Today, the world faces the most significant migration in its modern history. Data from the United Nations highlights the existence of more than 244 million international migrants around the world, as a result of wars, natural disasters, political and economic factors.

Simultaneously, technological advances in machine learning steer a wave of automation, driven by the opinion that machines are more efficient and objective and make fewer errors. However, do machines have a point of view? are machines objective? Or is their objectivity based on our biases and assumptions about what should be accepted or rejected by society?


As a speculative machine learning assignment, ABC (AI Border Control) explores not only the playful potential of a trainable machine but also the wider questions and debates revolving around A.I development. ABC exists at an airport, where border control agents have been replaced by trained, intelligent machines.

Travelers and immigrants crossing into another country first need to activate the machine with their biometric passports (fingerprints). After which they proceed to walk through a full body scan and finally a face scan. Based on five intentionally random variables – Happiness Index, Terrorist Tendencies, Laziness Index, Mental Diseases and Genetic Perfection, ABC accepts or rejects the travelers’ entry into the country.

This project was designed in an effort to highlight the biases embedded into trained systems and the risk of handing over decisional power to such machines.

The prototype is composed of two main components:

  • A fingerprint scanner programmed using an Arduino microprocessor and created using a force resistor sensor and an RGB LED.
  • A  pre-existing facial tracking algorithm, altered and trained using p5.js.