Grimes (c) in the zone while producing [and AGI too]

From a conversation with physicist Sean Carroll:

Sean Carroll: 00:10:13 What you’re given in the workspace is the individual sounds and you put them together to a beat? Or are you given whole beats?

Grimes: 00:10:13 Yes. You can construct the sounds from scratch, but you don’t have to. Or you can buy out a sample pack and get sounds, or you can get an 808 or a hardware and make sounds. The way I describe production is, it’s sort of probably like what Mozart used to do except you can actually see it in real time so you don’t have to be able to imagine everything. You’re basically making the whole thing. You’re composing, you’re scoring. You’re writing the melodies, harmonies, whatever. You are making the drums and you construct a piece of music from scratch.

Sean Carroll: 00:10:55 Is it intentional that you start with the beat? Is that a creative choice or it’s just easiest to find the rest of the structure that way?

Grimes: 00:11:05 I just need to know the BPM and I cannot compose to a click. That doesn’t work, so I need to make the beat. I’ll often delete it and make another one that’s better. What I usually do is, I make kind of a crappy one, like a midi. Then I do chords, whatever. Blah, blah, blah. Make sounds. Then I’ll do weird wordless vocals that are like, “Blah, blah, blah, blah.” Then I go back. It’s sort of like a weird religious experience because it feels like, “What’s happening?”

Sean Carroll: 00:11:40 You get in the zone of something.

Grimes: 00:11:40 You get in the zone, it feels like it’s being written for you. Then you go back and then it’s like a weird puzzle that you go back retroactively and solve. Can we change the synth to a much nicer synth or maybe a hardware synth instead of a software synth? Write lyrics that roughly sound like what I was mumbling. That’s kind of how it works.

Grimes: 00:21:10 Mindfulness is way too wholesome, is basically my thought. Mindfulness is not edgy, it’s not cool, it’s not dark.

Sean Carroll: 00:11:59 I can’t help but think of a Jane Austen novel where after dinner, one of the ladies would play piano and sing. Every woman in society had to have those skills and we’ve lost those since there are records and things like that.

Grimes: 00:12:13 That is interesting. Because women used to, or a geisha used to be a thing. Women would be like fucking … Sorry. Am I allowed to swear?

Sean Carroll: 00:12:19 Yeah. We’ll mark it as explicit. Don’t worry. It’s fine. [...]

Sean Carroll: 00:12:34 What I’m wondering is, does this electronics idea help us bring back the idea that we’re all making music? People, it can be much more widespread?

Grimes: 00:12:42 I do think it should be more widespread and the one thing I love about stuff like Audacity or Ableton or whatever exists, super democratizing music. Because for a while it used to be, in the ’70s or whatever, it’s millions of dollars to make an album. It’s really expensive to go to a studio, record on tape, whatever. That’s all really expensive. It’s still, I would say prohibitively expensive at the moment for a lot of people, but in ten years I think everyone will have access to the same tools, basically.

Here's comes an interesting little bit of generational awareness:

Grimes: 00:14:23 That’s another thing I really wanted to talk about, because I feel like we’re in also the end of art. Human art.

Sean Carroll: 00:14:29 Okay? Elaborate on that.

Grimes: 00:14:32 Once AI, once there’s actually AGI, it’s just going to be so much better at making art than us.

Sean Carroll: 00:14:39 Artificial, sorry, AGI?

Grimes: 00:14:41 Artificial General- once there’s not just AI, it can do one task or basic things.

Sean Carroll: 00:14:49 Not just playing chess.

Grimes: 00:14:50 Once AIs can totally master science and art, which could happen in the next ten years. Probably more like 20 or 30 years. I don’t know what you think. I’m actually curious what you think?

In the beginning it was just Artificial Intelligence (AI) and the idea was to craft a full-scope intelligent machine. But all the actual programs were highly specialized. So in the 1990s someone (Ben Goertzel?) coined the term "Artificial General Intelligence" (AGI) to cover that full scope, that is, the scope of what AI had aspired to in the beginning.

Carroll is skeptical about the coming AGI and so am I. But that's beside the point.

And yet, it's interesting to hear how she thinks/feels about it:

Grimes: 00:15:38 I do feel a different strategy … Well, I feel like we’ll probably get to a point where it’ll be building itself and then it’ll be much better building itself than we are, which I think they’re already doing at DeepMind and stuff.

Sean Carroll: 00:15:48 I’ve heard about that. I don’t know. You think that’s a likely thing to happen?

Grimes: 00:15:52 Yeah, because it’ll just be so much better at … It’ll be able to do ten years of work in a day, or whatever. I do think there’ll be also a critical point where there’s a runaway effect and we become irrelevant. Maybe it’s still far away from becoming what it will be, but it’ll just get there.

Sean Carroll: 00:16:08 Well, I know Max Tegmark, a friend of mine who was trained as a cosmologist as I am, but he’s now doing AI research. His project with his students is, create an AI which will act like a theoretical physicist. That is to say, find equations that govern a set of data and it can be done. Art sounds on the one hand, easier to make some art, maybe harder to make great art. I don’t know.

Grimes: 00:16:33 I agree, but I think ultimately AI will get to a point where it will be able to emulate all our hormones, all our feelings, all our emotions. Be able to see great art and it will be able to understand what true innovation is, probably even better than we are. Can.

Sean Carroll: 00:16:46 Maybe, yeah.

Grimes: 00:16:48 I think this is both great and bad, but I think part of the reason it’s great is that I feel like we’re in this amazing time where we might be the last artists ever, which feels fun. [...]

Sean Carroll: 00:18:29 This is what you’re saying. If, on the one hand we’re democratizing our own abilities to make these things, but on the other hand we now have competition.

Grimes: 00:18:39 We’re also going to run out of … I feel like, luckily we’re in this sick spot where technology is allowing for new art forms and stuff. We are running out of ideas in some regards. I feel like AI are going to equally enjoy things in ways that we don’t enjoy things. It will make art for itself and it will just make art forms that are enjoyable in ways that we can’t even conceive of, at the moment. I think that will be the real innovation when it’s not even making art for us anymore. It’s making art for a new set of standards of what is good art because, yeah.

She doesn't offer any technical or quasi-technical argument about why AGI will happen – but then, does anyone? Rather, she thinks it will, it's gotta' happen because it would be so cool. Not a bad attitude, wrong, but not bad.  

The remark I've highlighted in aqua –  It's as though we are corrupt, but the computer will not be, it will rise above. Sounds rather like the Japanese pop culture notion of the Newtype.