HOW MACHINES SEE THE WORLD
How does artificial intelligence perceive its environment?
For the installation ‘How Machines See the World’ I trained a machine learning algorithm solely on the images of one plant. Thus the plant became the only thing the computer knows and recognizes.
There is a lot of buzz on how smart algorithms decide on our lives and take control, while it is often forgotten that any algorithm, no matter how advanced, is still created by people with certain agendas and biases.
To display this thought I trained a machine learning algorithm solely on the images of a single plant.
Hence the plant became the only thing the machine knows and recognizes. The algorithm then tries to turn everything it sees into what it knows.
Visitors standing in front of the installation are live recorded by a webcam and can see their own image converting into a plan by the smart, yet bias algorithm.
RESEARCH & EXPERIMENTS
Can Artificial Intelligence be creative?
A philosophical contemplation about machine vision of intelligent technologies -
Do machines only perceive the shadow of our reality or did they already move beyond?
At the beginning of my experimentations with machine learning algorithms my goal was to co-create output with the help of a trained algorithm.
While my ambition was to just 'throw' data at the computer and let it create, I did realize that training the algorithm was much more complex.
For a creative output I was expecting the computer to surprise me with its interpretations, but I soon realized how important the dataset is to train the algorithm and the impact of the bias I inititated by choosing the machine's input.
Input were simple drawings of passport photos (see above).
The algorithm was then able to create drawings with the same style from any new passport photo with similar properties. (left)
Furthermore I was able to input a fictional drawing of a face I drew myself, that the computer could then translate into a photograph. (right)
‘’In the end, algorithms are tools. People build them, determine if their output is accurate, and decide when and how to act on that output. Data can provide insights, but people are responsible for the decisions made based on them.‘’
Jennifer M Logg, "Using Algorithms to Understand the Biases in Your Organization", Harvard Business Review, 2019
Unconditional learning algorithm trained on the Google results for Marcel Duchamps fountain
Experimentation with live image translation of webcam input