Chat GPT and Music – threat or friend?
Here are my thought s on the subject:
First and foremost we have to understand that there is no “Artificial Intelligence” – “AI” is just a marketing gimmick. What??? Let me explain. Here is a definition of intelligence as described in Wikipedia (full article here)
Intelligence has been defined in many ways: the capacity for abstraction, logic, understanding, self-awareness, learning, emotional knowledge, reasoning, planning, creativity, critical thinking, and problem-solving. More generally, it can be described as the ability to perceive or infer information, and to retain it as knowledge to be applied towards adaptive behaviors within an environment or context.
The only characteristic of the above described that might apply to “AI” is learning. “AI” is in reality ARPP: an Algorithm Responding to Preset Parameters. There is no capacity for abstraction. No self-awareness. No reasoning, planning, creativity, critical thinking and not even problem solving. As for an ARPP there is no problem to solve. That is entered by the human that starts a query. But “Artificial Intelligence” sounds so much sexier, marketable than Algorithm Responding to Preset Parameters. “AI” is thought provoking, scary, enigmatic, mind-blowing! ARPP is – well – just another software code. “AI” is Emotional Marketing ABC. Emotional Marketing, or, as Forbes put it “The Super Weapon of Marketing and Advertising” (read article here)
Now will ARPP have an impact on music (you will not see me using AI again) Of course it will.
I have dabbled with AI
Call it a fact finding mission. GPT was neither the first, nor the only one. The one I use is called WriteSonic. And yes – I do have a subscription. So here is the ARPP response to the following query I used to give you an example: “Write a 300 word post on music and AI”
Music and AI: The Future of Music Production
The potential of AI in the music industry has been a popular topic of discussion lately. AI has the potential to revolutionize the way music is produced, recorded, and distributed. With the help of AI, music producers can create more complex and interesting sounds and music styles, which can reach a wider audience.
AI can also be used to automate certain aspects of music production, such as song composition, sound design, and mixing. For example, AI can be used to analyze a song and automatically create a corresponding drum pattern. This can make it much easier for producers to create custom drum patterns for different songs. Additionally, AI can be used to analyze the complexity of a song and suggest the best ways to mix it, allowing for a more seamless workflow.
AI can also be used to analyze a song’s structure and suggest changes or improvements. This could potentially reduce the amount of time it takes to produce a track by helping producers find the right sound or arrangement more quickly.
AI could also be used to help music producers distribute their work. AI-driven platforms could help musicians get their music to the right audience more quickly and efficiently, while also reducing costs associated with marketing and distribution. AI could also be used to recommend songs to potential fans, based on their past listening histories and preferences.
Overall, AI has the potential to revolutionize the music industry by providing producers with more efficient and powerful tools to create their music. From song composition and sound design, to mixing and distribution, AI could help producers create unique sounds and reach a wider audience. While this technology is still in its early stages, it is certainly worth exploring and could bring about some exciting changes to the music industry.
It took the ARPP less than a minute to write it. Impressive, right? The algorithm crawls through the web, pulls out information, rewrites it in lightning speed and voila! You have your requested piece for a post/ blog/ article. Just imagine that human would search through text on the web, in books, publications and articles, re-write it and then pass it off as their own work. We have a word for that. Plagiarism. Because no matter how sophisticated the ARPP is, it can only ever find what already was written.
We have a word for that. Plagiarism!
And this is where I see the biggest threat in ARPP. There is no accountability. No reference to the original source. No possibility to check if a reference is open to interpretation or if it was taken out of context. Let us – for the sake of the argument – play with this idea: in about a month someone starts a similar query to an algorithm:” Write a critical 300 word post on music and AI” and the algorithm finds this blog and the mention of ARPP as and alternative name for AI. There will be no mention of the source, no mention who came up with that alternative abbreviation, no credit for the creator (me). Now let us take that scenario and use it for a love song. My Query in WriteSonic: “Love song” Topic: “Forever yours” Genre “Alternative”. It gave me 3 alternative results:
You’re the one I want to keep around/ Even when the world’s against us/ You’re the one who makes me feel alive/ And I don’t know what I’d do without you/ I’m not sure I could go on/ Without you by my side/ I love you, forever yours
I don’t know what I’d do/ If you were to go/ I don’t know what I’d do/ If you were to leave me/ I don’t know what I’d do/ If you were to disappear/ And never come back/ I don’t know what I’d do/ If you were to stay with me/ And never let me go
You’re the one that I want to be with/ The one that I want to be with/ The one that I want to be with/ You’re the one that I want to be with/ Forever yours
Okay – none of these results blow my mind away, but – how can I even be sure some of these lines are not already used in a song and credited to someone else? How will I know if by using these lyrics I am not infringing on copyright? Remember, Algorithms can only regurgitate what has already been created.
When it comes to music creation – we already many steps further: AIVA (Artificial Intelligence Virtual Artist) – and algorithm that has even managed to get recognised by SACEM and an electronic composer! “By reading a large collection of existing works of classical music (written by human composers such as Bach, Beethoven, Mozart) AIVA is capable of detecting regularities in music and on this base composing on its own” (link to the full article on Wikipedia here). If you decide to use the algorithm you will even have to give it credit for the composition. Is this creation? Plagiarism? Whatever you decide it is that AVIA does – at least it is transparent. And you will see they only used classical composers for the “learning” process. Why? Because their work is no longer protected by copyright – are now public domain works, these classical music pieces may be freely copied, distributed, adapted, or performed in public without permission or paying a fee – royalty free (see source).
So how will it be in the music creation of popular music? When we start using ARPP to fill in sections of songs, help with writers blocks or create a riff for a guitar? Algorithms can only reproduce what already has been created. When will we see the first lawsuit against an algorithm that ripped off somebody else’s work and – more interestingly – who gets sued? The composer? The algorithm? (Unlikely as to my knowledge a lawsuit can be brought against a person, business, organization or even a government – none of which applies to an algorithm) Or the business that created the algorithm? I am sure it will not be the latter – they will have some fine print that dumps the responsibility on the users shoulders…
Chat GPT and other algorithms like it will have a massive impact on our lives, work and also on music creation. We will learn to work with it and utilise it as a helpful tool. But unfortunately, again, a new technological innovation has been released into the world without any consideration to the consequences. ARPPs will be used to take over the more mundane tasks in work. Legal documents, newsletters, resumes, ads (in my case I will definitely continue to use it for starters for my Instagram captions and hashtag search) …and in a positive outcome it will free up more time for our core tasks – as actually creating music! But no one thought to adress legal consequences (can we hold algorithms responsible for intellectual theft?), job loss or the impact on the educational system. Per example: should algorithms be obliged to provide a list of its sources? Similar to students who have to include a reference list in their research? For me the real danger in using ARPPs as Chat GPT lies in the undermining of intellectual property. We need to start developing systems applying the same thoroughness for releasing virtual products to the market as we do to physical ones.
Create and feel great!
PS: after long consideration I decided to leave a tip jar on my website. If people are eager to pay for algorithms, surely original thoughts are worth something 😉 I would appreciate a contribution!