AI's Progression into Song Creation
When discussing artificial intelligence, one of the things that artists will always tell you is that AI can't make art. In broad strokes, that is true - the refined talents that art requires is still, so far, a human-only domain. However, in the grander scheme of things, artificial intelligence has accomplished quite a lot. The History Channel notes that in 1997, an artificial intelligence trained to play chess beat a human being
, Grandmaster Gary Kasparov, for the first time. Today, we've come to acknowledge that computers and artificial intelligence are better than us as a few things. Art, however, remains the domain of humans, but how long is that fact likely to last?
From Computer-Generated Tones to AutoTune
According to Quartz, Alan Turing, the genius inventor that helped the allies win the war with his decoding machine, also invented a device that was able to generate tones
as music. Through the years we've seen a lot of artists delve into experimental music generation techniques. Vice offers insight into David Bowie's lyric-writing techniques using a sentence randomizer
called the Verbasizer. Even so, all of these elements still rely on human beings to do the final bit of composition. Art is, after all, humans speaking to humans on a level that goes beyond mere words. Even if a composer uses producer sound kits
for a sample, it's still the product of a human being.
The First Pop Album Composed by AI
In 2018, the first-ever AI-composed album called Hello World was created by Francois Patchett. Despite it being a landmark moment, the songs were less than spectacular, and it's unlikely it would ever see commercial success. The BBC mentions another engine called Flow Machines
, a system that has been used by the likes of Belgian superstar Stromae to help him produce tracks using artificial intelligence. At the end of the chain, however, there's always a human for input and cleanup.
Functional vs. Experimental Music
Flow Machines forms a part of a system of technology that aims to produce functional music. Another example of this is a system known as Amper, which companies like Reuters use to create stock music for use in the background of videos and ads. Amper takes input from a producer based on genre, mood, and tempo and creates a generic piece of music that is indistinguishable from a professionally produced bit of stock music. Functional music is only one half of the equation, however. Like Stromae, a few other modern artists rely on AI to guide their music production with experimental results.
On a few pop songs, you'll probably notice that they credit a composer named Dr. Luke. This "Doctor" is, in reality, an algorithm that deals with creating catchy, generic pop music hooks that people are almost guaranteed to love. The problem is that it boils music down to forgettable cotton-candy music that isn't deep or reflective at all. It's catchy and has an excellent beat, but doesn't do anything that music does. Other composers are looking at AI in the exact opposite way - to help them push their boundaries. By using AI to develop a template, the band YACHT was able to mix and match music to create a sort of Frankenstein's Monster vibe with their last album. The result was a lot of trippy dance-pop that seemed more akin to an LSD trip than a studio album. It did, however, result in YACHT's first Grammy nomination.
A Danger to Jobs?
There's already a massive concern about whether AI will replace people in 9-5 jobs, but most creators and composers haven't considered the possibility. In the past, the music industry has seen its share of anti-technology movements. In the 1970s, guitarists formed a petition to stop bands using synthesizers for fear that they would be replaced in the recording studio with robots. However, human music is still something that AI can't quite grasp, no matter how much it mixes and matches. For the time being, the jobs of composers and songwriters are safe. How long that'll be the case remains to be seen.
Curiosity Breeds Innovation
As it stands, artists have started embracing AI in developing their music. Composition tools such as FL Studio already have ways for producers to automatically detect tempo, and create chord progressions based on any number of scales. Experimental coders are getting their work in the hands of the right people more often because of how easy it is to get in contact with someone and sell an idea over the internet.
However, there are a lot of ethical issues to consider when it comes to introducing AI into music. Creating an AI that "sounds like" an artist isn't a difficult feat anymore. In such a case, if an AI artist covers a song written by a real person, does copyright law extend to the copy? Issues such as these may hamstring attempts to bring AI into the world of music, but it's unlikely that it will stave off AI's adoption in the industry forever.
Subscribe to get your daily round-up of top tech stories!