Holly Herndon’s third album, Proto, sees her collaborating with an ensemble of talented vocal performers… and the digital voice of an AI program called Spawn.

Spawn was developed by Herndon, and her collaborator and partner Mat Dryhurst. Most AI music projects focus on creating automated forms of songs. But Spawn is different. Rather than trying to replace a human, Spawn works through a voice-modelling approach, observing and augmenting human-voice inputs to change the possibilities of the way that music sounds.

Spawn’s musical capabilities were first heard on Herndon’s collaboration with JLin, Godmother. On Proto, Spawn’s contributions – learned through call-and-response training sessions with her human bandmates – fuse with vocal melodies and harmonies that call back to Herndon’s experience singing in Tennessee church choirs. The result is an evocation of folk musics from many parts of the world, while pointing to a fruitful future of human-AI cooperation.

Herndon completed the album while working on her PhD dissertation at Stanford, and is now Dr Herndon to you. The album has just been released, and live dates start at Pioneer Works, New York, on 18 May.

What was Spawn’s most surprising contribution?
I think there were two pivotal points for us. One of them was Birth. [Spawn says an unprintable word.] It's not something that we trained. It's just kind of like a reconfiguration of my language. So that felt like a really funny moment of “Kids say the darndest things!" And also Godmother. I knew it was going to be percussive in that way, but I didn't know how insane it was going to sound.

There are a lot of sacred and folk vocal styles on Proto, too.
The whole album is drawing on various folk traditions, and a lot of it has to do with the individuals who were in the ensemble and the different experiences they have. I got really interested in this idea of all these different vocal traditions that happen all around the world as this almost inherent human technology inside of us, that found its way out as a way of communication in hunting. And just part of human intellectual evolution, essentially, human intelligence as being this collective project through the ages, and artificial intelligence as the next phase in this collective project.

I’ve assumed that Spawn was a she...
That is the pronoun that I've been using as well, and the reason for that is because the first training that she did was on my voice. There's of course this long history of female digital assistants. And then there's all kinds of research about people mistreating the digital assistant, and then that behaviour seeping into the physical world as well. But I don't think that I have to tether myself to that necessarily. Like, I see myself reflected in her, and so for me she's a she. And I'm not mistreating her!

Beyond calling her an “AI baby”, though you built Spawn and understand the workings, do you find yourself anthropomorphising?
I mean, we say child, but we don't think of her as a human child. We think of her as an inhuman child. We just really like this useful metaphor, because we see the state of the technology as pretty nascent. It's really powerful and could potentially do a lot, but it still sounds pretty scratchy and rough at times, which you can hear all over the record, the roughness of her logic trying to figure out how to sculpt from sample to sample. So, we really like this useful narrative, the importance of instilling the parents' value systems in the child early on, and “it takes a village to raise a child”. But there were several human babies born during the making of this record... there's a very obvious and palpable difference!

Artificial intelligence is already so much in our lives. Do you think fear of that technology and retreat from it almost leaves people more vulnerable to it?
Oh, 100%. I think that if you rely on 90s, cyberpunk, dystopian narratives, it's almost like saying that we don't have an alternative vision or we can't imagine anything else. And so that almost becomes something that we can entertain ourselves with on the side, while our digital selves are being fully scraped and captured, and turned into training sets for god knows what. I think we have to have a critical lens on what's happening, and there are some really good institutions that I follow that do deep research work, like AI Now, run by Kate Crawford and Meredith Whittaker.

But I also think that artists and writers have this duty to come up with new ways of thinking about how it might be. And that's where I see my role. Because otherwise it’s handing over human agency to the powers that already have so much control, if we are distracted by existing narratives or are unable to formulate a new fantasy of what we would like to see happen. Like, new demands, y'know?

Where can you see yourself going next, what has this work opened up for you?
Oh, that's always this question! So JLin who I collaborated with on Godmother... if anyone ever asks her, when she's just finished an album, what's next, she just loses it! She's like, “What more do you want from me?” It's an inside joke that we have: it's never enough.

Oh, it's plenty enough. I can't believe you were finishing a PhD while you were doing all this.
No, I won't take that out on you... I don't know where it's gonna take us. I mean, touring with a lot of people is definitely opening up new challenges, but it's also rewarding in a new and exciting way. Mat and I were touring for a long time, just the two of us, and then we invited Colin [Self] to join us, and it was really exciting to have a third person. We were really missing that kind of human contact. We both really enjoyed working with players, and I could see that expanding beyond the voice.

If you liked this story, sign up for the weekly bbc.com features newsletter, called “If You Only Read 6 Things This Week”. A handpicked selection of stories from BBC Music, Culture, Capital, Future and Travel, delivered to your inbox every Friday.