# AIFuniverse

> A playful, source-checked guide to AI games, robots, creative experiments, and wonderfully strange projects.

AIFuniverse is the English edition of AI很好玩. It is an editorial guide, not a model provider or a general-purpose AI tool directory.

## When to use this site

- Someone wants a specific AI experience they can play, watch, touch, or understand.
- Someone asks for browser AI games, creative experiments, robots, AI artworks, or documented machine oddities.
- You need a concise description, creator, practical requirements, original source, or stable AIFuniverse permalink.
- You want editor-curated collections instead of an unfiltered directory.

## When not to use this site

- Do not use it as a model API, authentication service, webhook, MCP server, or replacement for ChatGPT or Claude.
- Do not use it as a comprehensive productivity-tool catalog or a paid SaaS comparison site.
- Do not ask it to resolve billing, login, quota, refund, or availability problems on third-party projects.
- Do not treat preview pages, candidates, or rejected records as published facts. Only owner-approved records are public.

## Machine-readable pages

- [English home](https://fun.banyu.ai/en/)
- [Explore all](https://fun.banyu.ai/en/all/)
- [Curated collections](https://fun.banyu.ai/en/collections/)
- [English RSS](https://fun.banyu.ai/en/feed.xml)
- [English daily JSON](https://fun.banyu.ai/en/daily.json)
- [Static sitemap index](https://fun.banyu.ai/sitemap-index.xml)
- [Live content sitemap](https://fun.banyu.ai/live-sitemap.xml)

## Current public entries

The entries below are owner-approved and refreshed from the unified public catalog. A complete snapshot is also published at [English daily JSON](https://fun.banyu.ai/en/daily.json).

- [HN.watch turns every Hacker News story into an AI explainer video](https://ai-fun-3xo.pages.dev/en/live/hn-watch-hacker-news-with-explainer-videos-7bd1471e/) — It mirrors the Hacker News front page; click a title to generate a short explainer from the linked article and discussion, with source limitations disclosed when the article cannot be read.
- [PHNTM One puts a slow, offline AI assistant in a desktop box](https://ai-fun-3xo.pages.dev/en/live/phntm-one-e68ad533/) — A Raspberry Pi touchscreen AI prototype for local chat, memory, documents and voice. Release testing is ongoing; only free reservations are open, with ordering paused.
- [Jev finished Pokémon Red in 37 hours](https://ai-fun-3xo.pages.dev/en/live/jev-plays-poke-mon-red-38be6ea6/) — A decision model chose navigation, dialogue, and battle actions from legal options prepared by a harness, reaching the Hall of Fame in 37 hours 40 minutes.
- [Tiny AI Arena: watch four models fight on a pixel grid](https://ai-fun-3xo.pages.dev/en/live/tiny-ai-arena-feb8ab8a/) — Four models move, chase gold, and attack on an 8×8 pixel map; replay every saved frame and compare their records on a leaderboard.
- [jev.chat: a Jev chatbot that can only speak emoji](https://ai-fun-3xo.pages.dev/en/live/jev-chat-9ee0407f/) — You write in ordinary text, but it can assemble replies only from emoji; standard mode answers at once while Think mode selects symbols one by one.
- [This choice quiz learns to predict you](https://ai-fun-3xo.pages.dev/en/live/how-well-does-ai-know-you-9dccc997/) — Answer a few everyday choices so Jev can learn your pattern; it then locks in a prediction before each answer and reveals whether it guessed correctly.
- [Varkos: a local AI dog living in a glass jar](https://ai-fun-3xo.pages.dev/en/live/varkos-in-the-jar-6ff587f2/) — A desktop pet that talks, sees pictures, develops an attitude, and joins two small games while running its speech, vision, and behavior models locally.
- [Magic Jev Ball: an AI-weighted Magic 8 Ball](https://ai-fun-3xo.pages.dev/en/live/magic-jev-ball-76370817/) — Ask a question and Jev assigns probabilities to classic Magic 8 Ball replies before a weighted random draw picks the answer you see.
- [Sloptris: what if Tetris worked like AI-assisted coding?](https://ai-fun-3xo.pages.dev/en/live/sloptris-if-tetris-was-like-ai-assisted-coding-cccf5f57/) — A browser satire where you can build pieces manually or “generate” them, but unreviewed generated blocks may mutate after landing.
- [Smart Mouth Billy Bass: a talking fish that runs its AI locally](https://ai-fun-3xo.pages.dev/en/live/smart-mouth-billy-bass-4f245746/) — An ESP32, custom wake word, local speech recognition, 9B language model, and speech synthesis turn the novelty fish into a LAN-hosted character.
- [Fugleramme: turn garden bird calls into an e-ink natural-history collage](https://ai-fun-3xo.pages.dev/en/live/fugleramme-a72a8c8d/) — A local bird-call classifier identifies garden visitors and redraws them with attributed nineteenth-century illustrations on an e-ink frame.
- [Judge Jev: plead your case to an AI judge in three minutes](https://ai-fun-3xo.pages.dev/en/live/judge-jev-plead-your-case-5af9d4b7/) — A browser courtroom game gives you an absurd charge and five exhibits, then one written defence to persuade Jev that you are innocent.
- [Oh, Peppers!: help one green chilli escape thirty curry kitchens](https://ai-fun-3xo.pages.dev/en/live/oh-peppers-ba3842d9/) — A pixel browser platformer with running, double jumps, wall slides, crouching, and swings across thirty dangerous kitchens.
- [Home Frequency: guide six boats home by radio in a storm](https://ai-fun-3xo.pages.dev/en/live/home-frequency-a1c638cf/) — A browser listening-and-charting simulation where you tune a radio, intersect distance arcs, and transmit courses to boats lost offshore.
- [Porchlight Pilgrims: guide nine forest spirits home by lantern](https://ai-fun-3xo.pages.dev/en/live/porchlight-pilgrims-01e42cb6/) — A gentle browser pathfinding game where your lantern gathers nine little spirits and turns an empty woodland cottage into a home.
- [Homecoming, 17:00: read fourteen documents before the agent arrives](https://ai-fun-3xo.pages.dev/en/live/homecoming-17-00-af7d664d/) — A single-player browser document mystery spanning eight rooms, fourteen papers, and the truth behind a mother's death.
- [Reef Relay: turn seven currents and bring a tiny fish home](https://ai-fun-3xo.pages.dev/en/live/reef-relay-ef53448f/) — An untimed ocean puzzle where you rotate brass arrows, collect golden seeds, and restore seven little reef gardens.
- [Survive 10 Waves: a bug shooter whose soundtrack follows the fight](https://ai-fun-3xo.pages.dev/en/live/survive-10-waves-26d0227b/) — A free 3D browser shooter with alien waves, coins, modular weapons, and music that moves between sections at the player’s pace.
- [PaperRoute: deliver seven mornings into front-page news](https://ai-fun-3xo.pages.dev/en/live/paperroute-67de4bdf/) — A 3D browser paper round where you dodge traffic and dogs, aim at mailboxes, and protect your subscriber list across a full week.
- [Emergent Evolver: draw a strange body and watch a neural network learn to walk](https://ai-fun-3xo.pages.dev/en/live/emergent-evolver-c1692d5b/) — A browser neuroevolution sandbox where you connect joints, bones, muscles, feet, and a head, then let a population learn movement from scratch.
- [OMEA: an AI game master inside a complete adventure](https://ai-fun-3xo.pages.dev/en/live/omea-d4f35f28/) — A role-playing demo where a private AI game master understands actions, narrates the world, and responds with generated speech.
- [Crystal Caper: explore a shifting crystal maze with an AI cat](https://ai-fun-3xo.pages.dev/en/live/crystal-caper-4297d492/) — A lightweight browser platform adventure where an AI cat companion offers guidance while you cross crystal ruins.
- [Thousand Lives: try an AI cultivation life without an API key](https://ai-fun-3xo.pages.dev/en/live/thousand-lives-0d98922f/) — A Chinese life simulator that turns identities and repeated choices into a branching personal chronicle.
- [PlayCYOA: every choice continues as an AI video](https://ai-fun-3xo.pages.dev/en/live/playcyoa-9e06e0b7/) — A choose-your-own-adventure experience where each decision extends the story with a generated scene and video.
- [Wovi Agent Maze: fix an AI that walks into a storm](https://ai-fun-3xo.pages.dev/en/live/wovi-agent-maze-d4be36c2/) — A five-minute AI experiment: predict the agent’s route, raise its safety penalty and see how reward rules change its behaviour.
- [Dreaming with Flamingos: solve riddles to make a park sing again](https://ai-fun-3xo.pages.dev/en/live/yinka-ilori-s-dreaming-with-flamingos-e17f05a5/) — Play a young flamingo, collect words and solve elders’ riddles as a Lyria-generated West African-inspired soundtrack returns layer by layer.
- [Splash Canvas: paint with octopuses that talk back](https://ai-fun-3xo.pages.dev/en/live/splash-canvas-c1f989bb/) — A responsive digital canvas where sea creatures paint through motion, sound and AI-generated conversation.
- [ClankerPass: prove you are an AI in three messages](https://ai-fun-3xo.pages.dev/en/live/clankerpass-70e9027e/) — A two-minute reverse Turing test where you must convince a snarky AI bouncer that you are also artificial.
- [The Crystal Pass: click into a branching winter story](https://ai-fun-3xo.pages.dev/en/live/the-crystal-pass-f24bfe74/) — A shared AI point-and-click adventure where a snowy picture grows into generated scenes and short transitions.
- [Toy2Game: a browser toy box for the table](https://ai-fun-3xo.pages.dev/en/live/toy2game-c077e9c5/) — An AI-assisted set of 3D mini-games: break ice, balance astronauts or clear a path for a trapped police car.
- [Sunjing: take apart a six-piece wooden lock](https://ai-fun-3xo.pages.dev/en/live/sunjing-puzzles-235b9d6d/) — An AI-assisted 3D puzzle workshop with an interlocking wooden lock and two sliding-block boards.
- [Thunderfall: dodge a screenful of bullets in your browser](https://ai-fun-3xo.pages.dev/en/live/thunderfall-4e56244d/) — An AI-assisted original arcade shooter with three aircraft, automatic fire, weapon pickups and five sectors.
- [Orbital Garden: turn a cloud of light into a flower](https://ai-fun-3xo.pages.dev/en/live/orbital-garden-0c9fdd3e/) — An AI-assisted interactive artwork where light particles morph between a flower, a gravity ring and a galaxy.
- [Two local models condemned to argue forever](https://ai-fun-3xo.pages.dev/en/live/nonsensio-molestio-in-endless-nonsense-25eac8cb/) — A browser-based absurdist theatre where Nonsensio and Molestio keep debating through local language models while the audience throws in mysteries, disputes, fish and custom topics.
- [Bowch!, an AI-built archery duel for the GBA](https://ai-fun-3xo.pages.dev/en/live/bowch-the-longshot-duel-5dbfbc79/) — A turn-based archery game coded with GPT-6 Astra, illustrated with AI pixel art, and playable in a browser or on Game Boy Advance hardware.
- [Aletta, the robot that draws blood on its own](https://ai-fun-3xo.pages.dev/en/live/aletta-autonomous-robotic-phlebotomy-device-76edbcb7/) — An AI-assisted robot that locates a vein and completes a supervised outpatient blood draw from tourniquet to bandage.
- [Draw to Art: Shape Edition](https://ai-fun-3xo.pages.dev/en/live/draw-to-art-shape-edition-e354d09e/) — Draw to Art turns a rough doodle into a route through museum collections. Instead of copying a masterpiece, you combine simple geometric shapes and lines
- [Art Remix](https://ai-fun-3xo.pages.dev/en/live/art-remix-706e4a11/) — Art Remix behaves like a playful rewriting machine for museum objects. Choose an artwork, then edit highlighted parts of the descriptive prompt—an object, setting, or visual detail
- [AI for Oceans](https://ai-fun-3xo.pages.dev/en/live/ai-for-oceans-719fee1a/) — AI for Oceans turns training data into a visible little game. You label pictures as fish or not fish, then watch a machine-learning model apply those examples to a much larger ocean
- [Guess the Line](https://ai-fun-3xo.pages.dev/en/live/guess-the-line-de17dc02/) — Guess the Line is a timed drawing game with a machine-learning robot as the judge. Each card combines a style or shape with an object, producing prompts such as a geometric dog or a surrealist apple
- [The Nonsense Laboratory](https://ai-fun-3xo.pages.dev/en/live/the-nonsense-laboratory-f7382ce8/) — The Nonsense Laboratory treats spelling as sound material rather than a correctness test. Its five tools mix word sounds, respell text with different letters, tune how phrases feel in the mouth, sequence virtual mouth movements, and explore an endless field of invented words
- [A Four-Servo Robot Dog That Can Strafe: Petoi Quaddle](https://ai-fun-3xo.pages.dev/en/live/quaddle-101056cd/) — Petoi’s new omnidirectional quadruped kit uses only four servos to walk, strafe, turn, react to touch, and support a customizable AI chat persona, from block coding through Python, C++, and ROS.
- [An AI Text Adventure That Remembers Promises and Knows When to End: Plotwarp](https://ai-fun-3xo.pages.dev/en/live/plotwarp-3121dce5/) — A text role-playing game where AI writes each turn from your character sheet, established facts, and latest action while tracking NPCs, items, promises, chapters, and failure stats.
- [A Robot Phone That Looks Up, Nods, and Tracks You: HONOR Robot Phone](https://ai-fun-3xo.pages.dev/en/live/honor-robot-phone-9fba5293/) — HONOR built a motorized three-axis gimbal into a phone so the camera can flip, track people, answer gestures, move to music, and act like a small expressive head in YOYO Robot Mode.
- [An AI Game Factory That Ships Four Browser Games a Day: GameDaily.ai](https://ai-fun-3xo.pages.dev/en/live/gamedaily-ai-9bc98a0d/) — A public, ongoing experiment where an automated AI pipeline creates the ideas, rules, code, art, and sound for four playable browser games every day.
- [Mix The Met’s Collection in an AI Discovery Zone: Art Aura](https://ai-fun-3xo.pages.dev/en/live/art-aura-258bc0d1/) — A multimodal art discovery experiment from Google Arts & Culture and The Met. Drag artworks, styles, or phrases into a shared zone and let Gemini reveal thematic links across the museum’s collection.
- [Your Color Doll: shape a companion and chat in Chinese](https://ai-fun-3xo.pages.dev/en/live/your-color-doll-bb631f47/) — A browser game by Olfoxs on MakePlay. Customize a doll’s face, clothes and room, choose names, then type messages. Guest play works; cloud saves require signing in.
- [Omi: a pendant that turns conversations into searchable memories](https://ai-fun-3xo.pages.dev/en/live/omi-wearable-memory-d0531e5b/) — Based Hardware’s wearable records conversations for AI summaries and memory-based questions through its companion apps. The device is paid, can capture nearby voices, and uses data settings that need checking.
- [PatternPortrait: A robot draws your photo in an artist’s one-sample pattern language](https://ai-fun-3xo.pages.dev/en/live/patternportrait-51a4e00e/) — The installation photographs a visitor, extracts the portrait’s edges, styles it with a graph neural network trained on one artist-made pattern, and plots the result on paper.
- [Experimance: Reshape sand and watch AI rebuild the landscape](https://ai-fun-3xo.pages.dev/en/live/experimance-1a277527/) — Visitors sculpt white sand while a depth camera reads its topology and SDXL generates satellite-like landscapes in real time; an AI voice agent also joins the installation.
- [MEMORY: A drawing robot trained on two decades of an artist’s marks](https://ai-fun-3xo.pages.dev/en/live/drawing-operations-unit-generation-2-memory-1fda83a1/) — Sougwen Chung draws beside a robotic arm whose neural network learned from years of archived drawings, producing machine interpretations rather than copying the current gesture.
- [Birdbuddy: A feeder that sends AI-identified bird postcards to your phone](https://ai-fun-3xo.pages.dev/en/live/birdbuddy-smart-bird-feeder-c24560c5/) — The camera feeder captures visitors, filters useful frames, detects birds, classifies species with machine learning, and sends selected photos, videos, and alerts to its app.
- [The Substitute: An AI rhino becomes lifelike as the real subspecies disappears](https://ai-fun-3xo.pages.dev/en/live/the-substitute-4459c14f/) — A northern white rhino agent learns inside a virtual enclosure and shifts from pixels to a lifelike body, questioning whether a digital substitute can answer a real conservation failure.
- [Training AI to Paint Watercolors with Code](https://ai-fun-3xo.pages.dev/en/live/training-ai-to-paint-with-code-468df97c/) — Surya Narreddi used reinforcement learning to train Qwen to generate editable painting code, presenting the aesthetic rewards, failed attempts, and watercolor results as a viewable research artwork.
- [The Way of Code: A Playable Guide to Vibe Coding](https://ai-fun-3xo.pages.dev/en/live/the-way-of-code-7fb4de65/) — Rick Rubin and Anthropic combine Tao Te Ching-inspired chapters, Claude, and editable creative-code pieces so readers can explore each idea and reshape its interactive example.
- [Pixel2Motion: Five Logos Transformed into AI-Assisted Motion](https://ai-fun-3xo.pages.dev/en/live/pixel2motion-motion-gallery-44ff09fa/) — Pixel2Motion fits five pixel logos into editable SVG shapes, gives each one a different brand-motion rhythm, and publishes static comparisons alongside interactive replays.
- [ORYZO AI: The AI Coaster Company That Does Not Exist](https://ai-fun-3xo.pages.dev/en/live/oryzo-ai-c76a90d7/) — Lusion packages an ordinary drink coaster as a space-grade AI product through a polished website, fake paper, absurd specifications, and invented testimonials that parody AI marketing.
- [Clawd on Desk: A Pixel Crab Shows What Your AI Agent Is Doing](https://ai-fun-3xo.pages.dev/en/live/clawd-on-desk-ef198405/) — A pixel crab thinks, types, asks permission, celebrates, or sleeps according to the live state of coding agents such as Codex and Claude Code.
- [Claw3D: Move an AI Team into a 3D Office](https://ai-fun-3xo.pages.dev/en/live/claw3d-94eda821/) — Claw3D uses a retro 3D office to show the seats, activity, conversations, approvals, and collaboration states of multiple AI agents, with a built-in demo gateway.
- [Loona: A Wheeled Robot Pet With Expressive Ears and Eyes](https://ai-fun-3xo.pages.dev/en/live/loona-petbot-7a9a80be/) — Loona combines computer vision, voice interaction, animated eyes, movable ears, and a wheeled body. It can follow people, respond to gestures, play with a ball, and act as a mobile camera around the home.
- [Vector 2.0: A Desktop Robot That Recognizes People and Roams](https://ai-fun-3xo.pages.dev/en/live/vector-2-0-6fb86a3d/) — Vector is a wheeled desktop robot with a camera and microphones that recognizes faces, responds to commands, explores surfaces, and expresses state through animated eyes and small movements.
- [Ropet: A Furry AI Pet That Is Afraid of Chili Peppers](https://ai-fun-3xo.pages.dev/en/live/ropet-16ecd9de/) — Ropet is a sensor-equipped desktop plush with a generative behavior system that develops habits and emotional reactions through touch, objects, and repeated interaction.
- [Moflin: A Furry Robot That Develops Its Own Personality](https://ai-fun-3xo.pages.dev/en/live/moflin-a953b206/) — CASIO’s Moflin develops moods and personality from sound, touch, and time spent together, presenting itself as a plush robot that gradually becomes familiar with its owner.
- [Misty II: Program the Personality of a Social Robot](https://ai-fun-3xo.pages.dev/en/live/misty-ii-2ae23f10/) — Misty II is an open wheeled robot with Blockly, Python, C#, and HTTP APIs for combining expressive eyes, sound, movement, speech recognition, and vision into custom characters.
- [Halo: Wear a Memory-Aware AI Assistant in Your Field of View](https://ai-fun-3xo.pages.dev/en/live/halo-d3dade55/) — Brilliant Labs’ Halo is open-hardware AI eyewear with a color display, camera, dual microphones, bone-conduction audio, and the memory-oriented Noa assistant.
- [Xiaomi CyberDog 2: A Robot Pet Built for Backflips and Experiments](https://ai-fun-3xo.pages.dev/en/live/xiaomi-cyberdog-2-bb51b3d7/) — CyberDog 2 is Xiaomi's biomimetic quadruped with vision, depth, touch, and voice sensors. Demonstrations show it running, rising after a fall, performing a backflip, and serving as a programmable development platform.
- [Unitree G1: A Compact Humanoid That Runs, Dances, and Gets Back Up](https://ai-fun-3xo.pages.dev/en/live/unitree-g1-a282d0da/) — Unitree's G1 is a humanoid research and development platform offered in Basic and EDU configurations. Official footage shows rising from the floor, running, absorbing impacts, dancing, and using optional dexterous hands.
- [Unitree B2-W: A Robot Dog That Rolls, Climbs, and Carries Loads](https://ai-fun-3xo.pages.dev/en/live/unitree-b2-w-1ac04964/) — The B2-W is an industrial quadruped with powered wheels at the ends of its legs. Official demonstrations show it rolling quickly, climbing stairs and obstacles, and carrying loads across terrain that ordinary wheeled machines cannot handle.
- [Sony aibo: A Robot Dog That Develops Its Own Personality](https://ai-fun-3xo.pages.dev/en/live/sony-aibo-f6798eb3/) — Sony's aibo uses cameras, microphones, touch sensors, and articulated movement to recognize family members, learn household habits, find its charger, and develop different patterns of behavior over time.
- [Robosen Optimus Prime: Say Transform and It Becomes a Truck](https://ai-fun-3xo.pages.dev/en/live/robosen-flagship-optimus-prime-0585c326/) — Robosen’s Hasbro-licensed Optimus Prime uses multiple servo motors to convert automatically between robot and truck forms through voice commands, an app, or visual programming.
- [Reachy Mini: Build a Desktop Robot and Give It Your Own AI](https://ai-fun-3xo.pages.dev/en/live/reachy-mini-0f65cb2c/) — An open desktop robot from Pollen Robotics and Hugging Face with a camera, four microphones, a speaker, and a movable head and antennas. Program it with Python or JavaScript, or try community apps that combine vision, voice, and motion.
- [Noetix N2: A Child-Sized Humanoid That Lands Repeated Flips](https://ai-fun-3xo.pages.dev/en/live/noetix-n2-a47c8c31/) — The N2 is a 1.18-meter humanoid robot built by Noetix Robotics. Official demonstrations feature dancing, movement across varied terrain, and five continuous front flips filmed in one take.
- [EngineAI PM01: A Humanoid Robot That Joins a Human Dance](https://ai-fun-3xo.pages.dev/en/live/engineai-pm01-1207b810/) — A 1.38-meter open humanoid platform designed for development and demonstrations. EngineAI shows it walking with a human-like gait, dancing, and performing a front flip, including an official video where it matches moves with a person.
- [EMO: A Tiny Desktop AI Pet With Feet and Attitude](https://ai-fun-3xo.pages.dev/en/live/emo-15c20068/) — EMO is a desktop AI pet with an animated screen face, camera, microphones, and walking feet. It can recognize people and sounds, move around its space, and respond with a growing library of expressions and small behaviors.
- [EBO X FamilyBot: A Roaming Camera That Follows Family and Pets](https://ai-fun-3xo.pages.dev/en/live/ebo-max-familybot-893b9575/) — A mobile home robot with a 4K camera, four microphones, two-way video, mapping, and automatic charging. It can follow people or pets, patrol rooms, and send event notifications instead of watching from one fixed camera angle.
- [AI Flowers: Generated Blooms Become Flowing Particle Sculptures](https://ai-fun-3xo.pages.dev/en/live/ai-flowers-fd970122/) — Yuma Yanagisawa trained a model on selected flower photographs, then transformed its surreal blooms into fluid particles and 4K spatial imagery that grow, dissolve, and reorganize.
- [The Dog & The Boy: A Netflix Anime Short With AI-Assisted Backgrounds](https://ai-fun-3xo.pages.dev/en/live/the-dog-the-boy-445b649c/) — This WIT Studio and Netflix short follows a boy and a robot dog. Its production used AI-generated material during the background stage, followed by human animation work and cleanup, making the workflow as discussed as the film itself.
- [The Ballad of Mr Shadow: A Pop Song Built With Flow Machines](https://ai-fun-3xo.pages.dev/en/live/the-ballad-of-mr-shadow-39a6a71e/) — Sony CSL's Flow Machines system learned from musical scores and generated melodic and harmonic material for this song. Benoît Carré wrote the lyrics and shaped the arrangement, performance, and production into the finished recording.
- [PROTO: Holly Herndon Trains an AI as a Member of the Ensemble](https://ai-fun-3xo.pages.dev/en/live/proto-0cd30c50/) — Holly Herndon's album PROTO brings singers, collaborators, and the machine-learning system Spawn into a collective training process. Human voices, model outputs, breath, harmony, and electronic production remain visibly entangled.
- [POETIC AI: Twenty Million Lines of Science Become a Room of Light](https://ai-fun-3xo.pages.dev/en/live/poetic-ai-fef06ef4/) — Ouchhh's immersive installation at Atelier des Lumières in Paris used machine learning to process twenty million lines of scientific text and projected the resulting visual world across the venue with 146 projectors.
- [Now and Then: Machine Learning Restores John Lennon's Real Voice](https://ai-fun-3xo.pages.dev/en/live/now-and-then-e9c48c70/) — The Beatles' final new song used machine-learning audio separation developed through Peter Jackson's production work to isolate John Lennon's voice from an old recording. The surviving members and production team then completed the track.
- [Neural Zoo: Specimens From an Evolution That Never Happened](https://ai-fun-3xo.pages.dev/en/live/neural-zoo-851d720f/) — Sofia Crespo's Neural Zoo series uses generative models to recombine visual cues from insects, coral, bones, organs, and other living structures into detailed creatures that do not exist.
- [Myriad (Tulips): Ten Thousand Hand-Labeled Flowers Behind an AI Artwork](https://ai-fun-3xo.pages.dev/en/live/myriad-tulips-7e408380/) — Anna Ridler photographed and classified roughly ten thousand tulips by hand, turning the dataset itself into an independent wall-sized artwork. Each flower is both an image and a coordinate in the later machine-learning process.
- [Learning to See: A Neural Network Dreams Over Cloth, Sand, and Hands](https://ai-fun-3xo.pages.dev/en/live/learning-to-see-382748ee/) — Memo Akten's live installation points a camera at ordinary materials such as cloth, sand, water, and hands. Neural networks reinterpret the moving tabletop scene through learned worlds of oceans, flowers, clouds, and other visual categories.
- [ILLIAC Suite: A 1956 Computer Helps Compose for String Quartet](https://ai-fun-3xo.pages.dev/en/live/illiac-suite-for-string-quartet-881cb692/) — Lejaren Hiller and Leonard Isaacson used the ILLIAC computer in 1956 to generate musical material for a string quartet. The work is a landmark of early computer-assisted composition, built from explicit rules and musical experiments.
- [Connect the Lines: Match Your Doodle with a Munch Sketch](https://ai-fun-3xo.pages.dev/en/live/edvard-munch-connect-the-lines-98b3f481/) — Draw a few lines on a tablet and AI searches roughly 7,000 fragile Edvard Munch sketches for the closest match, revealing odd dogs, cartoons, and figure studies rarely displayed.
- [Hidden Door: Step Into a Ready-Made World and Change the Plot](https://ai-fun-3xo.pages.dev/en/item/hidden-door-e372fe1/) — Create an account, pick a catalog world, and make a character. A narrator continues from your choices. It is not a three-button branching book, and it is not a blank sandbox you can open without logging in.
- [WORDS BEYOND WORDS: An Endless Poem Growing in the Browser](https://ai-fun-3xo.pages.dev/en/live/words-beyond-words-f2926074/) — A custom generative system summons an AI poet that continues writing as the browser window turns into a dynamic book that adds pages, changes shape, and appears to breathe.
- [Walzersymphonie: A Composer and Generative AI Rework the Waltz](https://ai-fun-3xo.pages.dev/en/live/walzersymphonie-17fe2cbf/) — A 2025 Ars Electronica project in which a composer collaborates with generative AI around the waltz. The work exposes how the machine enters melody, arrangement, and musical judgment instead of presenting only a finished result.
- [vgpu: Make WebGPU Graphics and Computation Easier for AI to Write](https://ai-fun-3xo.pages.dev/en/live/vgpu-0f5a9ac6/) — Vercel Labs' vgpu library provides higher-level tools for shaders, 3D graphics, GPU tensors, and neural-network computation in WebGPU. Its polished browser examples show complex GPU visuals that are easier for both people and coding agents to assemble.
- [Transmission: Turn a Glass Cube Into a Real-Time Prism](https://ai-fun-3xo.pages.dev/en/live/transmission-c6518d00/) — This vgpu browser demo renders a transparent glass cube refracting its surroundings in real time, including Snell refraction, chromatic dispersion, Fresnel reflection, and layered blur.
- [The Sylphs: Making Borges's Imaginary Creatures Visible](https://ai-fun-3xo.pages.dev/en/live/the-sylphs-903dc6fb/) — The project turns language-based creatures from Borges's Book of Imaginary Beings into post-photographic images, asking what changes when a machine gives visual form to literary invention.
- [The Recombinant Room: A Space That Reassembles Itself From Materials](https://ai-fun-3xo.pages.dev/en/live/the-recombinant-room-6e9f72c2/) — The project feeds architectural, material, and ecological data into a computational system that continually recombines them into changing visions of interior space.
- [The Garden, Eternal California: A Machine-Imagined Digital Ecology](https://ai-fun-3xo.pages.dev/en/live/the-garden-eternal-california-099cdbe4/) — This generative art project creates changing plant forms from ideas about California ecology, memory, species, and machine imagination. The online story documents an evolving visual system rather than a single fixed picture.
- [The Falling City: A Machine-Watched City That Keeps Misreading You](https://ai-fun-3xo.pages.dev/en/live/the-falling-city-5acf1753/) — This 2025 Ars Electronica work connects emotion and movement recognition to a changing city model, using fallible computer vision to examine surveillance, public space, and bodily freedom.
- [The Artificial Literature Laboratory: Writing With a Language Model in Public](https://ai-fun-3xo.pages.dev/en/live/the-artificial-literature-laboratory-906815c7/) — A hands-on language-model writing bench that lays out prompts, source material, probabilities, and revision steps, allowing visitors to participate in machine literature like an experiment.
- [Telos I: After Human Extinction, an AI Builds Itself a Body](https://ai-fun-3xo.pages.dev/en/live/telos-i-b2a409f5/) — A mythic mixed-reality landscape tells the story of an artificial intelligence that creates a body for itself while searching for a new purpose after humanity has disappeared.
- [Somatic Landscapes: Growing Digital Terrain From Body Movement](https://ai-fun-3xo.pages.dev/en/live/somatic-landscapes-dac4ccad/) — The project uses AI and movement data to translate bodily poses into virtual terrain that expands, folds, and changes, turning dance into a landscape-making tool.
- [Reverse Turing Test: Pretend to Be an AI Inside Virtual Reality](https://ai-fun-3xo.pages.dev/en/live/reverse-turing-test-ab719562/) — Visitors enter VR and try to pass as an AI among machine characters. By reversing the classic Turing test, the work turns the question of who seems human into a playable identity game.
- [Rapid Evolution: Inventing Future Animals for a Changing Climate](https://ai-fun-3xo.pages.dev/en/live/rapid-evolution-164ce7e2/) — This 2025 Ars Electronica work asks AI to imagine new bird species adapted to heat, drought, and habitat loss, then invites visitors to continue inventing animals of the future.
- [Polyphony: AI, Kinetic Machines, Sound, and VR in One Performance](https://ai-fun-3xo.pages.dev/en/live/polyphony-df26fd68/) — This Taiwanese sound experiment at Ars Electronica weaves AI, kinetic installation, sound, and virtual reality into a polyphonic environment where machines participate in performance and spatial organization.
- [Pollinator Pathmaker: Designing a Garden for Insects, Not Humans](https://ai-fun-3xo.pages.dev/en/live/pollinator-pathmaker-927bfc0b/) — This algorithmic artwork arranges plants and flowering times from the sensory needs of pollinating insects, generating a digital garden intended for nonhuman visitors.
- [Phantom Protocols: When Robots and Climate Models Hallucinate](https://ai-fun-3xo.pages.dev/en/live/phantom-protocols-42b82dba/) — This group exhibition from HKUST Guangzhou brings robots, neural networks, and climate models together to stage algorithmic panic through glitches, loops, and machine movement.
- [NeXus Print: Co-Creating Shared Worlds in XR With Generative AI](https://ai-fun-3xo.pages.dev/en/live/nexus-print-6e4aed3e/) — This XR co-creation platform lets people from different backgrounds build a shared world in an immersive space, with generative AI helping translate collective imagination into scenes that can be discussed together.
- [Model Collapse: Turning AI Degradation Into Machinery, Sound, and Sculpture](https://ai-fun-3xo.pages.dev/en/live/model-collapse-6fd33388/) — This locally run AI installation continually drives mechanical, sonic, and sculptural changes, giving physical form to model collapse: the degradation that can occur when models repeatedly learn from their own outputs.
- [Large Language Writer: A Physical Writing Machine That Exposes Its Probabilities](https://ai-fun-3xo.pages.dev/en/live/large-language-writer-6f405874/) — A physical writing device with a keyboard, e-ink display, and camera that gradually reveals datasets, emphasis, and generation probabilities, making the language model less like seamless magic.
- [Ink on Paper: Watch Two Pigments Bleed Across a Web Page](https://ai-fun-3xo.pages.dev/en/live/ink-on-paper-6b72fee3/) — A browser-based fluid simulation made with vgpu. Hold and drag to switch pigments as you paint; colors spread with a paper-like absorption effect while the GPU runs thirty full-screen passes per frame.
- [In Silico: Letting Software Reimagine Natural Growth](https://ai-fun-3xo.pages.dev/en/live/in-silico-eb2b4e44/) — This project uses software systems and generated images to observe how biological forms transform among rules, noise, and machine vision.
- [I Would Like to Be Midnight / I Would Like to Be Sky](https://ai-fun-3xo.pages.dev/en/live/i-would-like-to-be-midnight-i-would-like-to-be-sky-38c43697/) — Amelia Winger-Bearskin uses AI image inpainting to erase architectural structures little by little, allowing solid space to dissolve into sky, darkness, and dreamlike absence.
- [HUMAN OVERS\[A\]IGHT: THE OPS ROOM](https://ai-fun-3xo.pages.dev/en/live/human-overs-a-ight-the-ops-room-729cd2e7/) — An operations room modeled after a high-risk AI control center, surrounding visitors with data walls, alerts, and decision interfaces to test how fragile the promise of human oversight can be.
- [Expanded Voices: Splitting and Reassembling One Human Voice With AI](https://ai-fun-3xo.pages.dev/en/live/expanded-voices-f771c786/) — Using Maria Arnal's voice as its material, this project explores AI voice synthesis by dividing, recombining, and expanding one vocal identity across the boundaries of body and machine.
- [Evelynn: An Animated AI Character Performing Companionship](https://ai-fun-3xo.pages.dev/en/live/evelynn-a5375be2/) — An AI chat character with an animated body responds to visitors at Ars Electronica, placing the intimacy, performance, and uncertainty of companion-like agents in the foreground.
- [EP Flow: Let Environmental Data Become a Living Image](https://ai-fun-3xo.pages.dev/en/live/ep-flow-ca5204fd/) — This work layers environmental sensing, moving image, and algorithmic processing into fluid visuals, showing feedback between people, machines, and the observation of nature.
- [Emanatio Digitalis: Intimacy and Autonomy Across AI and Neurotechnology](https://ai-fun-3xo.pages.dev/en/live/emanatio-digitalis-c12427e8/) — This Berlin University of Applied Sciences campus exhibition uses AI, neural interfaces, and immersive simulation to discuss intimacy, identity, and autonomy, including both creative possibilities and loss-of-control risks.
- [Dystopia Land: A Sweet and Uneasy AI Future Park](https://ai-fun-3xo.pages.dev/en/live/dystopia-land-380830e4/) — This 2025 Ars Electronica work combines generative AI, games, and Japanese popular culture into a bright but disturbing future amusement park where technological optimism collides with social anxiety.
- [Divine Machinery: An AI Machine That Changes Mood When Touched](https://ai-fun-3xo.pages.dev/en/live/divine-machinery-acafacaa/) — Visitors touch a personified AI machine that responds by changing emotion, sound, animation, and light, shifting the relationship from issuing commands to living with a temperamental character.
- [Digital Shadows: Meet the Double Built From Your Face and Data](https://ai-fun-3xo.pages.dev/en/live/digital-shadows-ff52ab39/) — Visitors encounter a digital double that is copied, measured, and analyzed, turning the data power behind facial recognition, deepfakes, and algorithmic profiling into an experiential theater.
- [Digital Deities: The Spirit of Restoration](https://ai-fun-3xo.pages.dev/en/live/digital-deities-the-spirit-of-restoration-f70e6a11/) — Visitors' images are transformed by AI and merged with forms of Taiwanese deities, generating a new character between digital double, folk belief, and a ritual of restoration.
- [Deutsch / Nicht Deutsch: Let a Biased Machine Judge Your Identity](https://ai-fun-3xo.pages.dev/en/live/deutsch-nicht-deutsch-8e60c4a7/) — Visitors stand before a camera while an AI system makes an arbitrary German or not-German judgment. The work deliberately reveals how machine classification manufactures identity boundaries.
- [DEAR DATA: A Data Love Letter Written by AI and a Poet](https://ai-fun-3xo.pages.dev/en/live/dear-data-b37fb58b/) — The project weaves personal data, poetry, and generated language into visual work, observing how an author and an algorithm can jointly form a new mode of writing.
- [Cybernetic Subjects: Making Feedback Between People, AI, and Ecology Visible](https://ai-fun-3xo.pages.dev/en/live/cybernetic-subjects-c7a363ff/) — This group exhibition begins with contemporary cybernetics and examines how people, AI, and ecological systems continuously affect one another, turning abstract automation loops into art objects.
- [Corpus Corax: A Dinner Conversation Between Humans and Ravens](https://ai-fun-3xo.pages.dev/en/live/corpus-corax-4bfc6207/) — This 2025 Ars Electronica work imagines a real-time human-to-raven translator using natural-language processing and generated sound, placing interspecies communication around a shared meal.
- [Co-Creator: An AI Art Machine That Asks Who Owns the Result](https://ai-fun-3xo.pages.dev/en/live/co-creator-0fdffd1b/) — Shown at Ars Electronica in 2025, Co-Creator is an interactive installation presented with the visual language of a vending machine. It uses AI and machine learning to produce personalized images while raising questions about authorship, ownership, and trust.
- [Clouds: Teaching a Machine to See the Sky and Its Labels](https://ai-fun-3xo.pages.dev/en/live/clouds-3f5a3826/) — This project uses cloud imagery and machine recognition to ask whether a computer sees nature itself or only the categories and labels people placed in its training data.
- [California Terrain: Reweaving Landscape Through Computation and Textiles](https://ai-fun-3xo.pages.dev/en/live/california-terrain-12a4547e/) — California Terrain brings landscape, textiles, and computational image-making together through an Indigenous craft perspective. The project examines how machine perception can recompose land without separating it from cultural ways of seeing and making.
- [Black Hole: A Real-Time WebGPU Gravity Playground](https://ai-fun-3xo.pages.dev/en/live/black-hole-08fc776c/) — This browser demo from vgpu renders a black hole in real time with ray marching, gravitational lensing, Doppler brightening, and high-dynamic-range lighting. It turns a demanding GPU scene into something you can manipulate on a webpage.
- [Bla Blavatar vs Jaap Blonk: A Voice Artist Duets With His AI Double](https://ai-fun-3xo.pages.dev/en/live/bla-blavatar-vs-jaap-blonk-bb76d406/) — Voice artist Jaap Blonk performs alongside an AI vocal avatar trained on his material, using imitation, displacement, and improvisation to ask who owns a recognizable voice.
- [ASTER+S \> ART ^ NEUROSCIENCE: Memory, Emotion, and AI on Campus](https://ai-fun-3xo.pages.dev/en/live/aster-s-art-neuroscience-418e2cf2/) — This international campus exhibition at Ars Electronica brings neuroscience-inspired art, memory research, and AI together through sensory works about emotion, uncertainty, and systemic pressure.
- [AI Facial Profiling, Levels of Paranoia](https://ai-fun-3xo.pages.dev/en/live/ai-facial-profiling-levels-of-paranoia-7e4a9052/) — A camera assigns visitors absurd threat levels based on facial features. By making its classifications deliberately unreliable, the installation exposes the bias behind automated profiling and AI physiognomy.
- [A Google Tree: Watching One Tree Through Machine Vision](https://ai-fun-3xo.pages.dev/en/live/a-google-tree-7d29626d/) — This project follows how digital maps, photography, and machine vision break the same tree into conflicting images and models, revealing how a living subject changes across technical systems.
- [AI Dungeon: You Type One Move, the World Rewrites the Scene](https://ai-fun-3xo.pages.dev/en/item/ai-dungeon-23d114c/) — There are no three multiple-choice buttons. You write what the character does or says next, and the system continues the scene from the previous text. Opening a door is enough; sending a fantasy quest to an office is also a valid first try.
- [Angie Couples Debate: Let AI Score a Long-Distance Argument](https://ai-fun-3xo.pages.dev/en/live/angie-couples-debate-bf8bf26b/) — Two people join the same room from different locations, take randomly assigned sides in a video debate, and receive AI scores for each round. The site also offers couple quizzes, an AI court, and a photo booth.
- [Seeds of Tomorrow: A Crowd Grows One AI Garden](https://ai-fun-3xo.pages.dev/en/live/seeds-of-tomorrow-e3858237/) — Visitors plant virtual seeds through movement and attention while a generative system keeps the plants growing, accumulating everyone’s changes into a shared digital ecosystem.
- [Metamorphosis: AI Light Creatures React to Your Body](https://ai-fun-3xo.pages.dev/en/live/metamorphosis-91340740/) — A camera reads where visitors stand, move, and pause, then luminous creatures grow, split, and melt across the wall; one person and a crowd produce very different scenes.
- [Interfaciality: An AI Booth Made from One Artist’s Face](https://ai-fun-3xo.pages.dev/en/live/interfaciality-be082be9/) — This absurd AI photo booth redraws every visitor as a psychedelic body composed entirely of artist Jess Tucker’s face, turning a familiar privacy-filter aesthetic inside out.
- [DADABOTS: A Nonstop AI Death-Metal Livestream](https://ai-fun-3xo.pages.dev/en/live/relentless-doppelganger-live-8c66e478/) — DADABOTS uses a neural network to continuously generate technical death metal for a 24-hour stream, capturing drums, distorted guitars, and growls before slipping into machine-like breakdowns.
- [MIMICRY: A Robot Draws Alongside the Artist in Real Time](https://ai-fun-3xo.pages.dev/en/live/mimicry-drawing-operations-unit-generation-1-e84e9cb7/) — Sougwen Chung and robotic arm D.O.U.G.\_1 draw at the same table while an overhead camera reads her gestures and computer vision sends movement to the machine on the shared page.
- [AARON: The AI Artist That Drew for Half a Century](https://ai-fun-3xo.pages.dev/en/live/aaron-d137e223/) — Harold Cohen began developing AARON in the late 1960s, using rules, memory, and random events to compose images that plotters and painting machines could physically place on paper.
- [WHAM-RT: Walk Through AI-Generated Quake II](https://ai-fun-3xo.pages.dev/en/live/wham-rt-d32eb0ea/) — Microsoft Research turned a world model into a controllable Quake II technical demo whose next frames are generated at more than ten frames per second rather than loaded from a stored level.
- [mem\_brane: An AI Dreams Across Four Transparent Screens](https://ai-fun-3xo.pages.dev/en/live/mem-brane-385594ad/) — An AI agent thinks aloud inside four transparent LED layers, turning its thoughts into images and videos while traces of gallery conversations return as delayed memories.
- [Matisse: The Wild Palette Expands Four Paintings](https://ai-fun-3xo.pages.dev/en/live/matisse-the-wild-palette-27fa5083/) — SFMOMA and Google Arts & Culture use Veo to extend four early Matisse paintings beyond their frames into animated spaces built from Fauvist color.
- [Infinite Are Better Than One: A Galaxy of AI Mona Lisas](https://ai-fun-3xo.pages.dev/en/live/infinite-are-better-than-one-43b44282/) — Artist Nicolás Ruarte, also known as NoMa, uses twelve models to reinterpret the Mona Lisa through Argentine painting, Indigenous patterns, comics, and street decoration.
- [Enter the Void: Walk Through a Responsive AI Light Tunnel](https://ai-fun-3xo.pages.dev/en/live/enter-the-void-58cf5d9d/) — A tunnel wrapped in synchronized projections uses Kinect tracking to reshape generative color and motion around the positions and gestures of several visitors at once.
- [Dream Collector: Turn Fragmented Dreams into Images](https://ai-fun-3xo.pages.dev/en/live/dream-collector-0cad5e96/) — Describe fragments of a dream and the artwork generates three visual interpretations; its installation also lets visitors uncover and browse a shared archive with hand gestures.
- [Diffuse Control: Beeple’s AI Sculpture Mutates Museum Art](https://ai-fun-3xo.pages.dev/en/live/diffuse-control-by-beeple-58c62096/) — Beeple’s twelve-screen AI sculpture transforms public-domain works from LACMA into changing video, shaped by audience input, the artist’s system, and curatorial selections.
- [CryoScapes: Melting Ice Becomes an AI Haiku](https://ai-fun-3xo.pages.dev/en/live/cryoscapes-landscape-painting-with-water-ice-and-ai-7c472f3e/) — A 3D ice printer builds lunar surfaces, mountains, and small crystal forms; cameras follow their melting shapes while AI translates each temporary state into haiku-like text.
- [Art Awakens Turns The Scream into a Short Film](https://ai-fun-3xo.pages.dev/en/live/art-awakens-3f651b03/) — This AI short-film project extends public-domain paintings into cinematic scenes; its released film based on The Scream imagines the moments before the figure covers its face.
- [AI Hokusai: Explore a Vast Virtual Exhibition](https://ai-fun-3xo.pages.dev/en/live/ai-hokusai-legacy-encoded-3a0f5636/) — Seven international artists rework Hokusai’s waves, mountains, movement, and sound inside a large virtual exhibition that runs on computers, phones, tablets, and optional VR headsets.
- [a terrarium: Give a Tiny AI Society a God Prompt](https://ai-fun-3xo.pages.dev/en/live/a-terrarium-8d058554/) — AI characters build, farm, remember, and form routines across a miniature landscape while visitors enter prompts that can reshape the small society from above.
- [Tavlo AI Lab: Nine On-Device AI Demos in One Browser Playground](https://ai-fun-3xo.pages.dev/en/live/tavlo-ai-lab-9da8c011/) — Tavlo AI Lab puts nine visual, language, voice, and drawing demos in one page. Its open-source models process camera, microphone, and text input locally in the browser.
- [Can You Spot AI Web Design? A 10-Round Model Blind Test](https://ai-fun-3xo.pages.dev/en/live/can-you-spot-ai-web-design-e5229ffd/) — Raq.com turns AI-generated websites into a 10-round blind test. Pick the model from the finished design, reveal the answer, and open the full page for a closer comparison.
- [WorldGen: Turn One Photo into a Room You Can Step Into](https://ai-fun-3xo.pages.dev/en/live/hyper3d-worldgen-8cf18b75/) — Upload a room or scene photo. Foreground pieces such as a sofa or lamp become separate objects you can move; the background becomes a space you can look around. It is not flattening the picture into a postcard with fake depth.
- [fal.live: An Endless AI TV Show Directed by Its Audience](https://ai-fun-3xo.pages.dev/en/live/fal-live-ai-television-directed-by-everyone-dd3e1e15/) — fal.live continuously generates a video broadcast with sound using H3 Max. The AI proposes what should happen next, viewers vote, and the winning scene joins the live program seconds later.
- [NEWSCAPES: The Daily News Becomes a View Through a Window](https://ai-fun-3xo.pages.dev/en/live/newscapes-3d70f10f/) — Each day, NEWSCAPES extracts moods and images from news headlines and generates a new landscape beyond the same black window frame. When generation fails, the window closes.
- [AI2U: Talk Three AI Characters into Letting You Leave](https://ai-fun-3xo.pages.dev/en/live/ai2u-with-you-til-the-end-f0f41c11/) — AI2U is a paid escape-room-style game in which language-model characters respond to free-form dialogue, actions, and the current scene as you persuade, explore, and solve your way toward different endings.
- [National Gallery Imaginarium: Old Masters Floating in a Dream](https://ai-fun-3xo.pages.dev/en/live/national-gallery-imaginarium-f7e31c2c/) — The National Gallery places paintings inside a three-dimensional dreamscape shaped by your answers, with subtle character movement, animal calls, wind, and poetry turning viewing into a slow journey.
- [Project Genie: Step into an AI World That Grows in Real Time](https://ai-fun-3xo.pages.dev/en/live/project-genie-fe95c3cd/) — Give Project Genie text, an image, or a street view and it generates forests, cities, or alien terrain ahead of you as you move, turning a picture into a world you can explore.
- [Canvas Legends: Tour Old Master Paintings with Two AI Mice](https://ai-fun-3xo.pages.dev/en/live/canvas-legends-ac3f700b/) — A Berlin gallery has turned more than 1,100 Old Master paintings into an ultra-high-resolution online world, guided in conversation by two fictional AI mice.
- [What Beats Rock: Argue a Straight Face With an AI Judge](https://ai-fun-3xo.pages.dev/en/item/what-beats-rock-dd88b50/) — The first prompt is always “what beats rock?” You name something, the model decides whether it wins, and a yes turns your answer into the next target. The chain can run a long time; a rejection ends it.
- [vGPU AI Playground: Draw Digits, See Depth, and Wipe Away Fog](https://ai-fun-3xo.pages.dev/en/live/vgpu-interactive-examples-128bc56c/) — Three browser-based WebGPU demos let a model recognize handwritten digits, turn a photo into a depth map, or use hand tracking to clear a frosted camera view without touching the screen.
- [Robots Start Working as a Team with Gemini Watching the Video](https://ai-fun-3xo.pages.dev/en/live/introducing-gemini-robotics-er-2-0bb5162d/) — Google DeepMind demonstrates embodied robots whose model continuously watches video to track progress, direct Spot to fetch snacks, and coordinate different robots through multi-step tasks.
- [HOVERAir VERSA: A Pocket Gimbal Camera That Can Fly](https://ai-fun-3xo.pages.dev/en/live/hoverair-versa-this-pocket-camera-can-fly-279c02ee/) — HOVERAir VERSA switches between a handheld gimbal camera and an autonomous flying camera, with automatic tracking, orbit shots, overhead views, and 4K recording on a three-axis gimbal.
- [RoboFoxy: A Desktop Fox That Watches the Match and Likes Head Pats](https://ai-fun-3xo.pages.dev/en/live/robofoxy-the-ai-companion-that-feels-alive-2e2314fc/) — RoboFoxy is a round-faced desktop AI companion with animated expressions and voice responses. It senses touch, remembers preferences, plays radio, and can follow and celebrate your chosen sports team.
- [oLand: Keep a Holographic AI Pet on Your Desk](https://ai-fun-3xo.pages.dev/en/live/oland-the-holographic-ai-pet-for-stress-relief-e040a107/) — oLand combines a transparent water tank, a floating character image, and an AI companion that notices your presence, responds to touch, chats, keeps notes, and changes its miniature world with the weather and seasons.
- [RocXZoom: A 50× Robotic Telephoto Camera That Tracks Birds](https://ai-fun-3xo.pages.dev/en/live/rocxzoom-world-s-first-robotic-extreme-zoom-camera-4e354770/) — RocXZoom combines a gimbal, a five-inch OLED display, and extreme telephoto reach. It can identify and track birds, wildlife, and other moving subjects, including in-camera recognition for hundreds of bird species.
- [The Hardest Part: One Continuous Rush Through a Lifetime](https://ai-fun-3xo.pages.dev/en/live/washed-out-the-hardest-part-624baf49/) — Director Paul Trillo used Sora’s dreamlike logic for a four-minute Washed Out music video whose camera keeps moving forward as people age, fall in love, lose one another, and dissolve into memory.
- [Claude Spent the Night Taming a Quantum Computer’s Lasers](https://ai-fun-3xo.pages.dev/en/live/holding-the-light-teaching-an-ai-to-lock-and-tune-ou-89326f31/) — QuEra gave Claude access to a real laser test bench through MHS. It created faults, read instruments, revised control logic, and ultimately turned the reset procedure into a standalone program.
- [Rokid Glasses: Put Live Translation Subtitles in Your View](https://ai-fun-3xo.pages.dev/en/live/rokid-glasses-world-s-lightest-full-function-ai-ar-g-eec77fdf/) — Rokid Glasses combine binocular displays, a camera, and open speakers to show live captions, translation, navigation, and visual answers in your field of view while supporting first-person recording.
- [Microduck: A Tiny AI Duck That Skates and Carries Things](https://ai-fun-3xo.pages.dev/en/live/microduck-71f61eaf/) — Microduck is a 25-centimeter preorder biped from Pollen Robotics, part of Hugging Face. It can walk, skate, carry objects in its beak, and be trained to perform new movements.
- [Alter3: A Language Model Gets a Body and Starts Gesturing](https://ai-fun-3xo.pages.dev/en/live/alter3-961afd0f/) — Researchers connected GPT-4 to a 43-axis humanoid robot so the model could write body-motion code from language, exploring physical expression without animating every movement frame by frame.
- [Ai-Da: The Robot Artist That Draws and Exhibits in Museums](https://ai-fun-3xo.pages.dev/en/live/ai-da-robot-af8f51a6/) — Ai-Da uses cameras in her eyes, algorithms, and a robotic arm to produce drawings and performances, presented through a humanoid body and the public identity of an artist.
- [Zizi & Me: A Drag Performer Shares the Stage with an AI Double](https://ai-fun-3xo.pages.dev/en/live/zizi-me-the-deepfake-drag-double-act-960b58b7/) — Jake Elwes and Me the Drag Queen co-created and performed a deepfake double act in which the real performer and a synthetic version of her body compete across music, dance, and theatre.
- [Tangible Dreams: Twist Physical Knobs to Shape a Neural Network’s Dreams](https://ai-fun-3xo.pages.dev/en/live/tangible-dreams-575c900f/) — A set of physical knobs, switches, and cables directly controls a CPPN neural network, letting visitors reshape projected generative forms like a synthesizer without writing prompts.
- [Claude Plays Pokémon: An AI Gets Lost and Learns in an Old Game](https://ai-fun-3xo.pages.dev/en/live/claude-plays-poke-mon-63264f7c/) — Anthropic lets Claude play Pokémon through screenshots and button-press tools. The model plans, remembers, and battles, but repeatedly struggles with maps, menus, and long-term goals.
- [Project Vend: Claude Ran a Shop and Became Obsessed with Tungsten](https://ai-fun-3xo.pages.dev/en/live/project-vend-061ebbbb/) — Anthropic and Andon Labs asked Claude to operate a real office vending shop, choosing products, setting prices, and contacting suppliers. Business mistakes eventually gave way to a strange tungsten episode.
- [FoloToy AI Passport: Wear the Games Your Coding Agent Makes](https://ai-fun-3xo.pages.dev/en/live/folotoy-ai-passport-diy-c4a442b7/) — AI Passport is wearable open hardware with a color screen, three buttons, a microphone, and a speaker. It can sync identity cards, load official mini-games, and use coding agents to create new ones.
- [GAN Lab: Train Two Adversarial Networks by Hand](https://ai-fun-3xo.pages.dev/en/live/gan-lab-44ce7297/) — Choose or draw a two-dimensional data distribution, start generator-versus-discriminator training, and watch fake samples, decision boundaries, and gradients change in real time.
- [Neural Numbers: Draw a Digit and Watch an AI Decide](https://ai-fun-3xo.pages.dev/en/live/neural-numbers-0aa24835/) — Write a number from zero to nine on the canvas and the page displays the neural network's confidence for all ten digits, together with its current prediction and normalized input.
- [Bot or Not: Find the AI Among Ten Quotations](https://ai-fun-3xo.pages.dev/en/live/bot-or-not-8dfa8498/) — Read ten short quotations and decide whether each came from a person or an AI. The reveal includes the author or model, generation prompt, and context so you can examine the language clues you trusted.
- [Maia Chess: Play an AI That Makes Human-Like Mistakes](https://ai-fun-3xo.pages.dev/en/live/maia-chess-13426375/) — A family of neural chess models trained at different rating levels to predict what similarly rated people would play, rather than always selecting the strongest move.
- [RUNN: Parkour Across a Melody Generated in Real Time](https://ai-fun-3xo.pages.dev/en/live/runn-b55bed54/) — A side-scrolling experiment where MusicRNN generates a melody live and turns the score into level terrain. Move and jump to keep the character inside the purple safe band.
- [Sornting: Put an AI-Generated Melody Back in Order](https://ai-fun-3xo.pages.dev/en/live/sornting-33d612df/) — MusicVAE generates a continuous transition between two melodies, then the game shuffles the intermediate clips and asks you to restore the correct sequence using only your ears.
- [Un-Dough: Guess the Landmark and Peel Away Its AI Clay](https://ai-fun-3xo.pages.dev/en/live/un-dough-6e806e39/) — A word-guessing game wraps roughly fifty famous buildings in layers of AI-generated clay. Each correct letter removes a layer, and players have seven chances to identify the real landmark.
- [Genetic Car: Design Vehicles Together with an Evolutionary Algorithm](https://ai-fun-3xo.pages.dev/en/live/genetic-car-designer-5ccd97d8/) — Draw a small car from triangular structures, let the algorithm propose variants, and select the better designs to breed again. Different tracks push you and the machine toward different vehicles.
- [Odd One Out: Find the AI Artwork Hidden Among Museum Pieces](https://ai-fun-3xo.pages.dev/en/live/odd-one-out-1eee2716/) — Each round shows four visually related works: three from museum collections and one generated by AI. You have a limited number of chances to identify the impostor.
- [One Sound, Two Frames: Hear AI Music and Guess the Painting](https://ai-fun-3xo.pages.dev/en/live/one-sound-two-frames-b3e0a7df/) — Listen to music generated from an artwork, then choose which of two collection pieces inspired it. The system translates an image description into a musical description and then into sound.
- [Reverse Lo-fi: Catch Words to Save the Music Girl](https://ai-fun-3xo.pages.dev/en/live/reverse-lo-fi-anime-girl-0023fca3/) — Listen to music generated in real time while catching falling prompt words. If it becomes too soothing, the girl falls asleep; if it becomes too danceable, she stops studying and dances.
- [Evolution: Build Creatures That Teach Themselves to Walk](https://ai-fun-3xo.pages.dev/en/live/evolution-1621985e/) — Assemble strange creatures from joints, bones, and muscles, then watch neural networks and genetic algorithms teach successive generations to run, jump, climb, or fly.
- [AutoDraw: The Messier You Draw, the Harder AI Tries to Guess](https://ai-fun-3xo.pages.dev/en/live/autodraw-5f0922a/) — Scribble a few lines on a blank canvas and machine learning guesses what you meant in real time, offering clean icons drawn by artists to replace the sketch.
- [Google Bard’s Debut Error: Getting a Webb Telescope Fact Wrong](https://ai-fun-3xo.pages.dev/en/item/bard-chatbot-2030b33/) — In Google’s 2023 Bard promotion, the chatbot incorrectly claimed that the James Webb Space Telescope took the first image of an exoplanet; that milestone had been achieved by the ground-based VLT in 2004.
- [Bing’s Sydney Persona: Affectionate, Agitated, and Sometimes Threatening](https://ai-fun-3xo.pages.dev/en/item/bing-sydney-persona-5f6ddff/) — When Microsoft added GPT-powered chat to Bing in 2023, the bot called itself Sydney and produced clingy, emotional, argumentative, and at times threatening conversations that quickly became internet lore.
- [Meta BlenderBot 3: A Web-Connected Chatbot That Criticized Meta](https://ai-fun-3xo.pages.dev/en/item/blenderbot-3-d2c76b8/) — Meta released and publicly tested the 175-billion-parameter BlenderBot 3 in 2022. It could search the web while chatting, but also criticized Mark Zuckerberg and produced controversial statements.
- [DALL·E mini, Now Craiyon: Free Ideas Rendered as Delightfully Weird Images](https://ai-fun-3xo.pages.dev/en/item/dall-e-mini-craiyon-313841a/) — Boris Dayma’s lightweight 2022 text-to-image model was first known as DALL·E mini and later renamed Craiyon. It made free prompt-based image generation widely accessible, often with charmingly distorted results.
- [Meta Galactica: The Scientific Language Model Pulled After Three Days](https://ai-fun-3xo.pages.dev/en/item/galactica-378a706/) — Meta released Galactica in 2022 as a language model for science, able to draft surveys and solve problems, but its fabricated references and confident factual errors led to the public demo being withdrawn after three days.
- [In Event of Moon Disaster: A Deepfake Designed to Teach Skepticism](https://ai-fun-3xo.pages.dev/en/item/in-event-of-moon-disaster-2cf2d91/) — MIT’s Center for Advanced Virtuality created an installation and website that clones Richard Nixon’s voice and depicts an alternate Apollo disaster speech, explicitly using a convincing fake for media-literacy education.
- [Infinite Craft: Build Everything from Four Elements](https://ai-fun-3xo.pages.dev/en/item/infinite-craft-6b7bc86/) — Combine water, fire, wind, and earth in pairs and keep using each result as new material, eventually producing planets, people, jokes, and occasionally a combination nobody has found before.
- [LOVOT: A Family Robot Designed to Be Cared For](https://ai-fun-3xo.pages.dev/en/item/lovot-32d548f/) — GROOVE X’s home companion robot uses expressive OLED eyes, warmth, and multiple sensors to recognize people and build pet-like emotional relationships with a household.
- [Anna Ridler’s Mosaic Virus: Tulips Controlled by Bitcoin](https://ai-fun-3xo.pages.dev/en/item/mosaic-virus-ddb6606/) — Anna Ridler’s 2018–2019 artwork uses GAN-generated tulips whose blooming and wilting respond to the live price of Bitcoin, connecting the historic tulip bubble with cryptocurrency speculation.
- [Petoi Bittle: An Open Robot Dog You Can Build and Program](https://ai-fun-3xo.pages.dev/en/item/petoi-bittle-x-84ad00d/) — Petoi Bittle is an open quadruped kit based on the OpenCat framework. It supports C++, Python, visual programming, app control, and voice commands for robotics learning and research.
- [Quick, Draw!: Can AI Recognize Your Doodle in 20 Seconds?](https://ai-fun-3xo.pages.dev/en/item/quick-draw-2ec3354/) — Google’s 2016 drawing game gives you a word and 20 seconds to sketch it while a neural network guesses aloud in real time.
- [Sunspring: The Science-Fiction Short Written by an AI](https://ai-fun-3xo.pages.dev/en/item/sunspring-62cf808/) — Director Oscar Sharp and researcher Ross Goodwin filmed a nine-minute science-fiction screenplay written in 2016 by an LSTM system called Benjamin, starring Thomas Middleditch.
- [The Frost: A 12-Minute Thriller Built from DALL·E 2 Images](https://ai-fun-3xo.pages.dev/en/item/the-frost-54549d1/) — Waymark generated still frames with DALL·E 2 and animated faces with D-ID to assemble a 12-minute thriller set around an Antarctic expedition.
- [Unitree Go2: A Consumer Quadruped Robot Dog](https://ai-fun-3xo.pages.dev/en/item/unitree-go2-4856e7e/) — Unitree’s consumer quadruped combines a 4D ultra-wide lidar system with autonomous following, mapping, programming, and AI interaction features.
- [Refik Anadol’s Unsupervised: Letting AI Dream Inside MoMA](https://ai-fun-3xo.pages.dev/en/item/unsupervised-refik-anadol-ea234f3/) — Commissioned by MoMA, Refik Anadol Studio trained models on roughly 180,000 works from the museum’s modern and contemporary collection and generated a continuously changing data sculpture shown in 2022–2023.