✘ Mapping the Heart: Is AI-driven media personalization a terrible idea?

- Mocap model blues - Gen Z listening habits - Fair Use questions - the value of music's expanded rights - and the last matriarchy in Europe

AI can make us our own little customized stories, tracks, and images with very little effort. This capability is usually portrayed as a positive development by tech thinkers/moneymen, such as investor Vinod Khosla in his ten predictions late last year. (See #5) Personlization is democratization—another inherent good!—and everyone is a creator. Soon, we will all be able to make our own books, music, films, images, or combinations of all of the above, perfectly suited to our peculiar quirks, kinks, and preferences, without anyone else involved.

But what if this is a terrible idea?

Ad-based culture and dopamine dreams

Personalization has some shady roots. It’s the holy grail of digital marketing, a force to hack the attention economy that forms the foundation of our future culture and entertainment, according to those who love this future, as well as those who despise it. The true currency that runs the world is our minds focused on something for a moment. This focus will lead to some desirable action that ultimately translates into monetary exchange, and that will drive economic vitality, which equals cultural vitality for some. 

At scale, attention can happen in tiny bursts and still work out fine for certain content. Short is actually great from an advertising perspective, as more varied objects of attention can be crammed into the same person’s finite conscious hours. Our ad-based internet optimizes for ads. In music, this attention contraction means shorter and shorter tracks, less and less direct association of a specific artist to their music in fans’ ears, and less deep listening. It’s more about distracting yourself.

Ads aren’t to blame, some claim. There’s something much more nefarious at work. The contraction of attention and shift to distraction is baked into the very design of tech products, both hardware and software. Dopamine is the secret engine of this design. Distraction-seeking activities trigger a little dopamine hit until the brain becomes dependent on this hit. It wants more, much like an addict’s brain, and this spurs us to seek out a culture to satisfy this lust for distraction. As Ted Gioia put it in a recent lament of this dynamic, “Even the dumbest entertainment looks like Shakespeare compared to dopamine culture. You don’t need Hamlet, a photo of a hamburger will suffice. Or a video of somebody twerking, or a pet looking goofy.” But this lament starts from the wrong premise.

Loneliness is the real crisis

The argument is misguided in several ways—and not only because Gioia seems to have missed centuries of the stupid, sensational, and prurient ephemera that pop up when you scratch the right surface of the past. This assertion misunderstands the mind via a reductionist, mechanistic pop-psych model of how our brains actually work. Which is why there is no evidence that “dopamine fasts” actually do anything except sell airport books. 

This is more than a “well, actually” point I’m making. We’re ceding too much power to the Khoslas and Marc Andreessens of the world. In fact, we’ve had a 30-year history of wildly overestimating tech companies’ abilities to manipulate us. We are nudged, perhaps firmly at times, but humans aren’t quite as malleable, and the principles of design and experience as incontrovertible, as many folks invested in propping up the businesses involved would lead us to believe. 

That said, we are facing a real crisis right now, and it is tied to technology. Young people, who in the past tended to be the happiest age group, are reporting more mental health troubles than in previous generations, and there is an association with smartphone use. Part of that increase may be an increase in reporting—it’s more acceptable to talk about depression and anxiety now—but part of it is because as a society we’ve become a lot lonelier.

Big Tech has not conspired to jack our brains. They are simply serving us ads. We’ve formed habits (not addictions!) that conform to the habits of the people around us. These habits, not the tech or content itself, keep us lost in our own little online worlds to the detriment of interacting IRL with other humans. This is a subtle but crucial observation: It’s more about what we don’t do because of smartphones or social media than what we do.

AI-powered personalization will only increase the solitary nature of these worlds. No one else will see or hear the same content as we do. Can we do anything to change this path?

Music is anti-social

Music should be the antidote, right? Music bonds us. It wants to be social! That’s true, but only to an extent. Technology has a role here, too.

There's a reason it’s been hard to create a truly social music experience online despite lots of dedicated attempts over the last few decades. Many of the major tech shifts in music creation and listening have atomized, not united us. 

Let’s look at recording. For most of the 20th century, recorded music experiences were group activities by default. You had to get a bunch of people together in a studio. Many of the recent breakthroughs in music tech were designed to let you make music all alone, using minimal equipment or working totally “in the box.” The legendary bedroom producer doesn’t need any social interaction to make a track. 

Music listening has also leaned away from social experiences. At the dawn of the recorded music era, people gathered around to listen to the radio or records together. Even if you were listening alone, the second someone entered that space, you shared an experience (for better or for worse). Then came headphones in the late 50s, which allowed you to listen in solitude.

Eventually, the Walkman appeared, and you could do this solitary listening anywhere. 

Now that we’ve had decades to use this tech, we’ve developed specific habits. Solitary listening is arguably our most common mode. Many of us purposefully isolate sonically, blocking out the rest of the audible world. 

Even in highly social environments, things have shifted. Arguably, music’s solitary side has invaded its most organically social spaces, live concerts. People gaze at the stage via phones, recording, not interacting with, those around them. The dynamics feel not just solitary but anti-social at times, when fans get so intent on their own activities (grabbing a selfie, yelling lyrics) they invade the space of other music lovers—or even the artist herself. 

Tech has unlocked music’s lonelier modes, and previous social norms surrounding musical experiences do not automatically override these new habits. Customization drills down on the solitary. Once you’ve generated a track to your specs, it’s done. Do you really need to hear it again? Do you want to play it for someone else—and do they want to listen?

We need new habits. If there’s no behavior or psychological pattern hardcoded into our digital spaces and devices, that means we can forge them. We can build something with intention, counterbalance loneliness, and break our solitude.

Breaking the model to map the heart

Creativity comes when things break, at the moment of slippage when the airtight skin around us cracks open. As listeners, we realize we long for something else, or that we’ve found something we didn’t know we sought. As makers, we find something new to say and a way to shake up our listeners.

Because often, we don’t really know what we want. Or we think we do, but we find ourselves listlessly scrolling until something comes along and challenges us, when a track or album upends our understanding of our own capacity to feel or comprehend certain music. We need, in a word, education, not slight preferences or “brand love.” This is a collective process that requires exposure to a wide range of work of varying intent, lineage, era, and medium, whether we like it at first or not. It can happen online if there are diverse enough players involved, but just talking to other Swifties, for example, you’re less likely to find out about June Tabor or Billie Holliday.

In building worlds, finding stories, turning feelings to frequencies and back again, we need the friction of other humans—friction because other humans can be annoying—to come to new art, to appreciate what we’ve never heard before, and to imagine what else we might connect with. 

An LLM can’t do this; only we can. We won’t get connected to something bigger and more compelling by super-serving our own assumptions and unformed desires; that leaves us open to manipulation, to wandering in the desert of lowest-common-denominator content. We need to know how to break the model and map the boundaries of our hearts, where they collide with others’, to make and discover meaningful art.

---

LINKS

💃 AI Is Being Built on Dated, Flawed Motion-Capture Data (Julianne Pepitone)

“Over time, a variety of flawed assumptions have become codified into standards for motion-capture data that’s being used to design some AI-based technologies. These flaws mean that AI-based applications may not be as safe for people who don’t fit a preconceived ‘typical’ body type…”

✘ We’re at a crucial moment for motion capture and its role a building block for media, entertainment and art, and that means we need to talk models and data. This piece makes the excellent point that we need to avoid the errors of the past, when only certain categories of bodies were used to design products.

🎧 How AI is shaping the music listening habits of Gen Z (Beatriz Ilari/Lynn Snyder)

“But what happens when, as we’ve observed, young people don’t know what their peers are listening to? And does it matter that teens aren’t necessarily choosing the music they’re using to understand themselves and the world, let alone that no humans are selecting songs they’re exposed to?”

✘ Music education researchers consider the role of the algo in teen musical identity, and what it means when teens attempt to find themselves via Discovery mode.

📼Fair Use Defenses in Disruptive Technology Cases (Dr. Harrison, Esq)

✘ How can we evaluate AI companies’ claims of fair use when it comes to copyrighted training data? How is generative AI like a Xerox machine or a VHS tape? This piece lays it out in wonderful methodical detail.

🌋Fan economy: expanded rights are worth $3.5 billion, now what? (Mark Mulligan)

“Along with non-DSP and vinyl, expanded rights represent part of the modern music industry’s multi-faceted fan strategy and 2023 was arguably the first year of this new music business era. Streaming is not going away. Indeed, it will be part of this future, but the consumption-focused approach of the 2010s is going to be shunted to the side as fandom takes centre stage.”

✘ $3.5 billion (which is 10% of global music business revenues) is already being generated by “expanded rights” including likeness and merch. This could grow even more rapidly if someone finally unlocks derivative work licensing, the key to many interactive new formats and fan-driven creative ideas. That’s one area that could build up superfans, not just milk them for more money.

👩‍🌾In a Greek Village, One of Europe’s Last Matriarchal Societies Is Near Death (Anna Pantelia)

“The women of Olympos have themselves become a symbol for the island. Upon entering the village, visitors are greeted by a sculpture depicting an Olymbiti woman, her gaze fixed upon the village with the pastel-colored houses and the stone windmills she calls home. This encapsulates the enduring strength of the women in Olympos who, over the years, have been the backbone of both the village and their households.”

✘ Not all tradition is easily sorted into our current political categories or historiographic models. Examples like this one, a story of both adaptation and continuity, point to different possibilities that might inspire us right now.

---

MUSIC

I am obsessed with LA-based composer, singer, and harpist Nailah Hunter and her surreal, deeply emotional soundscapes that bloom on her latest album, Lovegaze. If you need to suspend reality and restore something in your heart, Nailah’s got you.