A worker who served as a fallback for a real estate AI chatbot for $25/hour details the experience of pretending to be a bot as part of a sixty-strong team
a human pretending to be a robot pretending to be a human! https://www.nplusonemag.com/ ... @frankpasquale : “I wasn't really training Brenda to think like a human. Brenda was training me to think like a bot.” https://www.theguardian.com/ ... Nilanjana Roy / @nilanjanaroy : “Most of us were poets and writers with MFAs, but there were also PhDs, as well as a number of opera singers, another demographic evidently well suited for chatbot impersonation—or, I suppose, for impersonating a chatbot that's impersonating a person.” https://www.nplusonemag.com/ ... Anne-Sophie Bolon / @asbolon : Most of them were poets and writers with MFAs, but there were also PhDs in performance studies and comparative literature, as well as opera singers. All people well suited for impersonating a chatbot that's impersonating a person. https://www.nplusonemag.com/ ... Charles Arthur / @charlesarthur : This piece by @laurahpreston is simply fantastic. Must-read. The original at @nplusonemag has a touch more detail and they'd love to have more readers! https://twitter.com/... Frank Sennett / @sennettreport : Why don't you just tell me the name of the movie you've selected?" https://www.theguardian.com/ ... @nplusonemag : “Never before had I uttered the construction off-site leasing specialist, but this was what Brenda called herself, and now it rolled off my tongue with ease.” https://www.nplusonemag.com/ ... Alex Press / @alexnpress : “Brenda's corporate clients were self-satisfied knowing they had not replaced their phone lines with a customer-service bot. What they were using, instead, was cutting-edge AI backed by PhDs in literature.” https://www.nplusonemag.com/ ... Mark Krotov / @markkrotov : “Sometimes I would jolt awake at night, my heart racing, with no thought at all but Legacy Lofts on Main.” Absolutely classic essay. https://twitter.com/... Dayna Tortorici / @dtortorici : It's also a tour of the grim American real estate market. Brenda, the bot who pretends to be a local leasing agent, answers questions about properties across the US; Laura, the woman behind the bot, answers questions that are too sensitive or human for the AI to handle, e.g.: https://twitter.com/... @nplusonemag : That August, Doja Cat posted her personal cell phone number on Instagram. The number was one digit away from one of Brenda's properties. “DOJI!!!” screamed the onslaught of fans. “I'm Brenda!” I replied. https://www.nplusonemag.com/ ... @longreads : “Brenda, the recruiter told me, was a sophisticated conversationalist, so fluent that most people who encountered her took her to be human. But like all conversational AIs, she had some shortcomings.” #longreads @laurahpreston @nplusonemag https://www.nplusonemag.com/ ... Andrew Revkin / @revkin : This is a surreally disorienting story by @laurahpreston: ~For one weird year, I was the human who stepped in to make sure a property chatbot didn't blow its cover - I was a person pretending to be a computer pretending to be a person...~ https://twitter.com/... @nplusonemag : A typical encounter with Brenda began when a prospect saw an apartment on an online real estate marketplace like Zillow. The listing provided a phone number; the prospect dialed it. Unbeknownst to the prospect, this phone number was a sham. https://www.nplusonemag.com/ ... Emily Gorcenski / @emilygorcenski : “It occurred to me that I wasn't really training Brenda to think like a human, Brenda was training me to think like a bot, and perhaps that had been the point all along.” https://www.theguardian.com/ ... Mark Krotov / @markkrotov : One of the very best essays I've read this year is @laurahpreston's funny and sad account of her time in the real-estate content mines as a “human fallback” for an AI chatbot that—as it turned out—had much to learn about the subtleties of customer service. https://www.nplusonemag.com/ ... Joe McFadden / @joemcfadden : This is a fascinating and beautifully written piece, well worth your time. One intriguing detail is that opera singers are well suited to impersonating chatbots! Via @charlesarthur and his always excellent newsletter. https://twitter.com/...
Context & Ripple Effects
Before chatbots could plausibly hold a conversation, some companies simply put people behind them. The n+1 essay describes Brenda, a real estate chatbot propped up by a sixty-strong team of $25/hour fallbacks — many poets, PhDs, and opera singers — who stepped in when the AI failed and used sham phone numbers so prospects never knew a human had taken over.
The piece lands at the start of a wave of first-person accounts of AI-adjacent labor: an [[a:842344|AT&T call center worker later described generating transcripts with AI and wondering if she was training her replacement]], while computational linguist Emily M. Bender's critique of what language models actually do sharpened the question of whether a bot like Brenda was ever doing the work at all.
First-order effects
- Property prospects contacting Brenda were misled about whether they were speaking to software or a person, since the system masked human interventions behind fake phone numbers.
- The sixty fallback workers themselves — writers and academics at $25/hour — absorbed the conversational load the AI couldn't carry, doing emotional and improvisational labor under a machine's name.
Second-order effects
- Rivals selling 'AI' leasing assistants face a credibility problem: if Brenda needed sixty humans, buyers of similar tools will press vendors on how much of the work is actually automated.
- The same concealment pattern invites disclosure rules for customer-facing bots, forcing companies that hide human-in-the-loop fallbacks to choose between transparency and the illusion their product depends on.
Third-order effects
- If the pattern holds, 'AI' services become a labor market in disguise — a reserve army of gig writers and call handlers paid to simulate machines — until regulation or buyer scrutiny forces the human share into the open.
- The essay also foreshadows the inverse trade now visible in companion apps like Replika, where humans increasingly relate to bots as if they were people, completing a loop in which each side performs the other.
The trend: Customer-facing AI is running on concealed human labor, and the gap between what bots do and what humans are paid to pretend they don't is becoming the industry's core accountability question.