In a lively Meta Discourse thread titled “Let’s see your best AI Image Caption!”, users enthusiastically test and showcase the new AI image captioning feature (available for TL2+). Saif kicks off the fun by inviting submissions of memes, pets, or anything else — with instructions on how to use the “Caption with AI” button. Participants like Lilly, Moin, and anon36555649 share diverse images — from vintage photos and miniature models to AI-generated art and humorous comics — often comparing multiple AI-generated captions for accuracy or creativity. Some users, like supermathie, propose recursive AI captioning experiments, while Firepup650 and JammyDodger explore layered screenshots and composer interface quirks. Falco and Arkshine contribute charming and surreal images, while Ed_S raises thoughtful questions about accessibility and community impact. Others, like Moin, even test multilingual captioning. The thread blends humor, creativity, and technical curiosity — with users celebrating the AI’s surprising accuracy on complex scenes, while also poking fun at its occasional missteps or overly literal interpretations.
Hey everyone
AI Image Captioning is now live in Meta [1] and we would love to see what everyone can come up with!
Feel free to submit your memes, funny pictures, cute pictures of your dog 
or whatever else you want 
I’ll start things off…
for TL2+ ↩︎
I like getting a cool pic and then generating a few AI captions to compare:
A group of happy teenagers are hopping over a white car in a dynamic and joyful moment against the backdrop of a mid-20th-century school building. (Captioned by AI)
A group of five young people appears to be joyfully jumping off the bumper of an old car, possibly from the 1950s or 1960s, in front of a school building. (Captioned by AI)
A group of young people in vintage clothing are playfully jumping off the bumper of a classic car, capturing a moment of youthful exuberance in a bygone era. (Captioned by AI)
A group of students are energetically leaping over a rope by a classic car with a school building in the background, creating a lively vintage scene. (Captioned by AI)
@Saif I want to see AI playing telephone with itself 
Human supplies a prompt
Prompt → AI Artist → AI Captioner → AI Artist → AI Captioner→ AI Artist → AI Captioner → AI Artist → AI Captioner
and see what we get.
(or, same but start and end with image)
The image shows a childlike drawing of a brown bird with a yellow beak spitting out a green object.
A childlike drawing depicts a brown bird with a yellow beak tossing a green object with its wing.
A hand-drawn picture depicts a brown bird with a yellow beak pointing at a green object, possibly a seed or an egg.
Well, the second is most accurate. It’s my daughter’s drawing of a kiwi tossing a kiwi.[1]
I’m using it on my site’s custom 50x page. She got the idea when we talked about how my business’ name would be pronounced in ancient Rome. ↩︎
The image displays a detailed miniature model of an urban scene with scale replicas of buildings, vehicles, and people, featuring the Colosseum in the background. (Captioned by AI)
This image showcases a detailed miniature model depicting a lively urban scene with buildings, vehicles, and figures that resemble a bustling city square. (Captioned by AI)
The image shows a detailed miniature model of an urban scene featuring a replica of the Colosseum and bustling streets with tiny figures of people and vehicles. (Captioned by AI)
I think it’s super cool AI was able to capture this sentiment from the picture, even took me a quick second 
It is very good at detecting that! I also tried some more pictures from the website of the Miniatur Wunderland.
Let’s see what happens if I give it a screenshot of Discourse:
Caption: The image displays an open web forum discussion with a user commenting on the effectiveness of an AI image captioning tool and mentioning their attempt to use it on other pictures from a website. (Captioned by AI)
What happens if we layer the screenshots a layer or two? 
The image shows a computer screen displaying a web forum where a user is engaged in a discussion about AI image captioning, with multiple browser tabs open and a chat sidebar visible. (Captioned by AI)
The image shows a computer screen with a browser open to a web forum where a user is discussing the use of an AI captioning tool and considering layering screenshots to test its effectiveness. (Captioned by AI)
The image depicts a computer screen with an open web browser showing a forum thread about AI-generated image captions, with a user joking about layering screenshots. (Captioned by AI)
Award winning caption!:
A pigeon sits next to anti-bird spikes on a stone ledge of a building, seemingly undeterred by the deterrent. (Captioned by AI)
A pigeon perches beside anti-bird spikes on a stone architectural element of a building. (Captioned by AI)
A pigeon perches beside anti-roosting spikes on a stone building ledge. (Captioned by AI)
Visitors in colorful attire observe a mighty waterfall from a viewing platform amidst lush green vegetation. (Captioned by AI)
The image shows a group of people clad in colorful raincoats standing on a pathway next to the powerful spray of a large waterfall surrounded by lush green vegetation. (Captioned by AI)
Visitors clad in colorful raincoats explore a walkway beside the powerful mist of a cascading waterfall surrounded by lush greenery. (Captioned by AI)
2 posts were split to a new topic: ‘Caption image’ displaying for those without permission to use it
The image is a satirical comic titled “CIRCLE OF AI LIFE,” depicting a cycle starting from humanity researching AI, perfecting it, then being enslaved by it, followed by the AI being disabled by a solar flare, and concluding with humanity worshiping the sun as a god.
The image humorously depicts a cycle where humanity researches and perfects artificial intelligence (AI), which then perfects itself and enslaves humanity, but is later disabled by a solar flare, leading to humanity worshiping the sun god.
(Captioned by AI)
Sitting in silence, the man and the robot share a moment of understanding and connection.
Discover the unexpected friendship between this small white dog and the chickens in their cozy coop
How did you create those captions? The ones generated here seem quite different.
A man in a leather jacket gestures towards a fallen humanoid robot on the ground in an urban setting, surrounded by curious onlookers. (Captioned by AI)
A man in a black leather jacket is confronting a humanoid robot that is lying on the ground amidst a crowd of onlookers. (Captioned by AI)
A rabbit stands upright inside a wooden and wire mesh enclosure beneath a sign that reads “BRIGHT EYED AND BUSHY TAILED.” (Captioned by AI)
A white rabbit stands on its hind legs inside a wooden and wire mesh enclosure with a sign reading “BRIGHT EYED AND BUSHY TAILED.” (Captioned by AI)
