🃏 Recipe Card

<aside>

📌 Title — Catchy, descriptive name

</aside>

Machine Ways of Seeing: How Computer Vision Labels the World

<aside>

📌 One-liner — One sentence: what you do and why it matters

</aside>

Upload an image to a computer vision tool and probe how machines label the visual world — and why that same logic, run in reverse, is what powers every AI image generator.

<aside>

📌 Credits — Contributor name + link

</aside>

Recipe by Nadia Piet - independent researcher, designer, co-founder of AIxDESIGN, and the initiator of this Cookbook.

🧂 At a Glance

<aside>

📦 What you'll make — Two outputs, one or both depending on your session

</aside>

Three sets of labels for the same image — yours, a peer's, and a computer vision model's — plus a comparison and reflection on the gaps between them.

<aside>

⏱️ Suggested timing — Allow roughly equal time for each phase

</aside>

For 15 mins: move quickly through labeling and focus discussion on 2 questions. For 30 mins: go deeper into comparison and group discussion.

<aside>

🛒 Ingredients — Everything needed: tools, links, materials, access

</aside>

• A single still image to analyze — a movie or documentary still, a close-up portrait, an abstract or ambiguous image, a meme, a photo of your own, or any random image on your computer
• Access to a computer vision tool — pick one:
  • Imagga Auto-Tagging Demo — no signup needed, instant tags with confidence scores
  • Google AI Studio — requires Google account, more detailed analysis
  • Azure Computer Vision — no signup needed, shows object detection with bounding boxes
  • Clarifai — requires free account, strong on scene and concept recognition
• Paper, notebook, or any digital tool where you can take notes next to an image