to help you make music, many musicians develop deep relationships to their instruments. I have many friends who have told me the “names” of their instruments, and I definitely think of mine as at least several notches above the way I think of my power drill or hammer, or even in many ways my computer. When you spend that many hours getting that good at coordinating a complex machine or instrument, it is a different experience. Hours spent struggling to achieve the vision (or is that sound?) you have in mind, hours spent having some of the best experiences of your life when you can let go of all the work and just play and experience being in the moment like little else can do. It does sometimes feel like you become a conduit for something else to work through you and your instrument to create something special. But all of it leads to a deep connection to the instruments you use.

    My violin and I have been through a lot of life together at this point - I’ve had it for nearly 20 years, from sophomore year of high school, and it’s been with me through the rest of high school, my entire college experience, the entire time I’ve been in Los Angeles (it was the first instrument I brought with me the day I came out here, along with my tin whistle collection at the time), and even came with me when I went to Ireland several years ago (see the cover photo on my homepage!). Between my violin and now my viola as well, they have been there for some of my most exciting moments, and helped me survive and recover from some of the darkest moments as well. So while they are tools to help me make music, maybe they aren’t just tools.]

  4. By the way, the fact that it takes thousands of hours of hard, focused practice to become a good performing musician, or dozens to hundreds of pieces (again, thousands of hours of work) to get to be a good composer isn’t a problem or some inefficiency to “solve”. It’s part of the point. Many musicians like the process most of the time - if we didn’t, we’d quit! (And I also know many who have…). Even when it’s frustrating and you’re beating your head against a wall because you just can’t get this one passage to flow smoothly, or you can’t find the right way to modulate between sections and you’ve just deleted the third attempt at your transition section in a piece (or your 43rd attempt!) you are aware enough of the overall picture to persevere and the reward at the end makes all the toil worth it. That reward is at least as much “soul-satisfaction” as it is financial - you don’t need to be a paid professional to experience it, and if money is really primarily what you’re after, you probably won’t survive in this field long enough to get there. Music can’t just be about a job to pay the bills, it needs to be something burning inside you to put up with all the other nonsense that goes with it to make it worth it (although I will point out that “all the other nonsense” goes with every other job too, and I’ve never met anyone who’s done any job who didn’t have stories to tell…). I like to joke that you have to be at least a little crazy or weird to even try to make it in a field like music, but I wouldn’t have it any other way! I know I can be at least a little weird sometimes, and so can all my other friends in the industry, but they’re “my people” and I love that about us! I like nothing better than to spend hours making music with people and talking shop with other composers or performers. To be clear, I don’t mean any of this in a bad way - I think the world would be a far more boring and much less interesting place if we were all the same “normal” people. And frankly, I’m not sure there actually is such a thing as a “normal person” in real life anyway, maybe we’re all just weird in different ways 😃.

    Also, If it really were easy to make it in music then we wouldn’t value these skills the way we do (culturally if not always economically!). It’s not elitist to say that someone who puts in the necessary time and energy should be recognized as being better at something than someone who doesn’t (or if it is, then that kind of elitism is good). That shouldn’t devalue the other person as a person, but if you can’t or won’t put in the work, then you shouldn’t get the reward, either social or economic. I’ve spent decades getting where I am today, and so have the other people I know who are proficient at music. I’ve had a lot of fun along the way, and I deeply love making music in a variety of forms, for myself and especially with other people, hence some of the ideas of Capital-M Music! This is a process that never ends, and the journey really is the point. You never actually “win” at music, even if you do win some competitions. My Berklee String Improv professor Eugene Friesen says in his book Improvisation for Classical Musicians, “This is not about ‘accomplishing’. This is about internalizing. It takes time, curiosity, and of course, love for music!” (Friesen 32). He was specifically speaking about learning to hear and feel scale degrees by using drone practice (see Ex. 1 in the Musicianship part of my main paper for a similar exercise), but I think this concept applies to many aspects of learning music. But it does take a lot of work to get to a high degree of proficiency.

  5. Holding on the idea of human music for a minute, I remember reading Jeff Rona’s book The Reel World (now in its 3rd edition, though I’ve only read the first two) when I first got interested in film music as a field when I was in high school. That was my favorite of the film scoring books I read back then, and in the section about studio technology there was a chapter called “Sounds Just Like The Real Thing, Until You Listen to the Real Thing”. This was about how to make mockups that you’d present to a director to get approval for a cue before the recording session with the live orchestra, so that if there were any issues, the director or producer could address them before the session so that you weren’t on the podium trying to rewrite the cue with a $10,000 overtime bill hanging over your head. There are always issues - no one expects you to get the cue completely right the first time. But one point of this section was that however realistic you could make your mockup, it always paled in comparison to hearing a cue with live players (or hearing any kind of piece with live players). This was just talking about going from midi mockup to live recording, but it’s even a big difference to go from a recording of human players to hearing a live performance in person, in the same room as the players.

    Every time I go to a live concert, I’m reminded that however good our systems are now (and they’ve improved substantially even since I read that section of Rona’s book about 15 years ago in the late 2000s), live performances are still better, and maybe always will be. There’s just something about being in a room with people doing this that makes it more exciting and visceral. Some of it also just can’t be reproduced on a recording very well, especially with a normal playback system. Super deep sub-bass pipe organs (32’ pipes) or even big bass drums make the air throb when you’re in the room with them in a way that your playback system is unlikely to replicate. Just recently I was in the organ practice room at CSUN with a small pipe organ that had a 16’ bass, and even that pipe made the air shake in the small room we were in in a way I don’t get to experience that often. Bass drums and other big bass instruments also generate sound that you feel in your body as much as you hear with your ears. Earbuds simply cannot reproduce that experience physically (they don’t vibrate the air around your body, only in your ear canal), and even big mid- or far-field speakers couldn’t do that quite the same way. If you want to try this, listen to the initial Dies Irae movement of Verdi’s Requiem - I’ve never really found a recording that reproduces those epic bass drum hits correctly, as it’s supposed to punch you in the gut as much as be a big low thud. They can sound pretty cool, but it’s not really the intended effect (the linked recording does sound good, and is actually one of the best I’ve heard on youtube or off, but depending on your playback system it will sound better or worse, and it still won’t sound (or feel) like it would in a hall). Remember that Dies Irae is latin for “Day of Wrath” (Judgment Day) - there’s nothing subtle about this piece of music!

    Even in situations that could in theory be reproduced, playing or listening live is still a better experience. At our most recent Scottish Fiddlers of Los Angeles show in April 2024, we got to play to a full house with Alasdair Fraser and Natalie Haas, one of the most well-known acts in Scottish music in the past couple of decades, and it was one of the best performance experiences of my life to date. I think we played better than we usually do, but the energy and groove that Alasdair and Natalie brought were just intense, and both we and the audience loved it! I’ve been a fan of theirs for several years now, and getting to sit 10 feet away from two of my fiddle-world heroes and watch and listen to them play, let alone to play with them myself, was an experience I am not likely to forget any time soon. I was getting concerned that I was overhyping this concert to myself, being super-excited about it for about six months before it happened, but it lived up to (my) hype! I can’t really have this experience with their recordings, as much as I like them, and computer-based music will never be able to rise to that level or give me an experience even of the same type, let alone the same quality.

    Audience feedback is also a crucial aspect of this - one reason that playing in a live concert is such a different experience than producing a studio recording is that the audience can give you real-time feedback and the players will play off of that to a large degree. There’s perhaps less direct feedback in orchestral concerts - usually the audience is supposed to sit there quietly and only make noise at certain (pre-determined) times, clapping at the end of a piece (and not even the end of a movement, though even the “best” audiences sometimes get caught by that - I’ve never played the Tchaikovsky Violin Concerto and not had the audience clap after the first movement, and given the nature of the piece I think that’s appropriate and it would be weird for them not to clap there). Even in a classical context though, the audience does give subtle feedback, and playing in a concert feels different from a run-through at a rehearsal with no audience. One other aspect of this phenomenon is the awareness that you have when you are playing that you can’t “redo” anything - in a rehearsal, even if you are attempting a run-through you can always stop and regroup if necessary, but in a concert setting that’s not a realistic option. Technically you can stop and start again, but the audience will be aware of it and we do our absolute best to not stop in the middle of a concert, so the adrenalin of being “live” itself fuels you as a player in any live performance, classical or otherwise. In other contexts with more direct audience feedback and even participation, you can really feed off the energy from the audience and take your playing to new heights because of that. Recording sessions usually don’t elicit the same feedback as there’s usually not much of an audience present and it’s a more technical setting, and computer midi production is most often done by yourself with no one else present and there’s no feedback from an audience other than what you yourself feel from moment to moment. That’s not nothing, but it’s not at all the same kind of experience, and that’s one of many reasons why I love playing with other people and for other people so much!

(Life Tips General Tips No. 10): On Writing Papers

Sometime between my undergraduate degree at Berklee and my graduate work at CSUN, I figured out why we write papers for school, and it has made the process of writing for grad school more enjoyable for me than it ever was before. I really thought about this in detail with the rise of ChatGPT and similar Generative AI systems, and why we should or shouldn’t allow them to write papers for us. I would never use one of those systems to write a rough draft of a paper for me today, and there are at least two reasons why that is (you know, besides the whole “cheating” thing, and if you get caught you’ll fail the assignment, possibly the class, and get written up for “academic dishonesty”). The first is that they make things up as a matter of course (“hallucination” is the word that AI researchers have adopted for that phenomenon), so that you can’t trust anything they say at this point in time. In the time that you take to fact-check it you could have written a better paper yourself anyway.

But the real reason why I wouldn’t want to use one to write my papers, a reason that advancements in AI technology can’t solve, is that the rough draft is where you actually do the thinking and learning. Especially at the undergrad level, but really with any kind of research writing (or even a lot of reflection writing), the point of a paper should be to think through what you’ve learned and integrate it with everything else you know in a tighter way, so that hopefully some of it sticks with you later. The rough draft is where you do that part of the work, so that you miss both the benefit and the entire point of writing a paper if you allow ChatGPT to write the rough draft for you, even if you then edit it and do all the fact-checking. Really at that point all you have is another source that you can read, which may or may not be correct, and while you might learn more from that source by doing detailed fact-checking than you would get from many sources just reading them, you haven’t done the actual work and gotten the actual benefit that doing research writing is supposed to lead you to.

Ezra Klein from the New York Times recently did an episode of his podcast The Ezra Klein Show with Nilay Patel from The Verge where they were talking about this point from the perspective of writing articles, which is a similar process to writing research papers in many ways. On the show, Klein said:

“…[O]ne way I think that A.I. could actually not make us more productive, more innovative, is that a lot of the innovation, a lot of the big insights happen when we’re doing the hard thing, when we’re sitting there trying to figure out the first draft, or learn about a thing, or figure out what we’re doing.

One of the messages of the medium of A.I. is ‘Be efficient. Don’t waste your time on all this. Just tell the system what to do and do it’. But there’s a reason I don’t have interns write my first draft for me…

They could do it. But you don’t get great ideas, or at least not as many of them, editing a piece of work as you do reporting it out, doing the research, writing the first draft. That’s where you do the thinking. And I do think A.I. is built to kind of devalue that whole area of thinking.” [1:02:30 in the episode]

This is why you should take writing papers seriously in school, even when it’s frustrating and you’re spending 30 minutes finding the page number for the quote that you know word for word, but you have to cite it (this particular issue may be less of a problem today than when I was in school earlier as search functions for pdfs and ebooks work better - we didn’t have as much of that back when I was at Berklee and before, but I’m still using print books for some of my sources today in my grad program). Integrating ideas from different people is another skill that paper-writing is developing. One of the things I’ve been working on in my graduate writing is trying to integrate ideas from different people in defense of my argument, rather than just verbatim repeating people from their own arguments. I do it all the time when I think to myself, but for some reason I have a harder time with it in a formal paper context. Building research skills is a useful part of paper-writing, and this applies to many jobs outside of academia as well as in school. I’ve even done research in my day jobs with production music library operations, so just being able to think holistically is an important ability.

I sent a version of this “mini-essay” tip to Joanna Parypinski (horror author Jo Kaplan), a friend of mine who is also an English / Creative Writing professor at Glendale Community College in Glendale, CA, and in her email back to me she also pointed out:

“Writing is absolutely part of the “thinking” process, and drafting (contrary to what some students may think) is not just the creation of a finished product but the process of working through ideas on the page. You know, I often have students say their biggest issue with writing is just taking their thoughts and writing them down. I imagine it’s because they haven’t had much practice or instruction on writing as a process rather than a product.”

The idea of using a system like ChatGPT to do your papers shows that you have a “writing as a product” mindset, and if that’s the case you may find it makes papers easier to tolerate and more interesting to research and write to change to a “writing as a process” mindset, or a “writing as thinking” mindset. You can have an AI system churn out a paper for you, at least in rough form, and get a product that you could in theory turn in (though again, if they catch you doing that there could be academic consequences and they could be severe depending on your institution’s policies, so it could be dangerous). But as I mentioned earlier, you miss all the benefit and learning that would theoretically be the reason why you are taking a class to begin with. And even if the real reason you are taking a class is because it’s required for your degree, one would hope that such required classes would have a point that will help you down the road, whether you can see how in the present moment or not. I can look back on my educational experience now several years and the early part of a career later and see why they had me do certain things the way they did, even if at the time they seemed pointless or counterproductive. Not everything is perfect (most things aren’t!) but there really is a reason for most of it.

This idea of “writing as a process” or “writing as thinking” is actually a significant part of why I’m writing my Capital-M Music project (and I promise you that AI did not write a single word of any of this to date, and I’m not planning on using AI for any of it going forward either!). I hope that anyone who reads all this gets some new ideas from it or considers things from a different perspective than they did before. But I’m actually writing it all as much for myself as for anyone else. It is helping me to think through the way I see music and even life in a much more detailed and focused manner than I have previously, and I have already been able to use some of these ideas to help the students I’m tutoring at CSUN this year. As I’ve mentioned elsewhere, I’ve been thinking about some of these concepts for several years now under the name “(Capital-M) Music”, but writing it down has forced me to make it a bit more coherent than it was in my mind, even after spending hundreds of hours thinking about all of this on walks by myself, or in conversation with other people at random times when these ideas would come up.

On Apple Ad “Crush!” [From Facebook post on May 10, 2024]

So about that Apple ad everyone has been talking about the past couple of days… It’s interesting timing for me, as I’ve been working on a personal essay reflecting on my first year of grad school which I’m planning to post here sometime after next week once our semester is actually done. I’ve chosen to structure it around three keywords (I call them “watchwords”) that I keep coming back to this year, and the main one I spend time on in that essay is “Human”. I’ll save most of that for that essay, but I will say that when I first saw people complaining about this ad a couple of days ago, I could see the point people were making intellectually but I thought people were overreacting. So I decided to watch it last night to see what they were actually complaining about and I found it harder to watch than I thought I would. Having seen it now I’m much more fully on the side of people who don’t like the ad. The ad is linked in the article I share below, or you can find it on Youtube - it’s an Apple iPad ad called “Crush!”

If you haven’t seen it, they show a whole bunch of physical objects that people have used over the centuries for various art and culture activities, like musical instruments, paint buckets, and even video game consoles, and then they crush them all with a hydraulic press and in the end, all that is left is a very thin iPad (“the thinnest ever!”). The point is supposed to be that all these things that people used to have as separate objects can now be done with a single very thin iPad, which even itself is only kind of true, but many people took it as symbolic of big tech companies crushing culture and human creativity. (Also, that ending with the little creature getting squished was much more creepy than fun…).

As someone who is currently working on an advanced degree in a highly creative field, and who has spent his whole life in some version of this field (and who has also been very techie for my whole life as well, as I get into in the other essay), I have been thinking a lot recently about the coming AI future, if in fact it will really exist. I’ve written about this in a couple of places on my website recently, as well as in this reflection essay I’ll post soon. At the moment, those posts on my website are under richardbrunermusic.com/life-tips, then go to “Philosophy of Music Theory” Tips Nos. 9-12 [See above!], and “General Tips” Tip No. 10 (Also see above) (that one is about why we should still write even the rough drafts of essays for college classes ourselves even with the existence of ChatGPT etc.). And for anyone who looked at my website when I posted earlier this year about my Capital-M Music project, the Life Tips guide in particular has been expanded dramatically since then.

But one of my thoughts in reaction to this ad is based off of something I’ve thought about before in a different context over the last few years. One of my topics I was thinking about several years ago was “pros and cons of print books vs. ebooks”, and I came up with several practical reasons why ebooks were better than print books, mostly having to do with how much room ebooks (don’t) take compared to print books, and also generally being cheaper. But one of the big points I had in favor of print books was that no one ever talked about how nice the ambience of a room filled with a single kindle or iPad was, but they did talk about the ambience of a room filled with books! Watching all these objects get crushed down, to be replaced only with a single shiny iPad reminded me of that notion. Maybe you can do many of those things on an iPad now, but sometimes it’s just nicer to be in a room with real objects that you can really interact with in a visceral, physical (human) way that you can’t really do as readily with virtual objects in a touch screen. I do use a digital piano at home for my piano (with the physical modeling synthesizer Pianoteq as my sound source so it feels much more acoustic than your typical keyboard piano sound), but it’s still a very physical object in my room, and even if you use an iPad for that sort of thing, you still need a physical keyboard to play it from. Have you ever tried playing Clair de Lune with a touchscreen keyboard?

My favorite article that I’ve read in the last couple of days about this ad is one of the articles from The Verge, “Apple Doesn’t Understand Why You Use Technology”. They had a couple of other good articles too about the reaction to this ad. On this week’s episode of The Vergecast (The Verge’s main podcast) that came out this morning, they also have an extended discussion of the ad (cued in link), where they point out that the reaction is really about something much deeper than this specific ad - it’s really a flashpoint in the ongoing angst over where we might be headed with generative AI (for evidence of said angst, see the 17 pages of single spaced text I have above this post on this page!). I think this is probably true, but that doesn’t change the fact that we’re having this ongoing conversation about the future of a lot of our lives, which this ad does hit on in a way I’m not sure Apple was really thinking about when they released it (though they can’t be unaware of the feelings that many people have about this).

Initial posting August 31, 2024

Here are some links to articles that I’ve found that examine Gen AI from a similar philosophical lens to what I’ve described in my writing above. You can think of this as a Further Reading list if you want to go even deeper than I have above. There are a lot of people talking about this today, from both skeptical and evangelical viewpoints (as in “tech-evangelists” - tech supporters, not right-wing Christians!). You can tell I’m on the skeptical side, and most of these articles will be as well. I fully admit that I’m biased, both in my outlook on life as someone who has dedicated his life to the arts and in the way I see the current hype cycle around this technology and what it’s doing to our society today, and I’m not trying to hide the way I feel (as you no doubt noticed above!).

So this is not aiming to be a balanced list, though if I run across articles from the tech-evangelist side that seem to be reasonable and make sense to me, I might include them as well. But I do think that each of us as people need to establish our own values and have principles to believe in. The importance of (human) thinking and philosophizing, education, and expressing ourselves through the arts and particularly through music are some of the values I hold dear personally, and some of them are under attack today in many arenas, including Gen AI and the tech world more generally.

As I mention in my reflection essay Watchwords: Reflections on a Year of Graduate School, I have also been a child of the computer and internet age, and I’ve considered myself a technologist as well as a creative artist. Until recently, I could at least pretend that there wasn’t really a conflict between those aspects of who I am. I am well aware that the way I make music has been heavily influenced by technology, that there were ways of doing many of the things I do before computers came along, and that some people may have been (definitely were) displaced as the techniques they had trained in were supplanted by extremely different methods of achieving a similar end (hand copying parts vs notation software, for example, or the rise of personal PCs rendering recording technology much more accessible both in cost and ease of use compared to the studios of old, which displaced some of those institutions and the people who worked in them). But I came up on the modern side of those changes, and with all the other significant twists in technology in the 34 years of my life to date, this has been the first major change that has scared me. 

I’m not going to give up my computer studio for making music (that’s not a practical solution to any of these issues at least insomuch as it’s not even a solution for whatever “problems” we’re going to have in the future anyway). But I’ve found in the recent past that when push comes to shove, if you force me to “pick sides” in most debates, I’m siding with the artists and not the technologists (or businessmen, or anyone else who may be pitted against the artists). I hope that there will be ways to thread these needles so that I don’t have to pick sides entirely and jettison parts of who I am, but at the end of the day, I am a musician first, an artist second, and anything else after that.

On a more practical note in regards to this list, I do much of my news/article reading through Apple News, and I’ve chosen to pay for Apple News+ to have full access to many of my favorite publications on the platform, so some of these links will be to Apple News+ articles. I’ll try to indicate when that’s the case, and when possible to also link to the website directly, but most of those publications will have paywalls on their direct websites as well. These would tend to include articles from The Atlantic, New Yorker, New York Magazine, and Slate among the sources I read frequently on Apple News. I’ve generally liked the coverage by The Verge as well, which is not currently paywalled, so there will be several articles from them here as well. I believe that if you don’t have Apple devices and therefore don’t have Apple News, the Apple News links should take you to the articles online anyway. Not sure if that’s true about Apple News+ links.

I will update this list as I run across more articles. 

Finally, I will also note that many of these articles will be aimed at ChatGPT style Large Language Models (LLMs) and Dall-E style image generators more than music tools as both of those use cases are more mainstream (and tend to be more relevant to the journalists who are writing about these topics for these publications). See the Vox article below for a deep dive into music AI. But the philosophical implications of all of these Gen AI tools will be related, and I’ve written about LLMs above as well. I am writing papers as well as music in my grad school program, and I am back around undergraduates also writing papers these days, so that’s as relevant to me as the music tools are.

MIT Technology Review:

What is AI? - 7/10/24, Will Douglas Haven - This seems like a good place to start if you want more about this topic - a super deep dive (like a miniature book - it even has “chapters”) about what the technology is and where it stands today, including a lot of coverage of current debates (as of early-mid 2024) over the technology and what it should be.

The Atlantic:

Silicon Valley’s Trillion-Dollar Leap of Faith [Apple News+] - 7/29/24, Matteo Wong

Do Navigation Apps Think We’re Stupid? [Apple News+] - 7/4/24, Ian Bogost - Not quite a Gen AI article, but nonetheless an interesting one about the disconnect between how humans perceive the world and how computers can see the world. This maps onto GenAI by reminding us that systems that seem to us like they should work one way (how we see the world) may not in fact work that way, which I then apply as “systems that seem to create coherent, grammatical language [LLMs] do not necessarily ‘think’ the way we do” [my quote]

OpenAI Just Gave Away The Entire Game [Apple News+] - 5/21/24, Charlie Warzel - A look at some business shenanigans in the field, and a larger discussion about the ethics and self image of the industry based on that.

Would Limitlessness Make Us Better Writers? [Apple News+, Website] - 4/25/24, Rachel Khong - I referenced this one in the passage about AI in my essay Watchwords on my website. A novelist looks at what makes writing human, and whether LLM technology leads to better writing or just more writing. Addresses some of the same points I’ve made from a musical perspective as well, particularly in the Human Music part of my AI essay above (Tip No. 5). 

It’s the End of the Web as we Know It [Apple News+] - 4/22/24, Judith Donath, Bruce Schneier

Godel, Escher, Bach, and AI - 7/8/23, Douglas Hofstadter

AI Should Not Replace Thinking at my University - 6/22/23, Douglas Hofstadter

The New Yorker:

Why AI Isn’t Going to Make Art [Apple News+] - 8/31/24, Ted Chiang - This article has been getting a lot of coverage in the week or so since it came out (writing 9/7), both in agreement and in critique. Matteo Wong at The Atlantic had a fairly strong critique of this article. I find myself in more agreement with Chiang’s article than Wong’s critique, and I actually wrote my own response to Wong’s article in what I’ve taken to calling a “rage essay” the night that I read it. I’m not sure I’m going to post that essay (I haven’t yet), but there are certainly a lot of strong emotions in this space at the moment(!)

New York Magazine:

The Future Will Be Brief [Apple News+, Website] - 8/12/24, John Herrman

Scientific American:

ChatGPT isn’t Hallucinating - It’s Bullshitting! [Apple News+, Website] - 7/17/24, Michael Townsin Hicks, Joe Slater, James Humphries

The Verge [website links]:

No One’s Ready for This - 8/22/24, Sarah Jeong

The AI Photo Editing Era is Here, and it’s Every Person for Themselves - 8/19/24, Allison Johnson

Procreate’s Anti-AI Pledge Attracts Praise from Digital Creatives - 8/19/2024, Jess Weatherbed

Google Pulls Gemini Ad from Olympics After Backlash - 8/2/24, Victoria Song

We Have to Stop Ignoring AI’s Hallucination Problem - 5/15/24, Alex Cranz

Apple Doesn’t Understand Why You Use Technology - 5/9/24, Elizabeth Lopatto - Not an AI article per se but a reaction to Apple’s ad Crush! in May 2024, the reaction to which was largely experienced in the context of Gen AI as I mentioned in my Facebook post I copied above. I referenced this article in that post.

Vox:

What AI in Music Can - and Can’t - Do [website] - 7/1/24 rev. 8/5/24, Adam Clair - A deep-dive article about AI in music, how it works and some of the philosophical implications. Makes some of the same points I raised in my essay above, but also discusses some other points.

Books:

There are of course lots of books about this area today. Here are a couple I’ve read recently as of August 2024 that I’d recommend.

Literary Theory for Robots - Dennis Yi Tenen, W.W. Norton, released 2/6/24. A combination of a history of information processing leading from the Medieval period all the way to modern Gen AI LLMs, and a manifesto about ethical and unethical ways to use the technology, given that it’s not going away. He puts modern systems in context of historical developments and takes what I’d consider a relatively even-handed approach to the technology. I don’t agree with everything he says in the last chapter regarding lessons for using AI, but I do agree with some of it and it was certainly an interesting read overall. I referenced a couple of his concepts in Tip. No. 3 in my AI essay above.

Who Wrote This? - Naomi S. Baron, Stanford University Press, released 9/12/23. Another look at a slightly different version of the history of information processing / word processing technology. She takes a somewhat more critical look at the field than Tenen’s book, an approach that more closely aligns with my view on these issues, though as mentioned above I did like Tenen’s book too. Both of them tried to find both good and bad aspects, or even just value-neutral aspects that are facts of how the various forms of the tech works which are important to remember as we consider these tools.

One interesting tidbit I recall from this book was a discussion of using “AI” software several years ago to help grade the essay portion of the SAT. I think that was even in use during the period I would have taken the SAT, though I never took that test. We did the ACT at my high school. But Baron notes in her book that basically the way the system graded the essays was largely based on length of sentences, longer multi-syllable words, and higher scores on the Flesch-Kinkaid Readability test (Page 42). She notes that one glaring omission from the list is “accuracy”, such that a meaningless essay that used long sentences with big words would get a higher score than an essay that explained a more complex topic in simpler terms accurately.

I’ve read a lot of graduate level papers in the last year and I can say that there are times when I’m not convinced the authors actually know what they are talking about and they try to cover it with grand wording that I’m not sure means anything at all. I try to keep my writing relatively decluttered from that kind of language personally. My favorite word that gets abused in writing about art is “kaleidoscopic” - sometimes it almost seems like there’s some contract people sign that they will use that word in every piece of writing about music today, and I think I’ve found maybe one instance where I agreed that it was appropriate. I even found it on page 3 or so of the preface to an academic book about the practice of going for a walk as it relates to writers, both in stories they write and in creative practice itself to help them write. I’ve taken to highlighting it every time I see the word getting used, often with a wry note “well, that didn’t take long!”