Bullshit videos: a theory
Or: What happens when AI reveals that nobody wanted this content anyway
Hi everyone —
Hope you’ve had a wonderful break, recharged and reinvigorated to take on 2026 and your big dreams. Reminder to self: the year is already 1% over.
And wow, Efosa has a doozy for you today! Let us know what you think.
Wishing you and your families all the very best for the new year,
—Shamir
co-founder/ceo of Eddie AI
In 2018, late anthropologist David Graeber published Bullshit Jobs: A Theory. His provocation was that, despite massive technological advancement, we never arrived at the 15-hour workweek economists predicted in the 1930s. Instead, we invented entirely new categories of work to keep people busy.
Graeber defined a bullshit job as:
A form of paid employment that is so completely pointless that even the employee cannot justify its existence, even though they feel obliged to pretend that this is not the case.
It’s the receptionist in offices where everyone just waltzes in anyway.
The crisis PR managing crises that don’t exist.
The account manager who emails the client to “check in,” then emails internally to report they “checked in.”
The data analyst paid to Google phone numbers and paste them into a spreadsheet.
That one manager in every organization with a title no one can explain. Not even them.
It isn’t that the work is hard or unpleasant. It’s that it’s useless, and the person doing it knows it.
Graeber was talking about jobs. But jobs produce output. And if bullshit jobs exist, then so do bullshit deliverables: bullshit reports, bullshit presentations, bullshit emails.
And bullshit videos.
The “meet the team” video that’s just headshots and job titles.
The company culture video full of staged smiles and laptops.
The social ad for products no one wants.
Product demos that read spec sheets over stock music.
“Behind the scenes” content that shows nothing behind any scenes.
This sector produces millions of videos a year, employs hundreds of thousands of people, and remains largely invisible to the cultural conversation. Talk to the people making this work. Honestly, after hours, and they’ll tell you:
“I hate this work but what else can I do?”
“Nobody watches this shit, but the client needs it for some reason.”
“I went to film school to tell stories; now I make product demos that get 47 views.”
From Inside the Machine
The contradiction: I get paid to produce commercials. Or corporate. In a previous life, I spent plenty hours editing company culture videos to pay bills too.
I also write for Eddie AI’s substack and it’s building tools like AI logging to automate the repetitive work video editors don’t want to do.
So I’m writing, live and direct, from inside the contradiction because I make bullshit videos too and write for a company automating part of the bullshit-making process. But this, this essay, isn’t that.
It’s a call to arms.
I don’t have easy answers. There’s no how-to’s. Not for me, nor the hundreds of thousands of video people making a living inside this system.
Before we dive into the bullshit, let’s be clear about what isn’t.
Real estate virtual tours that replace hours of travel.
Software tutorials that actually teach people how to use complex tools.
Medical training videos showing surgical procedures.
Safety demonstrations for operating machinery.
Commercials that are genuinely entertaining or informative enough that people choose to watch them.
This kind of video can’t be replaced by an email, a PDF, or nothing at all without real loss.
It exists but it’s a small minority of corporate and commercial video work.
Which raises the deeper question: if so much of this industry produces work that creators resent making and audiences avoid watching, how did it survive for so long?
Metric Theater
The commercial content industry runs on a shared delusion, held together by metrics that measure everything except what matters.
Impressions count how many times an ad was displayed, views count how many times it played for more than 3 seconds, clicks track who clicked, and engagement rate divides likes plus comments by followers.
Each metric creates plausible deniability for everyone in the chain.
The creator can say they made content that got 100,000 views. The client can then say their campaign reached 500,000 people. Then the platform can say their ads deliver measurable engagement.
Yay.
For years, agencies could easily charge $50,000+ (and much more) to produce content that racked up impressive “view counts”, never mind that most were 3-second auto-plays. The system worked because everyone could point to the numbers and justify the spend.
But AI’s here and can generate the exact same 3-second scroll-past for $50. Suddenly the $50,000 wasn’t buying quality or effectiveness. It was buying theater.
AI on Acid
A single video person can now experience a psychedelic expansion of capability, with the ability to create videos that would have been impossible five years ago.
She can make a product demo with multiple camera angles, motion graphics and b-roll she didn’t shoot and didn’t spend thousands for. He can localize a CEO quarterly update into 27 languages with lip-sync in 30 minutes. You can iterate 50 variations of an ad tested against AI-simulated audiences before anything goes live.
The ceiling has disappeared and one person with vision can now do what used to require an army.
It’s exhilarating, disorienting, and terrifying all at once.
But expanded capability carries a side effect familiar to anyone who's tripped: ego death. Except here, it's excuse death. When cost and friction collapse, what’s left is the underlying value of the work itself.
And that’s where the acid test appears:
Can AI generate it for $50 instead of $50,000? Yes.
Can audiences tell the difference in low-attention contexts? No.
Do they care if they can? Not really
If all three are true, the video was bullshit.
The Spike Jonze Gucci film "The Tiger" fails this test, in a good way. It deliberately uses AI’s uncanny limitations as an artistic tool. You can’t generate that vision for $50. Audiences can tell it’s intentional and they’ll care because it’s remarkable.
Crucially, it had a multi-million dollar budget and top director attached so in this context, AI existed within an authored film where attention is already secured.
Most corporate or commercial videos don’t have that protection.
They don’t exist in environments where attention is guaranteed. They don’t play in theaters, screening rooms, or spaces where people have paid to pay attention.
It lives online. Embedded in feeds, autoplaying between posts or fighting for a few seconds before a thumb scrolls.
When a video has 2.5 seconds to register as meaningful, the difference between $50,000 and $50 collapses fast.
A Ghost Economy
Generative AI models are trained on decades of stock footage and corporate video. Archived human labor, stripped of authorship and repurposed through the cloud. Businesses can now generate infinite variations and deploy them at scale.
But we scroll past anyway, conditioned by two decades of interruption advertising teaching us how to not look.
What emerges is a ghost economy: content synthesized from dead labor and pushed toward audiences who either aren’t there or aren’t looking.
It’s ghosts selling to another species of ghosts.
Try this:
Open Instagram. Right now. Scroll for 30 seconds. Close it.
What did you actually see?
I had to scroll past seven posts before seeing anything resembling a human I actually know.
Posting Zero
Everything is here, now, endlessly, and near-indistinguishable from human-made work but just as forgettable.
The cognitive load is becoming unbearable. People can’t tell what’s real anymore and they’re exhausted from trying.
In his New Yorker essay, writer Kyle Chayka described this tipping point as “Posting Zero”: the moment ordinary people stop sharing their lives online because the effort no longer feels meaningful.
Recent surveys point in the same direction, with large portions of users posting less than they did in previous years. Instead, we’re retreating to group chats, DMs, newsletters (hi), and subreddits. Spaces that feel less like a marketing feed and more like a place to actually exist.
If there’s no guarantee your community will even see what you post, what’s the incentive? How social really is this media now?
As Chayka went on to warn, if we reach Posting Zero:
What we’re left with is dry corporate marketing, A.I.-generated slop, and dreck from thirsty hustlers attempting to monetize a dwindling audience of voyeurs.
Posting Zero feels like the fully realized version of what for years circulated as the so-called “dead internet theory”: the idea that most online content is synthetic, automated, or performative rather than genuinely human. What began as fringe messageboard paranoia hardened into mainstream anxiety within just a few years because the evidence stopped being hypothetical.
This ghost economy goes far beyond displacing workers as it makes audiences revolt against being marketed to this way at all.
That’s the real disruption of AI.
Not collapsing costs, but collapsing the system by flooding it with so much cheap content that we’re finally forced to look at modern advertising and confront a question we’ve avoided for twenty years:
Did we ever want any of this?
If the feed is now dominated by ghosts, the value judgment flips. Value no longer comes from polish, scale, or consistency. It comes from evidence that there is a human on the other side after all.
Usually in the form of a simple signal:
proof of life
/pro͞of əv līf/
noun phrase
1. Original definition Evidence that a person (especially a hostage or kidnapping victim) is still alive, typically provided during negotiations.
“The kidnappers sent a proof of life video.”
2. Internet slang A post, message, or content that demonstrates someone is still active online after a period of absence; evidence of continued digital existence.
“After three months of radio silence, she finally dropped a proof of life selfie in the group chat.”
3. Contemporary usage (2024–) In the context of AI-generated content, markers or signals that indicate human authorship; distinguishing features that separate authentic human work from synthetic or automated content. Includes mistaks, creative risks, or evidence that someone made choices and cared about the outcome.
No More Bull
In 2003, legendary marketer Seth Godin published Purple Cow, arguing that traditional marketing was already dead. His premise was that, in a field of brown cows, you need to be purple. You needed to be remarkable.
One of his favorite examples was Chris Bianco in Phoenix, Arizona, who milled his own flour when it was cheaper not to, rejected tomato shipments that didn’t meet his standards, obsessed over fermentation timing instead of following fixed recipes, and refused to scale for years.
All of this was friction.
People now travel far and wide to queue two hours for his pizza.
Pizza that earned him the James Beard Award for Outstanding Restaurateur aka the culinary academy Award.
The friction was the point because it filtered for people who cared.
So that’s the question facing businesses now: what friction will you embrace once audiences start boycotting ghost work entirely?
Because the creators who survive this reckoning won’t be those making better bullshit for brands that don’t care. They’ll be working with the rare clients who’ve already realized the ghost economy is dead, and only the remarkable can live on.
I don’t know how many of the 500,000 people making corporate video worldwide will successfully transition. Some will find clients who value friction. Some will pivot to work that matters to them. Some will have to find something else.
The timeline is uncertain, as is what that “something else” looks like.
What comes next won’t be easy, and it won’t be quick. And it certainly won’t save everyone.
But hopefully, it won’t be bullshit.
Devastating breakdown of the ghost economy. The three-question acid test perfectly captures why so much corporate video felt hollow before AI made it explicit. I worked on a few of those "check in to say we checked in" projects early in my career and always felt gross about it but couldn't articulate why. The Chris Bianco pizza example is perfect because it shows friction as signal rather than bug,proof someone actually cares rather than just executing a formula. That shift from metric theater to proof of life feels like a fundamental change in what audiences will tolerate going forward tho I wonder if brands will actualy adapt or just find new ways to simulate authenticity.