Doing Social Media in the AI Era: Your Boss Is More Hopeless Than the Algorithm
The real AI shock to marketing teams comes from the boss: outsourcing intuition to Claude, approving only Copilot-generated proposals, so employees disguise their ideas as Copilot interfaces to get sign-off.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Bosses have outsourced their intuition to Claude
Rachel interviewed about a hundred marketers and found that the main path for AI into teams isn't tool procurement, it's the boss themselves. A social media manager carefully makes a funny meme, sends it to the boss, and the boss throws it into Claude to ask "is this funny?" Even more absurdly, some marketers discovered their boss only approves ideas that come from Copilot, so they first make their own ideas look like a Copilot interface, then send them to the boss for approval. Rachel says the underlying issue is that bosses have outsourced their entire intuitive judgment to AI, and there's no training or guardrails in the team about how to use AI—she contrasts this with her own experience at Condé Nast spending months learning Asana.
— Rachel CartonConsumers don't hate AI, they hate being fooled
Taylor points out an overlooked fact: a lot of YouTubers now directly read AI-generated scripts, and the comments still praise "great analysis, well-written script." In other words, when people don't know they're consuming AI content, they actually like it. What really triggers anger isn't AI itself, but the feeling of "being played." Rachel adds an explanation for this double standard: AI fruit romance memes like Fruit Love Island are low-risk memes, and nobody cares about the craft of a meme; but if a brand you pay money to uses AI to generate a menu, you suspect how much else it's cutting corners on the product. The same person can curse brands for using AI while forwarding an AI fruit meme to a friend.
— Taylor LorenzHow-to creators are the first to be replaced by AI
Rachel divides creators into two categories. Those who built their following on how-to content are most at risk—from the earliest YouTube beauty tutorials onward, media democratization followed this path, but for questions like "how to fix a dishwasher," AI is a better authority, and people will just ask AI. Taylor gives the example of her own mother: she recently proudly fixed a toilet using ChatGPT, which would have been unimaginable before. The other category of creators is actually safe: those who do deep parasocial relationship content, commentary, or things that are hard to replicate can use these tools to make better content. The criterion is whether the content itself is replicable.
— Rachel CartonAnti-data-center and anti-extinction people haven't allied
Taylor observes that people opposed to data centers and people worried about AI causing human extinction are two groups with completely different motivations, and they haven't formed any coherent alliance to date. She even half-jokingly says they should team up, but their interests diverge too much. Her own position is clear: AI won't kill us, "it's actually very hard to kill all humans on Earth, even if we wanted to." She thinks what really deserves scrutiny is how people anthropomorphize AI and grant these products agency—this is especially dangerous in policy debates about accountability, because they are computer programs, not living things. She also quotes a former FTC commissioner: we survived the Black Death, and that only killed 50% of people.
— Taylor LorenzAI accounts are pulling Instagram's slot machine
Taylor thinks Meta is in a tough spot because it's essentially a 2010s social media company, its influence peaked in the 2010s, and Google is better positioned than Meta. In the short term, AI's impact on platforms is about scale and speed: one AI account can post 50 times a day, which is like constantly pulling Instagram's slot machine—it'll hit eventually. She checked the top 20 most-viewed reels at the time, and two or three were AI-generated, and all were very new. One of them used a "behind-the-scenes reveal" framing to show how a film crew shoots ocean waves and how the waves destroy a house, essentially using "proof of authenticity" as a shell to create confusion. Her judgment: this is monetizing confusion, with short-term gains and terrible long-term results.
— Taylor LorenzThreads is the most toxic social media platform
Taylor says she recently posted hot takes calling Threads the most toxic social media platform to date, far more toxic than Twitter—"Twitter compared to Threads is a beautiful wonderland." The mechanism is that the algorithm heavily prioritizes replies, so you get thrown directly into someone's angry reply about something, which in turn feeds the algorithm and quickly escalates into groupthink-style pile-ons. She also points out that Threads' AI-generated news summaries "are just lying, making things up outright." Asked who uses Threads, her answer is her mom.
— Taylor LorenzThe growth secret is turning your account into a series
Asked how to grow followers in the AI era, Rachel says accounts that are growing now all have a clear throughline of familiarity in their content, like they're making their own TV show. She gives the example of a restaurant chef wearing Meta glasses filming himself cooking, who grew to about 500,000 followers in two or three months, with the account name Gil Holy Cooks, posting one episode every day. Taylor adds that content is becoming fully serialized across the board, and you really have to think about how to entertain people; the era of "share your life" is clearly declining, and in 2026 you shouldn't become a lifestyle influencer—there's a lot of anger toward those people.
— Rachel CartonInstagram officially gave a division-of-labor chart for formats
Rachel says Instagram once shared a very useful chart with her readers explaining the purpose of each content format: carousels and reels are for reaching new audiences and get pushed to non-followers, and what the current algorithm really feeds on is non-follower views—if content is only understandable to your own audience, the platform judges it "not entertaining enough" and won't give it main traffic. Single-image posts only serve existing audiences. Stories and DMs are seen by the platform as where community happens. She also quotes Adam Mosseri on Bloomberg Screen Time: "social" on social media is no longer about sharing family photos, but about DMing friends things that remind you of them. She finds this regrettable, but it does clarify what "social" now means on social media.
— Rachel CartonIn their own words · checked verbatim
I wrote an article about this called Does Your Boss Have AI Brain? And I talked to like a hundred marketers who are dealing with an issue where their boss is asking them to use AI and their boss is using AI themselves and they're outsourcing all of their gut instincts to AI.
Rachel Carton4:06
I heard another example of a marketer who their boss only buys into ideas that came from Copilot. And so they actually will mock up their ideas into the Copilot UI to send their boss to get it approved.
Rachel Carton5:09
All these YouTubers are just reading AI-generated scripts now, and it's so... And you look at the comments, people will be like, amazing analysis, great script.
Taylor Lorenz7:14
But also just because something kills us doesn't mean it's smarter than us. You know, like a virus can kill a lot of people. It's not smarter than us.
Taylor Lorenz12:29
And so it was using the framework of proof of reality or BTS to essentially confuse people of how they film something. I think that's really bad to essentially monetize confusion.
Taylor Lorenz20:35
Well, the way that the algorithm works, it really prioritizes replies. And so you'll get sort of dropped off into somebody's really angry reply to something. And that's what fuels it, essentially.
Taylor Lorenz21:35
I think the era of like sharing your life is like very declining. Nobody cares. And I think like, so don't try to be a lifestyle influencer in 2026.
Taylor Lorenz22:36
Figures
| Daily posts by an AI account | 50 | 19:35 |
| Number of AI-generated reels among the top 20 most-viewed | 2-3 | 20:35 |
Glossary
- carousel
- An Instagram multi-image swipe post; the current algorithm uses it to reach new non-follower audiences.
- reel
- Instagram's short video format, which like carousels is mainly pushed to non-followers.
- slop
- Batch-generated, low-quality, quantity-over-quality AI content, such as "Fruit Love Island."
- BTS
- Behind the scenes; videos showing the production process, used by AI to fake authenticity.
How to listen
Brand social media and content marketing practitioners, MCN and creator economy investors, and anyone who wants to know how AI is changing the creative industry from inside organizations.
The first two minutes of guest self-introduction and the Coca-Cola truck bit; jump straight to 04:06.