In 2018, it’s hard to imagine a time when YouTube was making headlines as an upstart. But, that’s exactly what it was when it burst onto the scene in 2005, suddenly allowing anyone with a camera and a microphone to broadcast themselves to anyone else on Earth with a computer and an internet connection — and Taryn Southern was there for it. After going viral with her Hillary Clinton tribute video “Hott4Hill” in 2007, Taryn has become a digital mainstay, producing a decade’s worth of music, comedy, web series, and anything else that technology makes possible.
Today, the technology that’s caught her eye is AI. On her upcoming pop album, I AM AI, every track has been composed in part with the use of AI software — just another experiment from someone who’s made a career out of conducting them. We got a chance to talk to Taryn about what exactly it means to compose music with AI, where she sees technology going, and how YouTube helped her get around the usual Hollywood roadblocks that aspiring entertainers face.
Here’s our chat, which has been edited for length and clarity.
Your upcoming album, I AM AI, was created with the help of a few different programs, like Amper and IBM Watson, with the AI generating the beats and the harmonies while you compose the melodies and the lyrics. Can you walk us through what that creative process is like, and how much of it is tinkering with what the AI generates versus just letting the AI do its thing?
I liken the process of writing with these AI programs to the process of directing a film. I’m the director and the editor, and the AI is giving me the raw footage. It’s giving me some interesting stuff to play with, giving me more than I need, so I can find the little delicious nuggets of goodness and then piece those together into something that I deem exciting or useful or enjoyable.
That process, however, for each of these technologies is different. In the case of Amper, I’m primarily working off of a beat or a track that it spits out at me based on parameters I give the algorithm, so everything from genre and mood to BPM and instrumentation. It will give me a bunch of different tracks depending on the parameters I set, and then from there I can edit as little or as much of the song as I wish by going back and forth with the algorithm and making changes from there.
With something like Watson, in some ways I have a little bit more creative license because I’m giving Watson the actual data set that it’s learning off of versus Amper, where it’s using the rules that it’s been given by the engineers to create these sets of music. So with Watson, I’m picking the data set, so I could say I want it to just learn from Beatles music from the 1960s. Or, if I wanted to create a pop song using music only from the baroque period, I could feed it that kind of music. So, you can see how that sets up some interesting creative entry points into the music writing process that you wouldn’t otherwise have.
The other thing that working with a Watson or a Magenta does is that you’re working with a raw MIDI file that spits back at you, so I can actually shift that instrumentation to whatever it is that I deem fit for the song, so I could take a violin and turn it into a synth, and that would be my human free will choice that I make. With Amper, you can still make unique choices on instrumentation, but you’re leaning on the algorithm and the sounds that are within their library to get a finished product.
The songs on the album are all related to the interaction between humans and machines. Your latest single, “Life Support”, seems to ask the question of whether or not we’re over-reliant on technology. What would you say your take is on that?
I’m an optimist! I’m really excited about the future of technology, but I think we’re in the dark ages of how we use it. I think that our reliance on screens and these extensions of ourselves that keep us buried, head down all day, strained over a computer, are going to be seen as archaic and brutal in 30 years, when we’ll have our beautiful AR devices that we can just pop up, and it doesn’t have to disturb anyone else, and the sounds are not intrusive, and they don’t assault us at every corner! I think that we are at this funny place where we’ve had a decade of BlackBerry or smartphone usage, and it’s shifted the culture away from being present all the time, and now humans have to actually learn self-control when it comes to their devices, like when they’re going to be on and when they’re going to be off. So, we’re just in our growth phase, and I actually think that we’ll become more reliant on technology in many ways, but I think it’ll be there to serve our humanity, not make us a prisoner.
You’ve got experience in all things media, but up until now, a lot of what you’ve done on YouTube has been lifestyle or comedy videos, whereas I AM AI is your first full non-comedy pop album. Did the desire to make this album come from wanting to do a pop album and seeing AI as a way to make that happen, or was it more that you were curious about the potential of AI and saw a pop album as an interesting experiment?
It was a combination of the two. I had been wanting to make a project for several years that explored the relationship between humans and technology, and that went through a few different iterations. I saw that as a web series, I looked at doing this explorative spoken word series in virtual reality. I actually took that series to Google and pitched it to their creative grant program, and then I read an article about AI and the future of music in The New York Times about a year and change ago and immediately thought that I have to combine these two ideas, with AI as the mechanism to write the music, and then it became an album and took on a life of its own. I’ve always loved making music, so I probably would’ve continued to write music either way, just now I’m doing it with AI. But yeah, it was the perfect synergy of things that I’ve been toying with in my head for a while.
With your music videos, you include a lot of VR elements. What do you see as the potential future of VR versus AR? Do you see a lot of overlap, or do you see one becoming more ubiquitous than the other?
I’m equally excited about both for totally different reasons. I do think there’s a tremendous amount of overlap. If you’re thinking about it just from the creator perspective, the biggest difference is that one screen is translucent and the other isn’t! But you can essentially create and manipulate objects in a three-dimensional sphere, where VR and AR are using many of the same mechanisms. Of course, with AR, everything is contextualized with location, so that makes it a unique storytelling device. I think AR is going to be far more ubiquitous in use, just because people will be able to use it anywhere and everywhere. It’s going to be responsible for our transportation, our grocery shopping, our computer interfaces.
If you want to have a fully immersive content experience, then VR is probably the way you’re going to go. But, I imagine you’ll still be wearing your same AR glasses, you’ll just have something drop down so that it’s not translucent. I do think AR will see its boom way before VR really catches on. The headsets are still too expensive and too bulky, and not quite good enough. They’re not quite there yet.