AI AdventureAICreate with AIDay 6
🖼️ Create with AI · Day 6

Pictures Into Words

You've made AI write and AI draw. Today it does something wilder: it looks at a picture and writes a sentence about it — out loud, in words. Draw literally anything, hit describe, and a real AI on your own laptop will tell you what it sees.

↓ draw something
✏️ Lab · Draw & Describe

Draw it. Let the AI tell you what it sees.

Scribble a cat, a house, a spaceship — whatever. Then press 🔮 Describe it. A real image-captioning model (running 100% on your device) will look at the pixels and generate a sentence. No lookup, no cheating — it reads the picture and invents words to match.

…or drop in a ready-made picture:
Draw something (or pick a picture) and smash Describe it. 👆

The photo buttons need internet the first time. The drawn scenes and smiley work totally offline. 🌐

🐞 Break-It · The Confident Guess

Give it garbage. It'll still sound sure.

Make a random rainbow mess — pure abstract scribble that means nothing. Then describe it. The model will never say "uh, I don't know." It always produces a confident sentence, even when there's nothing real there.

Make a mess, then ask it what it sees. 👀
The lesson: it never says "I don't know" — it always guesses. When an AI describes something that isn't really there, that's called a visual hallucination. Real captioning AIs do this too — which is why you never fully trust one alone. 🧠
🏷️ The pro wordsTurning a picture into a sentence is called image captioning (or image-to-text). It's multimodal — it mixes two kinds of data, pixels and words. Inside, a vision encoder reads the image into numbers, then a text decoder generates the words one at a time. A model that can see pictures and talk about them is called a VLM (Vision-Language Model) — the tech behind every "describe this photo" feature.
🎯 Boss · Caption Judge

You're the quality checker

The AI captions a picture. Your job: is the caption a good match or wrong? If it's wrong, pick the caption that actually fits. Judge 3 pictures correctly to earn the badge. This is a real job — humans grade AI captions to make them better!

The AI says:
Correct: 0 / 3
🚀 Beyond the Basics

Where "pictures into words" really lives

Captioning isn't a toy — it quietly runs inside apps you use. Same core move every time: a machine looks, then writes.

🦯

Alt-text (accessibility)

Blind users browse with a screen reader. Captioning writes the "alt text" so they can hear what every photo shows. Genuinely life-changing tech.

🔎

Visual search (find by meaning)

Search your camera roll for "dog on a beach" with no tags. The photos got captioned, so now words can find them.

🛡️

Content moderation (safety)

Sites auto-describe billions of uploads to flag ones that break the rules — far faster than any human team could.

🏷️ Pro names to look upBig VLMs you'll hear about: GPT-4o, Claude, Gemini, and open ones like LLaVA and BLIP. The tiny one you ran today is ViT-GPT2 — a Vision Transformer encoder bolted to a GPT-2 text decoder.
🔬 Real Model Lab

The real captioner, on real photos

Everything above already used a real model — this is the same one, pointed at real internet photographs so you can see how it does on the hard stuff. It's ViT-GPT2, running 100% on your machine. Load it once, then describe any photo.

not loaded

First load pulls the model from Hugging Face, then it's cached for offline use. On a 4GB laptop give it a minute. ☕

🎉 Day 6 Complete!

You ran a real multimodal AI: a vision encoder read your pixels, a text decoder generated a sentence. You also caught it hallucinating — sounding sure about nonsense — the one thing to always watch for. That's exactly how alt-text, visual search, and every VLM work.

Day 7: The Meaning Map →
Tomorrow: how does typing "dragon" pull a picture toward dragons? We open the secret map where words and pictures live in the same place. 🗺️
AI Adventure · Create with AI · Day 6 — made for young inventors 🚀