When looking back on the creative trends of 2022 one of the most common trends, at least for me, was the boom and subsequent shunning of artificial intelligence (AI) art. With the rate machine learning is evolving, is it going to help creativity as a whole, or should genuine creatives prepare for the overthrow of the machine?
Let’s be honest, AI is set to change the landscape of the creative industry for better or for worse, and it’s not going away any time soon. What began as a flash in the pan when WOMBO launched its Dream AI app quickly turned into a plague for creatives across a number of industries, not just the art/design world.
Seeing the development of Open AI’s Dall-E, I was excited for the future of AI, especially because, after announcing and then opening Dall-E 2 to the public, I was wholly unaware of how things like this actually worked. I, like many people, presumed that the AI itself was creating its generated art from scratch and maybe using eras and styles of art (like impressionism, modernism, realism, etc.) to create original pieces.
Now I know that I was wrong, especially now Lensa AI has blown up and people have learned how exactly this AI art is being generated, and unfortunately, it’s not good news for creatives.
Speaking specifically about Lensa AI (as I’m unaware exactly how Open AI’s machine learning has been done), the way this art is generated is by AI Training, or AI Learning via Stable Diffusion, I.E. human operators feeding already existing art into the algorithm to use as the building blocks for user prompts. And the main issue here? The art being used for AI Training is being done so largely without artists’ permission.
To make matters worse, Lensa is charging users money to generate these portraits, none of which is going to the artists that this art is based. And while some might argue that AI is just a form of artistic expression that uses other artists’ work as “inspiration” – much like art and artists have done for years – I personally think this is a little different than simply being inspired by someone.
As demand for AI-generation progressed, we’re now seeing even more developments in the AI world a few of which could have both a positive or negative impact on the music industry as a whole.
ChatGPT, by the same creators of Dall-E, is an AI where you can enter a prompt and have AI come up with an entire batch of written text. It had a bit of a boom fairly recently thanks to TikTok and videos of users using this AI to create recipes using weird ingredients or creating a rap using nothing but a Denny’s menu.
While it might seem like a bit of harmless fun, the impact that this could have on the music industry alone is wild. Simply being able to create lyrics for entire songs just by typing in something like “emotional rap lyrics about being a loser in the style of Drake” and getting an entire song, pre-written, could be seen as insanely damaging to the music industry.
On the other hand, though, tools like this, if utilised as such, can be great for independent and upcoming artists. Being able to utilise ChatGPT to create the start of some copy for social media, create bios for electronic press kits (EPKs), create the framework copy of a press release, or even some ideas for an editorial pitch.
Hell, even today I was introduced to a platform that with a simple prompt I was able to come up with an almost cohesive piece of music which evolved over time:
Yeah, it sounds like absolute garbage, to begin with, but the longer you let it play the slightly better it sounds. Anyway, my point here is that even AI art at the beginning was trash, you could just tell it was generated by AI. Now though, just a year or so later, that AI has evolved and is creating much more genuine-looking pieces of art.
Imagine how AI-generated music will sound in a couple of years’ time.
As someone who is heavily involved in the instrumental mood-music scene, I and other producers are already facing a goliath of an enemy in the form of faceless fake artists released through labels that are seemingly supported wholeheartedly by Spotify – some of which are already utilising AI art for their many releases.
But now, we potentially face an even greater threat from individuals or companies willing to utilise AI to create entire catalogues of music to be pushed out on digital streaming platforms (DSPs) via labels (or the labels themselves) that have support from those platforms and pushing it out ahead of genuine artists who put their heart and soul into the music they create.
Now, to escape from the doom and gloom a little, let’s ask, could AI in music generation be a useful tool? Absolutely it can be. The example video above, at least for the time being, seems to be more of a demo or use case for this type of tech. There are companies, however, making strides with AI and music to help with production, such as Audialab with their Emergent Drums plugin.
Emergent Drums is an absolute spot-on example of how we as musicians and producers can utilise AI as a tool in our workflow. This plugin uses AI to “generate endless drum samples, all royalty-free”, which could absolutely shake up how we find our sounds as we produce.
They are even working on an update which can utilise already existing sounds to generate similar-sounding-but-unique hits, all of which can be customised and tweaked within the plugin.
Currently, there are almost endless avenues for artists to find sounds, be it Splice, Looperman, other producers, or even chopping individual hits from older records, but that results in absolutely HUGE libraries of sounds which most artists might not have the space for if they’re working from laptops, for example. So tools like Emergent Drums solve that issue.
With all of that, are we finally entering a world that was only dreamed up in Sci-Fi movies? Is this the start of a “man vs machine” generation where creatives are fighting against computers, not with computers, to create content and actually earn a living wage? Or is AI just the beginning of a technology which if utilised correctly, can aid the production of art, written content, or music?
Well, creatives across the globe have already kicked up enough of a stink to prevent companies and developers from using their content for AI Training. We’re seeing big hitters like Google making moves to develop tools to detect and penalise AI content. And I’d be surprised if the music industry isn’t developing similar tools to detect AI-generated music and figuring out a way of either demoting or preventing DSPs from being flooded with fake content.
So while I can’t say for sure, I’d hazard a guess that we’ll continue to see AI boom over the next year or so in some way, shape, or form as more technologies come out and wow us with their ease of content creation.
But as with most online trends, they’ll fizzle out eventually and we’ll probably instead begin to see the development of tools that utilise AI to aid with our creation of content rather than replace it. Adobe already use machine learning in some of their products to aid with creation, Audialab is another in the music industry, and AI tools for written content have existed for a while, so I’m somewhat hopeful for the future of AI as a tool, rather than a one-stop-shop for creation.