How Normal Am I? An experience to learn how AI judges your face
by Tijmen Schep
A new interactive documentary called “How Normal Am I” let’s you experience how AI judges your face. It reveals how algorithms that score us on beauty, age, gender, emotion, body mass index, and even life expectancy are increasingly finding their way into society.
Tinder, for example, uses these beauty scores to match people who are about equally attractive, while predicting your BMI from just a photo is used in the health insurance industry.
By giving access to your webcam you can also experience how these AI systems rate your own face. To protect your privacy the AI models were modified so that they can run in the browser. All the analysis happens on your own device, and no personal data is sent to the cloud.
The documentary’s maker and host, Dutch artist, technology critic and privacy designer Tijmen Schep, hopes the project will help society question the reliability of these systems. For example, the age prediction AI can be manipulated my moving your face up
and down, your beauty score may go up if you turn up the lights, and if you want to get a lower BMI score, all you need to do is raise your eyebrows. Besides being easy to manipulate, their biases are also revealed. The beauty scoring algorithm was trained on photos that were given beauty scores by Chinese students. This has given it a Chinese sensibility when it comes to what it deems attractive.
If you give permission to do so, then some anonymous statistics can be shared at the end of the documentary. Your anonymous scores are then used to compare you to all the people who have already experienced the documentary. This is more true to reality, Schep explains, as these AI systems are increasingly used to rank us in relation to others. Schep worries that In the long run these profiling systems may incentivise us to be as average as possible.
The English version was made with support from the European Union, where it’s part of the Sherpa Project, a research consortium which looks into AI issues that society might run into by 2025. SHERPA analyses how AI and big data analytics impact ethics and human rights. In dialogue with stakeholders, the project is developing novel ways to understand and address these challenges to find desirable and sustainable solutions that can benefit both innovators and society.