Machina Sapiens

Welcome, my son
Welcome to the machine
What did you dream?
It’s alright, we told you what to dream

Pink Floyd, “Welcome To The Machine”

In one of the machinations of life that we insist is coincidence, a neighbour who’s also a friend and I got into a discussion a few weeks back on machine learning algorithms, specifically those employed by social media channels. I’ll come back to that coincidence bit later. Social media algorithms feed you more and more of what you ‘like’ and reinforce your own beliefs with little or no room for a differing point of view, whether intelligent or not (I do not mean the comments but what comes by way of your ‘feed’). You’re a permanent resident of that echo chamber that one hears so much of these days where variety comes only in the tone and tenor of the voices but not in the content which remains monotonous and stultifying.

It was odd that this particular conversation came up just as I was thinking about the music service Pandora and then later on the same day, the episode of ‘Elementary’ that was up next on my repeat viewing of the series was one that dealt with this echo-chamber filling tech. I had heard much of Spotify’s recommendation engine at a time when the service was not legally available in my country. So when it did come in, I jumped on to it early. And found that it was a train that ran on unvaryingly familiar tracks that I had leapt into. Boring. Many years before that – around 2000/2001 – I had stumbled upon Pandora which at that time was subtitled the ‘music genome project’. BTW, back then it was an internet radio service. My experience of it makes me think that even after the passing of years it still had the most interesting and rewarding recommendations that I’ve come across thus far. I was introduced to the likes of Anna Nalick, Phoenix, Hawthorne Heights and scores of other artistes that were just breaking out into people’s ears and minds. The most important thing, though, was that many of the recommendations were quite orthogonal to and seemingly independent of what I was listening to and most of them worked. It was an absolute joy! And then Pandora became geographically restricted to the States and I couldn’t be bothered to go down a VPN tunnel to it; didn’t even try to see if that worked. There was a time when it had a huge number of subscribers as it grew over time and got taken over by Sirius XM but I read recently that it’s been in rapid decline in the last few years. Pity! A good product gone to waste, more or less. Instead we have ‘industry leaders’ throwing up vapid sound bites for us to deaden our senses.

Machine learning and artificial intelligence that gets talked up so much is still at a very rudimentary stage, at least in what’s seen in public. Whether its development into what is envisaged of it is a good thing or not is a moot point; I think it is eventuality. We will only get to the beginning of really serious artificial intelligence if the algorithms are able to reach tangential and unexpected points in their set of solutions rather than the relatively simple and limited like-to-like options that we see them come up with now. In the context of music, by way of a very simple example, that would be when one listens to the ancient Hindu Shaivite Rudram the algorithm exhorts, however gently, one to line up a dirge-like Doom metal or a nihilistic Death metal piece or a band like Frontierer next instead of another Hindu religious composition like Aigiri Nandini that it would currently deem appropriate. Or the sparse “When The Cellar Children See The Light Of Day” by Mirel Wagner to extreme metal fans. FYI, Shaivite chants like the Rudram are very much metal in their sonic depth albeit without the clamour of the genre and that album of very quietly delivered songs by Ms. Wagner is the scariest, most deeply distressing one that I’ve heard yet. Like I said, these examples are rather simple ones but when they turn real it will mark a significant scaling up of AI capability, when from being comfortably numb the movement is towards breaking open bounds of norms and what is familiar, to wreck things.

Of course, it is quite possible that this has already begun and the advancement in this area is way ahead of what we plebs think. Perhaps for now the social media and digital music pantheon of gods deems the current circle jerk content is what is convenient for them; the not-quite-proverbial opiate for the asses.

The machines are here, even if that’s not exactly music to the ears. We are the machines.


One Comment

  1. Vijaya Raghavan says:

    Dear Ram,

    This is a well written perspective on the technology of our age. While some of the nuances of musical qualities which you describe here, escape me, you make a fine point about how most recommendation engines today cannot really distinguish the intangibles that make up human appreciation for a piece of music, or art or any other composition that is created by human ingenuity.

    But we have come very far in the AI story and I suspect this is not the end of the journey. Perhaps, the day is not far away when more humanness is incorporated into the machines. But what would it mean to be human (or a machine) at that time, I wonder. Will nuance and appreciation for raw beauty be of any consequence? As the machines gather more human like qualities, will we humans lose them?
    Perhaps a factory in China and the billionaires of silicon valley will decide?

    I enjoyed reading this. Perhaps someone in Spotify should, as well.


Comments are closed.