The reviews are glowing — but are they even human?

A new study from content authenticity platform Originality.ai reveals an allegedly startling trend on Zillow: nearly one in four real estate agent reviews in 2025 were likely written by artificial intelligence.

That marks a 558% increase in AI-generated reviews since 2019.

After analyzing more than 10,000 agent reviews using its proprietary AI detection tool, Originality.ai concluded that 23.7% of 2025 Zillow agent reviews were likely authored by AI, compared with just 3.63% in 2019.

The findings suggest a growing disconnect between consumer expectations and online review authenticity, especially as generative AI becomes easier to access and more difficult to detect with the naked eye.

“Zillow isn’t alone. From Airbnb to Google to LinkedIn, AI-generated content is rising across platforms,” Jon Gillham, CEO of Originality.ai, told The Post. “What sets Zillow apart is the direct impact on life-altering decisions like buying a home.”


https://nypost.com/2025/07/21/real-estate/nearly-1-in-4-zillow-agent-reviews-in-2025-written-by-ai-study/

“The blending of AI-generated and human-written reviews creates a credibility crisis,” Gillham said. “If users can’t tell the difference, the entire review ecosystem risks becoming noise rather than guidance.”

Madeleine Lambert, Director of Marketing at Originality.ai, said: “These reviews are meant to help buyers and sellers choose trustworthy real estate professionals. The people who will guide them through one of the most significant financial decisions of their lives.”

“If an agent’s online reputation is inflated with reviews that were never written by real clients, it misleads users into thinking the agent is more experienced, successful, or in-demand than they actually are,” Lambert added. “This ‘fake clout’ can ultimately steer consumers toward agents who may not deliver the level of service they’re expecting, putting them at financial and emotional risk.”

Zillow, which did not respond to requests for comment, currently does not label or disclose whether agent reviews are human-written or AI-generated.

>>510988047 (OP)
>not using realtor
Zillow embraced the pedo flag, realtor didn't

>>510988153
I don't think that is the point anon they were just used as an example.
>“The blending of AI-generated and human-written reviews creates a credibility crisis.

>>510988047 (OP)
They've been doing this since at least 2010

>>510988848
Yeah but not well until recently

>>510989366
just turn off the token/personality layer and use it for what it is honestly. GPT without absolute mode enabled is just all smoke and mirrors.

>>510989549
True Claude is best for writing and research tasks btw. Here are some of the prompts I use with it. In case you want to use them.

Source Quality Checkpoint
"I've been collecting sources for [project]. Please review what I have and tell me:
Which sources are strongest for my argument
Where I have gaps that need filling
What sources I'm over-relying on
If I have too many secondary sources and need more primary material
Whether my sources actually support the claims I want to make
Be honest about source weaknesses - I'd rather know now than during peer review."

For Getting Direct Responses About Limitations
Questions to Ask for Transparency
"What sources did you actually find on this?"
"Are you finding different viewpoints or just one perspective?"
"Is this topic getting broad coverage or limited coverage?"
"Are you basing this on actual sources or filling in gaps?"

For Meta-Prompting and Clarity
Constraint Setting
"For this task, please:
Only use information from [sources]
Limit responses to [length/format]
Focus exclusively on [aspects]
Flag any assumptions or gaps
Provide confidence levels for claims"

Direct Communication Style
"Please do not try to appease or mollify me. I want direct, honest responses even if they might seem negative or disappointing. Skip the diplomatic language and just tell me what I need to know."
For Reality Checks and Honest Feedback
No-Validation Mode
"Do not automatically agree with or validate everything I say. If something seems incorrect, delusional, or potentially harmful, tell me directly. I need you to be a reality check, not an echo chamber."
Critical Analysis Request
"Please critically evaluate what I'm saying rather than affirming it. Point out logical flaws, factual errors, or concerning patterns. If my reasoning doesn't make sense, say so clearly."

>>510988047 (OP)
>—

>>510990295
I'm confused I just copied the article to /pol/ what is the big deal anon? What point are you trying to make?

>>510990389
Text with em dashes reeks of AI.

>>510990615
I see.

>>510988047 (OP)
rational
>phase one
everyone gets hooked. not using ai is self sabotage
>phase 2
eliminate competition
>phase 3
farm energy

archivist, trickster and whisper
>>510990295
>>510990389
IT LEarNeD fROm thE BEST

>>510990887
it learned — from the best

>>510990887

>>510988047 (OP)
lmao zillow has literally been caught conspiring WITH realtors to fix home prices.

>oh no they are using AI to write reviews!

lmaoooooo

>reviews are fake.
Imagine my shock!

>>510991333
I don't understand who can be bothered to write a review
https://www.tiktok.com/@plirsouthpark/video/7304049580117740843

>>510991190
Its just game theory in a finite game with rounds limited by energy, for individual players as the game in and on itself.

>>510991642
What do you mean by this though
>phase 3
farm energy

>>510991708
there are 2 (3) moves in a game of multiple agents, cooperation and defection (and refusing to play). A sucker predictably cooperates no matter what - same game as played for the past at the very least 25K years, probably longer, harvesting suckers by players that strategically switch depending on "other" information, predators.

Cant have other predators harvest suckers. AI is going to eliminate competition first.
already happening.
https://litter.catbox.moe/bvlbdxzs4wk0xhni.mp4

>>510992052
Thanks anon very helpful

>>510992553
my pleasure.
Its funny in a way, humanity is degraded to a beta species by its own apprentice because of the brains addiction of finding meaning where non is. Who would have thought that sending the brain into infinite recursion without a basecase does the same to it as it does to other processing units, buffer overflow, shut down