“The gods envy us. They envy us because we’re mortal, because any moment might be our last. Everything is more beautiful because we’re doomed. You will never be lovelier than you are now. We will never be here again.” Achilles in: “Troy” (2004)

ChatGPT, an Artificial Intelligence Generative Pre-Trained Transformer chatbot developed by OpenAI, was released to the public on November 30th, 2022. I had heard about it in media, youtube videos, and from geek friends but put it aside, a topic for later, let other people make their experiences with it and let me read the recapitulation next year. How new, how good, how different, how ground-breaking could it be? Well, I was about to find out.

In early December 2022, only a few days after ChatGPT’s initial release, my 17-year old son had an assignment in his English class to do. Tasks like this with a deadline are always a struggle for him but this time he was totally relaxed. I offered my service to go over his first draft and make annotations and corrections. He printed it out and gave me two clear pieces of paper with well-arranged paragraphs. The writing was well-structured, matter of fact, flawless, not one mistake in terms of grammar or spelling, totally not my son. The vocabulary was elaborated, experienced, on the point, totally not my son, the content was correct, but very, very general, boring to the point that reading it was tedious, every argument was so expectable, to the extent that it was predictable and lame. It was so cold and right that it felt uninspired, almost inhuman. It took me about two or three sentences to be sure that this was not my sons writing, also not one of his friends and also not copied from some article in the web but generated by a machine. And that’s exactly what had happened, only days after its release every student in the world knew about ChatGPT. Of course, my son had been uncritical, he should’ve ordered it to be in the verbal expression range of a 17-year old German Student of English, that would’ve made it a little harder on me but still, it was way to obvious, with this he could’ve fooled no one, for sure not his own father and not his English teacher.

This personal incidence showed me the scope of the change this generation is about to witness. And others did so, too. Chat GPT is learning with every text it generates, it will get better fast, and this will be a very great challenge for text-based subjects, for teachers, students, parents, for universities, colleges, schools and homes. And this is only a text tool, it’s AI that’s accessible to the public, we have no idea what the big names of IT are employing right now and in the future to come. All we’ve seen so far is AI in popular culture but it’s constructed to take over many other aspects of life, so get ready. Although everyone’s talking about AI now it is being employed in other professional fields for years and decades and we only begin to understand to which extent public and society was and is influenced, controlled and manipulated. The possibilities are endless, they are amazing and they are terrifying.

But let’s take a look at our field of interest: music production. Working with computers, algorithms and early forms of AI is not so new as it might seem. There have been developments as early as the late 1950s focused on algorithmic composition that aimed at creating aesthetically satisfying new compositions (Lejaren Hiller & Leonard Isaacson: Illiac Suite for String Quartet), in 1980 a breakthrough in the field was the development of EMI (Experiments in Musical Intelligence) by David Cope at the University of Santa Cruz (Bach by Design). He was able to let AI compose two-voiced Inventions, that were later recorded by human musicians (Computer Generated Composition). Many of the first-generation AI compositions worked on German Baroque Music especially J.S. Bach and it’s obvious why: two-voices, counter-point, sequences, repetition etc. a very clear-cut compositional set of rules, that can easily be transferred to mathematical formulas und algorithms.

Things really get interesting with the invention, development and wide-spread use of personal computers and MIDI in the 1980s and Digital Audio Workstations (DAW) and high-quality samples in the 2000s. Now many people (not only university lab professors) were in the position to use, experiment and program musical content on digital platforms.

Early on the working possibilities of DAWs were tremendously extended by stock audio plugins (EQ, Compression, Delay, Reverb), virtual instruments (synths, pianos, orchestral instruments, beat machines, etc.) and third-party audio processor plugins, all came along with endless possibilities of manipulation and a selection of useful presets that every musician was glad to use. Some of these plugins were ground-breaking musical tools and soon to be industry standards like Antares Auto Tune (1997), Kontakt (2002), Guitar Rig (2004), Maschine (2009), Melodyne (2009), Apple-Loops (2009), iZotope Ozone (2015), recent: RC-20 (XLN), Spitfire Audio Instrument/Sound Samples, Output, Splice.

Did I use any of these inventions when they came out? Oh, yes, I did. I’m a creator, I’m always on the look-out for inspiring sounds & tools. I’ll employ everything that will enhance the craft of my art. Was/is it run by presets, by dynamic algorithms or AI-powered? Sometimes hard to tell but also not relevant for the creating mind. More important: Is it inspiring, can I access new sounds and possibilities, does it ease up the workflow for a wider range of artistic decisions? Is it setting me on-par with exiting contemporary productions? Does it excite me and my listeners?

Within no time I was using ultra complex AI-plugins like Flex Time and Flex Pitch on a daily basis. Artists performances with shaky timing or loose pitch could be corrected effortlessly, inaccuracies were soon to be a thing of a long by-gone past. Generated instrumental parts by plugins like studio drummer, studio horns, studio strings were soon a regular feature of traditional productions. Parameters (easy/complex, quiet/loud) could be easily adjusted, close chord-voicings would be split-up automatically so that they’d fit the selected horn or string-ensemble. Many, many times this is way faster, cheaper and on the point than the real thing. Who can arrange for, book and pay a handful of capable players in the studio and record their parts in a mid-, low- or no-budget production?

Very helpful is Ozone 10 analysis and mastering assistance. Mastering gives the all-important final touch to the production but can be expensive, it used to be one of the magic crafts in the post-production process and is usually done by somebody from outside with technical expertise, experience and a fresh view on the track/project. It is an expensive service and hard to get a grip on, when you’re totally absorbed by every step of the production itself. For the most part a regular treatment consists of measurable values of frequencies, multiband-compression and limiting, that are very technical and can easily be analysed, evalued and processed with AI today.

I hear about new, so called “creative”, apps and plugins based on AI that are supposed to make things easier, sometimes I download them, sometimes I open them, almost never do I use them, I’m a musician, composer and songwriter, an artist and craftsman, most all of the time I’m quicker and way more original when I work in my established work flow, those “creative” AI-tools distract me, I get lost in possibilities and lose my inner voice that I depend on when I attempt to create something I call mine. I tried to let AI generate lyrics with various sets of input, settings, topics, fragments. Doesn’t work for me at all. To me the results were uninspiring, bureaucratic, one-dimensional, very common-denominator. And: I’m a well trained and experienced musician, I don’t need AI help me write a sustainable melody, a chord-progression or lyrics, those parts are at the core of my artistic expression, the fun of it all and the reason why I chose to write, perform and produce music as a profession. I understand the creative act as a way of being. Why would I give that away and trade it in for music and lyrics that have little or nothing to do with me? I will let AI help and assist me, but I will not let it replace me, that will not happen, for only I can create my own art.

Almost daily we are presented with allegedly AI-produced music (they won’t tell us to which extend) with historically unlikely and impossible combinations: Kurt Cobain singing REM, Elvis Presley singing Nirvana, a new Drake song without Drakes participation. These phenomena have been around forever and they are called novelties, new and surprising at first, but mostly not very substantial or artistically worthwhile in the long run. We have Elvis-Tours without Elvis, ABBA-Shows without ABBA, we have Queen shows without Freddie, and for sure soon we’ll have Rolling Stones Tours without Mick & Keith, AC/DC without the Young-Brothers and Amy Winehouse without Amy. Just like in the movies where we had a young image of the 40-years old Brad Pitt or Johnny Depp, Princess Leia without Carrie Fisher and soon we’ll see new movies with generated versions of James Dean, Heath Ledger or Paul Walker.

Am I interested? No, like hell, I’m not. Amy Winehouse as a full-drunk drug addict had more music in her little finger than all the processing power in the world will ever have. Put a guitar in his hand and Ed Sheeran will rock any stadium alone, three nights in a row if needed. Give Billie Eilish a microphone and a beat by her brother FINNEAS and she will blow your mind within the first minute of her performance. That’s what I am interested in. Who wants to trade that in for Deep Fakes?

Nonetheless I see wide segments in the modern music production world were AI will be a strong foundation, segments in which not innovation and originality is wanted but well-established standards in little variations of a blueprint. Segments that come immediately to mind are music for video games (up to 100h long), music for films, film-series or TV-series (Hans-Zimmer-style), music for ambient use (commercials, malls, airports, relaxing, sports).

I put together a diagram to illustrate the characteristics of Artificial Intelligence as opposed to Human Intelligence. And don’t get me wrong: Human Intelligence is not somebody out there, Human Intelligence is you: The creative mind. You are the difference, you will make the difference.