73 | Grimes (c) on Music, Creativity, and Digital Personae
What this covers
Sean Carroll hosts electronic artist Grimes to discuss how contemporary music production works and what technology means for the future of human creativity. The conversation moves outward from the mechanics of modern songwriting—how DAWs) like Ableton Live allow one person to write, arrange, and produce a finished piece—into broader questions about artificial intelligence, the nature of digital identity, and whether human artists face imminent obsolescence. Grimes brings both technical fluency and philosophical ambition to the discussion; Carroll anchors it with skepticism drawn from work on AI and complex systems.
The episode splits into three distinct territories. First, music and tools: Grimes argues that affordable software is democratizing production toward a new golden age, while noting that mastery of existing forms must precede rule-breaking innovation. Second, artificial intelligence and the future: both speakers explore whether AGI) will render human art-making obsolete within decades—Grimes leans toward yes, pointing to DeepMind's trajectory, while Carroll emphasizes that current AI lacks embodied motivation. Third, digital personae and psychology: Grimes contends that humans now inhabit two separate selves)—real and digital—and that consciously embracing this split through avatars and masks is healthier than chasing unified authenticity. She sketches emerging possibilities in digital fashion and explores her new album Miss Anthropocene as a deliberate embodiment of villainy rather than heroism.
Grimes argues that technology is simultaneously democratizing art-making and rendering human artistry obsolete, while also splitting the human psyche into real and digital selves that should be deliberately separated and embraced rather than reconciled.
- Digital audio tools are democratizing music production toward a new golden age of art
- AGI will eventually surpass humans at art, making current artists possibly 'the last artists'
- Humans now have two selves (real and digital), and consciously separating them via avatars can be freeing and creatively healthy
AI systems will demonstrate genuine creative capability across artistic domains including painting, writing, and unconceived art forms.
- When computers beat the best human Go players, they didn't just get better — they played differently and artistically, coming up with moves the lifelong devotees had never imagined, which is evidence that AI will be good at painting, writing, and art forms not yet conceived.
“They were devastated when the computer beat them, because not only did it get better, but it played differently. It came up with things that they had never imagined.”
- Douglas Hofstadter ran an experiment where a music-writing AI emulated Chopin, and when played for classical music fans alongside an early juvenile Chopin piece, everyone voted the computer's piece as more Chopin-like, illustrating that AI can already convincingly emulate human artistic style.
“Played that and an early juvenile piece by Chopin and said, "Which one sounds better?" Everyone voted for the computer to be more Chopin-like.”
Humans should embrace contextual mask-wearing and manage digital identity thoughtfully rather than pursuing a single fixed authentic self.
- Carroll makes a conscious decision to keep his digital/online self pretty close to his real self but not exactly, a choice Grimes remarked upon because most people's online and real selves diverge significantly.
“I actually make a sort of conscious decision to make my digital self pretty close to my real self, but not exactly.”
- Rather than always trying to be one's 'true authentic self,' the better approach (from a Pirandello reading) is to wear masks proudly — to wear the right mask for the situation and be aware of what you're doing — because humans are contextual beings whose personality genuinely differs with different people, and this is an authentic survival mechanism, not inauthenticity.
“what he actually went for was, "Wear those masks proudly. Wear the right masks and be aware of what you're doing."”
Current AI fundamentally lacks embodiment and biological motivation, so it will not spontaneously pursue knowledge without external command.
- Current AI is fundamentally missing embodiment, which matters because being in a body — getting hungry, sleepy, decaying without inputs, kept in homeostasis by the outside world — is the source of motivation; without motivation, an AI will just sit in whatever state it's in until commanded, and it will not spontaneously want to know the nature of the universe.
“one thing that current attempts at AI are really missing is that embodiment. AIs don't get hungry, they don't get sleepy.”
Mind-machine integration and direct cognitive enhancement will eventually become viable paths alongside standalone AI development.
- When giving a talk or seminar, one should think about what people in the room will actually remember five years later — and since the honest answer is usually 'nothing,' it reframes the purpose of the performance and what is worth conveying.
“when they give a seminar, think about what people in the room will remember five years later. And they instantly go like, "I'm not going to remember anything five years from now."”
- In addition to building separate AIs, there is the mind-computer interface path where rather than creating standalone AIs we will modify ourselves by merging mind and machine — for example to absorb a book's content directly — which Carroll believes is coming.
“in addition to AI, there's the whole mind-computer interface thing... where rather than building separate AIs, we're going to modify ourselves.”