Teachable Machine

A cinematic exploration of human–AI collaboration, Teachable Machine visualises how we train machines to recognise emotion, expression, and meaning. It bridges curiosity and responsibility — showing that intelligence learns not only from what it sees, but from what we teach it to understand.

Introduction

The Teachable Machine project was designed as a visually engaging introduction for a new course on AI learning and facial recognition. Inspired by Google’s Teachable Machine concept, the course challenges students to train a system to recognise drawn faces and emojis while exploring real-world implications of recognition technology.
The goal of the video was to inspire both teachers and students — showing the power, potential, and responsibility of creating systems that “see” the world through human input.

Process

The project began with the idea of transformation — from code to consciousness. Visually, this was told through a journey of light and data: beginning with a cube of glowing circuitry (representing a machine’s untrained mind), gradually learning to form patterns, faces, and eventually entire networks of understanding.

Each scene was carefully designed within Veo 3 to reflect both the literal process of training and the abstract concept of growth.

Scene 1: The cube awakens — digital filaments light up as a system comes online.
Scene 2: Abstract “faces” form from glowing particles, symbolising recognition.
Scene 3–4: Networks expand, threads connect, and the AI begins to perceive relationships between expressions.
Scene 5–6: A vast luminous field of interconnected faces — representing collective intelligence.
Final Scene: The journey returns to the cube, now evolved — glowing softly as a symbol of awareness and equilibrium.

The narration and music worked in tandem — balancing awe and calm. Vocals and deep atmospheric tones guided pacing, while slow fades and light transitions emphasised the emotional arc of learning and creation.

Challenges and Solutions

The biggest challenge was conveying “understanding” visually without resorting to human imagery or literal emotion. This was solved through the symbolic use of light, geometry, and movement — every decision expressed through the motion of particles and circuits rather than faces or text.

Another difficulty involved maintaining thematic balance: showing both the promise and potential misuse of AI without creating fear or negativity. This was achieved by dedicating most of the video to positive growth and unity, with only subtle hints at the ethical weight of control in the final lines.

Technical limitations from Veo 3 — such as enforcing no text or music generation — were handled through precise prompting, extensive negative cues, and custom editing in Premiere Pro, where all audio and pacing were manually aligned.

Conclusion

Teachable Machine became one of the most conceptually rich and visually cohesive projects in the series. It distilled complex topics — AI learning, ethics, and empathy — into a poetic visual journey that engages both students and educators.
By ending where it began, the film symbolises reflection and responsibility: that what we teach our machines ultimately teaches us in return.