AI Ads, Controversy, and Taylor Swift
Key Points
- AI‑driven marketing is booming, with AI now powering roughly 40% of Instagram feeds and companies like Meta investing billions in large‑scale models to tailor video ads.
- Brands are increasingly mixing real talent with AI‑generated elements—as illustrated by the Sydney Sweeney ad where a car scene was fabricated—to spark controversy and stand out in crowded spaces.
- Historical controversial campaigns (e.g., 1990s Brooke Shields Calvin Klein ads) show that “gen‑ads” must provoke debate because the product category is ubiquitous and hard to differentiate.
- High‑profile deep‑fake incidents involving Taylor Swift are being used by journalists and researchers as benchmark tests for AI safety, highlighting the rapid spread of synthetic celebrity content.
- The convergence of cheap, powerful AI creation tools and a willingness to leverage controversy creates a feedback loop that pushes advertisers toward constantly seeking new “local maxima” in consumer attention.
Sections
- AI‑Driven Ads and Controversy - The speaker outlines the rapid rise of AI in marketing, highlights Meta’s billion‑parameter model and AI‑generated imagery, and uses the Sydney Sweeney ad controversy to illustrate how brands deliberately leverage AI‑created content for attention.
- AI‑Generated Media and Marketing Dilemma - The speaker explains how AI‑created images like synthetic Taylor Swift concert shots blur the line between real and fake content, creating an uncanny‑valley challenge for marketers and prompting a discussion of the system’s stagnation and potential solutions.
- AI-Driven Parasocial Feedback Loops - The speaker explains how algorithmic content delivery and AI companionship apps generate layered feedback loops that alter user behavior and expectations while fostering intensified parasocial relationships with both real celebrities and AI‑created characters.
- AI Engagement vs. Authenticity in Marketing - The speaker argues that AI-driven engagement systems reshape human psychology, creating tension with brands' demand for authenticity, and suggests emphasizing inherently un‑AI‑replicable experiences—illustrated by Taylor Swift’s anti‑AI album reveal and vinyl pre‑sales.
- Experiential Retail Beats Controversy Marketing - The speaker contends that lasting customer value and cash flow stem from unique, in‑person retail experiences rather than short‑lived controversy‑driven hype, emphasizing the need to translate AI‑generated attention into tangible store engagement.
- Strategic AI Integration in Branding - The speaker urges marketers and technology builders to deliberately weave AI—particularly LLMs—into brand strategy, moving beyond algorithmic limits, guardrails, and local maxima to thoughtfully select ideas and cultivate long‑term brand relationships.
Full Transcript
# AI Ads, Controversy, and Taylor Swift **Source:** [https://www.youtube.com/watch?v=yqmdJ2lQw98](https://www.youtube.com/watch?v=yqmdJ2lQw98) **Duration:** 00:20:15 ## Summary - AI‑driven marketing is booming, with AI now powering roughly 40% of Instagram feeds and companies like Meta investing billions in large‑scale models to tailor video ads. - Brands are increasingly mixing real talent with AI‑generated elements—as illustrated by the Sydney Sweeney ad where a car scene was fabricated—to spark controversy and stand out in crowded spaces. - Historical controversial campaigns (e.g., 1990s Brooke Shields Calvin Klein ads) show that “gen‑ads” must provoke debate because the product category is ubiquitous and hard to differentiate. - High‑profile deep‑fake incidents involving Taylor Swift are being used by journalists and researchers as benchmark tests for AI safety, highlighting the rapid spread of synthetic celebrity content. - The convergence of cheap, powerful AI creation tools and a willingness to leverage controversy creates a feedback loop that pushes advertisers toward constantly seeking new “local maxima” in consumer attention. ## Sections - [00:00:00](https://www.youtube.com/watch?v=yqmdJ2lQw98&t=0s) **AI‑Driven Ads and Controversy** - The speaker outlines the rapid rise of AI in marketing, highlights Meta’s billion‑parameter model and AI‑generated imagery, and uses the Sydney Sweeney ad controversy to illustrate how brands deliberately leverage AI‑created content for attention. - [00:04:29](https://www.youtube.com/watch?v=yqmdJ2lQw98&t=269s) **AI‑Generated Media and Marketing Dilemma** - The speaker explains how AI‑created images like synthetic Taylor Swift concert shots blur the line between real and fake content, creating an uncanny‑valley challenge for marketers and prompting a discussion of the system’s stagnation and potential solutions. - [00:07:35](https://www.youtube.com/watch?v=yqmdJ2lQw98&t=455s) **AI-Driven Parasocial Feedback Loops** - The speaker explains how algorithmic content delivery and AI companionship apps generate layered feedback loops that alter user behavior and expectations while fostering intensified parasocial relationships with both real celebrities and AI‑created characters. - [00:11:19](https://www.youtube.com/watch?v=yqmdJ2lQw98&t=679s) **AI Engagement vs. Authenticity in Marketing** - The speaker argues that AI-driven engagement systems reshape human psychology, creating tension with brands' demand for authenticity, and suggests emphasizing inherently un‑AI‑replicable experiences—illustrated by Taylor Swift’s anti‑AI album reveal and vinyl pre‑sales. - [00:15:21](https://www.youtube.com/watch?v=yqmdJ2lQw98&t=921s) **Experiential Retail Beats Controversy Marketing** - The speaker contends that lasting customer value and cash flow stem from unique, in‑person retail experiences rather than short‑lived controversy‑driven hype, emphasizing the need to translate AI‑generated attention into tangible store engagement. - [00:19:03](https://www.youtube.com/watch?v=yqmdJ2lQw98&t=1143s) **Strategic AI Integration in Branding** - The speaker urges marketers and technology builders to deliberately weave AI—particularly LLMs—into brand strategy, moving beyond algorithmic limits, guardrails, and local maxima to thoughtfully select ideas and cultivate long‑term brand relationships. ## Full Transcript
All right, stay with me. We're going to
talk about local maxima algorithms, AI,
and Taylor Swift. I'm going to make it
all make sense. I promise. First things
first, I think everybody knows that AI
marketing is exploding. Something like
40% of the Instagram feed is AI now.
It's growing really fast. Meta is
investing heavily in AI tooling. Meta
just published a 1 billion parameter
artificial brain that's designed to
measure your response to videos and
simulate it so they can build more AI
videos for you. And it's not just the
tooling that's changing. It's also
actual ads that we see coming out all
the time that are controversial or AI
generated. And it's not even just the ad
space. It's also AI generated imagery as
a whole is exploding frankly because the
tools are getting cheaper and better.
I'm going to go through just a couple of
use cases to kind of freshen you up.
None of this will be surprising and then
I'm going to explain how it all comes
together. So, first the Sydney Sweeney
case, right? This was a famous case
where Sydney was a real person in an ad,
but certain components of the ad were AI
generated because it was cheaper. I
believe the car driving away was AI
generated toward the end of the ad. The
point here is that the brand chose to
lean into controversy and they've
admitted that, right? They chose to lean
into a controversial stance with their
with their particular angle on the GAN
ad. And they chose to lean into classic
controversial jeans ads from the past.
The Brook Shields ads from the 1990s
were deliberately evoked. We're not
here. I am not going to break down
marketing history for you. If you want
to look up the Brook Shields Calvin
Klein ads, it's a very famous example of
the controversy that genes ads have
historically done, which in and of
themselves are interesting because genes
are ubiquitous. Everyone has genes. And
so having to market them means you sort
of have to step into controversy to try
and differentiate yourself in the space.
And this is exactly what we're going to
dig into more as we get into this idea
of local maxima. The second thing I want
you to keep in mind, now we're getting
to Taylor Swift, is this idea of Taylor
Swift deep fakes. Now, deep fakes are
not new. They're not certainly not
limited to Taylor Swift, but because
Taylor Swift is a famous woman
celebrity, we get in a special case, a
special prevalence of deep fakes,
special visibility on deep fakes,
journalists tend to use Taylor Swift as
a test case when they are investigating
the safety of AI systems. And so that's
why we get an immediate test of Grock's
imagine feature and an immediate sort of
news headline around the world that
Grock's imagine feature will create
adult images of Taylor Swift as deep
fakes. You might wonder now where is
this going with ads? I'm sharing this
part because we are fundamentally in a
situation where the technology to create
ads and the technology to create
controversy, which is exactly what we
got at with the Sydney Sweeney case, are
converging and cheapening. In other
words, part of why we're at a local
maxima, is that the technology to create
this kind of product is cheaper and
cheaper and cheaper and cheaper. digital
ads that would have cost, you know, a
million dollars a couple of years ago
are a tenth of the price or less now and
they're going down further. We are going
to see more and more major brands jump
on the AI generated ad train. I am I am
fully prepared for a doubledigit share
of the ads in the Super Bowl in 2026
being AI generated at least in part. So,
there's a third piece here that I want
to get at before we start to dive into
the relationship between all of these
components. And the third piece is the
idea of the uncanny valley. Essentially,
you need to combine the controversy of
Sydney Sweeney, the ease of production,
and the parasocial nature of celebrity
relationships, which enables the
creation of deep fakes of Taylor Swift
and other celebrities, with the notion
that we can't tell the difference
anymore. And this is something that's
true even of Gen Z. It's not really an
agist thing. Gen Z in study after study
after study professes to care about
authenticity. But the reality is they
also struggle in study after study to
tell the difference between good AI
generated material and real material.
And I will tell you as a part of the
test case here, I went to midjourney to
create images of Taylor Swift in
concert. They were not inappropriate
images. They were just like Taylor Swift
in concert images. And if I had been
shown those images, I would not be able
to tell you if it was an artistic shot
by a photographer from a real Taylor
Swift concert or if it was made up. And
I think that that's part of what's going
on that makes it challenging here is
that we ourselves are unable to
distinguish artificial tokens from real
tokens in ad situations or other
situations today. It's true in words
too, but we're talking in marketing and
images are central to marketing. So,
we're going to stay focused on that for
the day. So, those are the three pieces
I want to bring together. We have Sydney
Sweeney, we have Taylor Swift, and we
have this idea of the uncanny valley.
Where does this take us? Fundamentally,
I want to suggest to you that we are
living in an evolved system that has
unintentionally
evolved to a local maxima. And I want to
walk through the technical reasons for
that. And then I want to walk through
because I'm Nate and I like practical
solutions. Some ways forward out of that
local maxima. So first things first,
what are the system dynamics that we're
all living in for marketing? And this
includes marketers. My job here is not
to blame marketers. It's actually to
talk about the system as a whole and how
AI is accelerating it. AI systems are
creating evolutionary pressure on human
content creation. So artificial
authenticity is going to out compete
genuine authenticity in algorithmic
environments. In other words, AI is an
accelerant on top of the social
algorithms that we have built and
optimized through social networks over
the last 20 years. This includes at
least four individual feedback loops
that are all accelerating. One is
sampling feedback loops. So,
controversial content tends to get more
distribution and creates training data
that biases towards more controversy.
That's the Sydney Sweeney use case.
Essentially,
the feedback loop samples and gets an
idea of what works well. Controversial
content, it's well known, does better in
the feed, and so it generates more
controversial content to accelerate
that, right? I we have talked about this
broadly as a society. We haven't
necessarily talked about the idea that
AI is an accelerant for this. AI
accelerates the uh production of
controversial content by making it
easier and cheaper cheaper to produce
and that in turn reinforces that
training data loop for the algorithms.
The second one is
feature feedback loops are a problem. So
when a user engages with content they
create a feature in the algorithm. User
engagement with manipulative content
will teach the algorithms that
manipulation equals quality. That one is
also not new as a feedback loop. But
again, AI is accelerating it because AI
is essentially able to produce content
that is more likely to be engaged with.
It's content that is more likely to be
latched on to by humans. And humans
can't tell the difference. This gets
back to the idea of the uncanny valley.
We can't tell. And so we click and we
engage and we're feeding both the
content algorithm that feeds us stuff
and we're also teaching the marketers
that AI works well. The third feedback
loop is an individual feedback loop for
us. Repeated exposure to optimized
content changes our behavior and
expectations. There's been a lot of
studies around sort of what the
Instagram feed has done to our
self-image and our social relationships.
I think we can also talk about this idea
of parasocial relationships here. A lot
of the Taylor Swift phenomenon that I
discussed is this idea that Taylor Swift
is not just a celebrity, but Taylor
Swift is someone with whom I have a
right to have a relationship, even if
it's a madeup relationship in my head.
So, this gets back to the number that I
gave you at the very beginning of this
video where we talked about Character AI
and other sort of AI uh companionships
apps and the and the money that they're
generating$ 220 million some dollars per
year growing very fast. They're
cultivating the idea of parasocial
relationships, but they let you create
the character, right? They let you
create the character you're going to
have an artificial relationship with.
Whereas traditional parasocial
relationships, it's your imagined
relationship with the celebrity. In both
cases, AI accelerates this individual
feedback loop in your own head where you
are changing your behavior and
expectations because of what you're
engaging with. The fourth feedback loop
is outcomes. Essentially, successful AI
content that runs through these feedback
loops becomes the baseline for normal
content. And so, in a sense, it's not
just that the AI content is changing the
way we consume behavior and changing
what is in our feeds. It is also
changing how we make real video content.
There's been a lot of work done on how
Hollywood movie making has shifted over
the last 20 25 years with CGI. In the
same way, the way we make real content,
whether it's movies or ads or what have
you, marketing assets, is changing
because of AI. I don't think we fully
process this, but we are going to see
camera angle shift. We're going to see
expectations for what you can do with
special effects. We're going to see
expectations for how you bring in people
versus obviously not people versus
people who are deep fakes, even if
they're not real humans, but they they
look so much like humans, we can't tell
the difference. That we are reshaping
the baseline of normal in our own movies
and ads by what we interact with. So
these four feedback loops, the idea that
controversial content gets more
distribution, the idea that user
engagement with manipulative content
teaches the algorithms that manipulation
is quality, the idea that repeated
exposure of ourselves to optimized
content or AI optimized content changes
our own behavior, and the idea that
outcome shaping shapes our expectations
of even nonAI movie making and nonAI
admitt, nonAI marketing. All of these
are feeding on each other and being
accelerated by AI and that is producing
emergent behaviors that nobody is
programmed for. I am not a big believer
in the idea that there is a bunch of
evil marketers somewhere who are all
deciding to sort of poison America or
poison the world with their ads. That is
a popular misconception. I have been a
marketer. I have worked with amazing
marketing teams. That is just not how
marketers actually work. Instead, I
think it's more useful to think about
emergent behaviors in a system that is,
as I've said from the beginning of this
video, locally optimized. It's a local
maxima system. So, in this case, what we
have is emergent optimization
designed effectively to adapt the human
psychological condition to AI content.
I'll say that one more time. You have AI
systems that are optimizing for
engagement and they inadvertently are
optimizing human psychology for
engagement with AI content. That is the
through line that ties together Sydney
Sweeney and Taylor Swift and this idea
of the uncanny valley parasocial
relationships. It's all around this core
thesis. Fundamentally, our AI systems
because we're building them on top of
optimized systems for engagement.
They're essentially adapting our
psychology to engage with AI content.
This then creates a problem because we
say as marketers and as brands that we
value authenticity. The humans buying
the product say they value authenticity.
In between disintermediating that is an
AI marketing system that values
artificiality. It values what's fake.
What do we do about that? How do we
optimize there? My suggestion to you is
that the way through this is going to be
for brands to emphasize what cannot be
faked with AI. I think the music
industry is a really interesting example
of this. Let's look at Taylor Swift and
her launch of her Orange album, right?
The new album that's coming out, TS12,
right? When Taylor Swift announced that
on August 12th, she blurred out the
image of the album in order to prevent
any issues with uh sort of album faking,
etc. from AI content generation. But
it's not just a defensive play. She's
also pre-elling physical media. And
she's not the first musician to do this.
Vinyl is making a huge comeback. She's
pre-selling the vinyls for the album.
You know what's interesting about
vinyls? You can't fake vinyls. They're
real. You can touch them. You can hold
them. That is part of why they're making
a comeback. My suspicion, my strong
suspicion is that one of the ways that
good brands can cut through the noise at
this point is by doubling down on
physical experiences that humans cannot
get any other way. You cannot have an AI
that gives you a vinyl of the Taylor
Swift album in the same way that Taylor
mailing you the album on the correct
date is going to get you. that that is
an irreplaceable human experience to
open the package and get the vinyl.
Well, that's a way through. That's a way
to keep authenticity that holds
long-term customer value in a world
where there are going to be so many
brands demanding your attention. And
that, by the way, is one of the reasons
I don't think this local maxima will
hold forever. If you're wondering, well,
should I throw up my hands? Should I
just despair? One, that's not what I do
on this channel, so go somewhere else.
And two, I don't think so. Because at
the end of the day, AI intelligence and
AI tooling is continuing to get cheaper,
which means every brand is going to be
able to make these, you know, fantastic,
weird videos all the time. We're going
to get a tremendous amount of content
flooding the zone that is all very high
quality and all very AI and it's going
to be so much that people are not going
to be able to consume it all. You're
going to have a massive signal issue.
People won't be able to find signal to
noise. So, what does a good brand do to
stand out? What do good marketing teams
do to stand out in the age of AI
optimized and accelerated algorithmic
relationships? Well, good brands double
down on actually giving you authenticity
that you can't get anywhere else. And
that's why I go back to physical
product, physical goods, physical
experiences, popups, events, things that
people can be in the space on that you
can't copy with AI. They can be very
Instagrammable moments. You can have the
sort of the Instagram wall with like the
special logo or whatever it is. You can
have fun with it. You can make it sort
of hybrid and online at the same time,
but you still, I think, are going to
need to have that irreplaceable human
touch. And that's why you have examples
like uh whiskey distilleries and other
things like going back to getting
physically close to their customers,
offering customer tours of the
distillery, offering special tasting
rooms, this that and the other thing.
And that's a luxury good. You might
think, well, if you're not in the luxury
goods space, how are you going to
replicate that? I think that that is
where physical store footprints are
going. We know that the mall is not
really a thing in the US anymore. We
know that the controversy that Sydney
Sweeney courted with the jeans ad was
primarily designed to drive attention,
drive traffic to the American Eagle
site, which would then convert. If
that's what happened, you want to be in
a position where you can give people
experiences that are in person that
leave them feeling like they want to
come back for another experience like
that that they can't get anywhere else.
Because part of the problem with
marketing for controversy is it buys you
the pop in the stock price. It buys you
the attention. And I'm not going to
pretend it doesn't buy you dollars. I'm
sure it bought them dollars, right?
Let's not let's not kid ourselves.
Controversy works, but it doesn't
necessarily translate well into
long-term customer value. And if you
want long-term customer value, which
ultimately sustains free cash flows and
sustains the valuation of the business,
I think you have to figure out how to
get people from AI spaces into online
spaces. And by the way, this is not me
saying that AI ads are bad or that AI
ads won't exist or that we shouldn't
have them. I think the reality is
everyone's going to be able to make
their own AI ads very very quickly.
We're just going to have them. Instead,
I want to think about how AI ads can tie
us into physical spaces that allow us to
engage with our whole senses, right? Our
our smell, our sense of touch, all of
the things that make us register real
experiences, maybe that help us build
community for the brand around with
other humans who are consumers of the
brand. That's going to be what builds
those long-term customer relationships.
So I think that like piece one is
getting physical. I want to suggest to
you another piece that I think will
help. To me, one of the things that we
are going to start to see as consumers
start to demand authenticity as a
differentiator is we're going to start
to see some kind of certificate of human
touch in some of these brands offerings.
I don't know exactly what that will look
like. This is a little bit of a peak
around the corner, but you can take a
you can take a tour through how the
beauty industry has handled authenticity
and beauty over the last 20 years and
the way marketing in the beauty industry
has shifted as a result. Similarly, I
think we're going to see marketing that
leans into authentically portrayed human
experiences. Marketing that says this is
a real human. We didn't AI their face.
Marketing that says this is a real car.
Look, a piece of it just fell off. we
didn't pretend. You're going to get more
and more of that as a way to signal to
consumers in a sea of AI content that
you are worth paying attention to. In a
sense, there's going to be an anti-AII
signal that is going to come up because
AI content is going to become so cheap
and easy to produce. I want to close
with a question for AI tool builders. If
you're building AI tools for marketers
in the space, most of the time you have
been building for the idea for the tool
chain to produce these ads. You've been
building for a cleaner, simpler pathway
from idea to ad. That's fine. I think
that pathway is fairly well trodden. I
want to suggest there's a lot of untaken
space with AI in helping marketers
exercise good taste and judgment over
what is the best long-term positioning
for the brand. What is the best
long-term positioning for a particular
ad campaign that supports the brand?
There has been precious little AI work
put into the connection between content
branding and long-term strategy. And
frankly, LLMs are getting smart enough
that they can at least be thought
partners on this. And I don't see
anybody thinking about what you do when
AI can produce a thousand ideas and you
have to pick one. Who's doing that part?
And so that's the challenge I want to
leave you with. I think we can build our
way out of and we can market our way out
of the local maxima we're in. We don't
have to tolerate the experience that
we're having now in our algorithms. We
can build our way out. We don't have to
tolerate a world where everybody can be
deep fake, not just celebrities, not
just Taylor Swift. We can put guard
rails in place. But the key to doing
that is our willingness to think more
deliberately about how we want AI to
weave into our marketing. And that's the
heart of this question. If we have a
local maxima, we have to be more
intentional with our AI usage and the
way we cultivate long-term relationships
with the brands. And that's up to us as
consumers to demand. It's also up to
marketers and it's up to builders who
equip marketers with tools to think more
intentionally. And that is the heart of
what I have to talk about