It seems we’ve reached an era where shitty pop music is no longer enough, so we had to get AI involved.
As a musician, tech journalist, and person who gives a fuck about quality music - I’ve written at length about AI’s impacts on all aspects of society. I know how AI has the potential to fundamentally change the future as well, but I want to set the record straight. I am not an AI-hating arsehole, just someone who is not ready to live in a world where we sacrifice quality for the sake of generic ‘musical’ content pumped out by machine learning.
I recently read an article highlighting a startup that focused on doing just that. Their goal, to continually produce as much mediocre musical content as possible for the sake of entertainment. For me, this not only angered me, but it made me slightly depressed. Along with the countless slippery slopes AI-produced music presents - it simply isn’t good - and falls short of the ultimate goal of music - evoking human emotion. Music also resonates with us on more than just a superficial level.
It seems in 2019, recreating the Turing Test was all the buzz. And it appears as if our new goal is to pass the exam. Thus, welcome to a new dawn, not at a beautiful, previously undiscovered location - but at a point in time where we merely want to produce inferior music. This could be in response to a need for a distraction from the countless, and almost-certainly detrimental worldwide political and humanitarian challenges we face as humans.
An Expert’s Take
I would argue, during times like these, music is exactly what we need. But good music. Not this. I recently spoke with futurist and fellow musician, Daniel Portal to get his take on things. The biggest issue for him, "AI will never be able to replicate human expression, nor will it be able to reproduce the connection we feel when we hear good music. How can you teach a machine to give humans goosebumps?,” Portal said.
He did add, there's a consensus amongst those who exist in the human race, and we agree on what music is and isn't. Portal attributes this to the feeling music gives us, adding: "Will AI ever be able to replicate this? Answering the very question he posed, he said: "I do not think it will, but I do not believe that current machine learning is capable of doing so.” During our time together, Portal posed, what for me, arose as a central theme in our conversation:
“How can we train these models to replicate such human behaviors? More importantly, why would we? This, for Portal, is where things come to a screeching halt. “For me, this is where AI falls short in the music creation process,” he noted. This left me scratching my head and wondering, How the hell did we get here anyway? The answer to that question could lie in a test developed in the 20th century by English mathematician Alan Turing.
Trying to Recreate the Turing Test
Developed in 1950, the popular test assesses a machine's ability to display intelligent human behaviours. The experiment first appeared in a scholarly article titled "Computing Machinery and Intelligence." And I will admit, since its debut, it has been met with both criticism and praise. You see, Turing's exam makes use of a human evaluator to observe natural language conversations between humans and machines to see if the evaluator can tell the difference between machine and human reactions.
Since the test is text-based and relies on a computer keyboard and screen, the test is subjective in nature. If the evaluator cannot reliably tell the machine apart from a human, the machine then passes the Turing Test. In the end, the results of the test do not depend on the machine's ability to provide correct responses to queries, just on how closely the machine’s answers mirror that of an old fashioned warm-bodied human. I propose, this is what we’re trying to do with music, which could lead us down a slippery slope of second-rate music.
The whole point of the test is that it attempts to answer a crucial question. Can machines think? If machines can think, they may have a higher probability of understanding how emotion works. He further argued, something as subjective as thinking is difficult to measure and define, which led to another pivotal mystery in his eyes: "Are there imaginable digital computers which would do well in the imitation game?" He believed this was the question that could be answered.
AI Is Good for a lot of Things, Just Not Music
Today, AI’s impacted massive positive change in the real-world, as well as the corporate realm where AI technology means big business. According to recent IDC data, global spending on AI systems should exceed $97.9 billion by 2023. On the less evil side of things, that is to say, in taking a less cringingly capitalistic approach - AI can be used for the greater good of humanity when there is solid leadership at the helm of ethical enterprises without ulterior motives driven purely by profit.
And there are many captivating examples of this, as companies on a global scale grow more aware of how their technology can be used for the betterment of society. Cleaning up polluted ocean floors, cigarette-infested beaches, and even removing unpleasant toxins from major cities - all good applications of AI if you ask me.
Since AI is not meant to supersede humans, merely to enhance our capabilities, this makes fully-produced AI music and even partially produced music cringe-worthy and offensive to those of us who care about the kind of music we consume; how it makes us feel, and the connection we feel to the artist who created it.
As we persist well into the future, it appears a growing number of companies want to replicate human-like traits in Artificial Intelligence. My conclusion and the conclusion of plenty of other respected musicians, producers, and tech-loving music nerds alike - AI does have a place in music but when it leads to this, we have every right to ask why we should even create something so subpar in the first place.
Has humanity lost its way? Have we become so numb and robot-like, that we’re willing to consume something just because it is trendy? If we have not lost our way, we’re certainly well on our way to doing just that. To leave you with a final thought, I asked Portal if he believed we were far off from AI leading the charge in the music creation and production process. His reply, “AI creating music that will genuinely make a difference in people’s lives (laughs) - we’re still a long way away from that."