Understanding AI

The Activation Atlas is a method for representing processes in an artificial neural network during image recognition.

Anatomy of an AI uses the example of Amazon Echo to show the countless components and factors behind the production of artificial intelligence systems.

In Artificial Senses, the unprocessed sensory data of the machines that surround us is given a live visualization.

In November 2018 China‘s state news agency, Xinhua, presented its first AI newsreader.

Machine-based image recognition and generation can already produce fantastic results, but machines cannot always handle data perfectly.

Autonomous machines: Ars Electronica Center’s self-driving Donkey Cars can maneuver from A to B without human control.

The Embedding Projector by TensorFlow shows a visualization of vector spaces.

Can the term “sensory deception” also be applied to computers? Video Experimental

EYES is series of experiments with kinetic sculptures.

The TUM Visual Computing Lab by Matthias Nießner at the Technical University of Munich is experimenting with a face-transfer software in real time.

Forma Fluens (Latin for “flowing form”) shows an overlapping collection of drawings made by over a billion people.

GauGAN is an application that is able to generate complex landscapes with only a few brushstrokes.

Joy Buolamwini and Timnit Gebru investigated into the bias of AI facial recognition programs.

In this application, certain external features of people can be manipulated using a controller.

GPT-2 is a machine learning model that has been trained to complete English-language texts on its own.

Advances in artificial intelligence are shaping our everyday lives in many ways. At this point, you will have the opportunity to familiarize yourself with current applications.

The video by National Geographic explains how the eye has developed over the course of time, and takes on the question of how animals would be able to see the world.

Experience the performance of deep learning algorithms in a live environment.

Learning to See is an ongoing series of works that use the latest machine learning algorithms to reflect on how we understand the world.

Two hackers, known as greentheonly and DamienXVI, managed to get hold of a development version of Tesla’s autopilot hardware.

The project aims to provide a critical perspective on machine learning image datasets.

Moral dilemmas presented by street traffic.

Train different Neural Networks and see how artificial intelligence behaves based on your input.

Painters were invited and asked to create portrait paintings of a person that would not be recognised by facial recognition software.

Jordan Peele transferred his own facial movements to Obama’s facial characteristics using deep fake technology.

Two industrial robot arms as puppet players. Via an app developed by the research unit “Creative Robotics” at the University of Art and Design Linz, visitors can control the two robotic arms with the smartphone and the puppets attached to them can be moved.

With the pix2pix procedure, users can draw free sketches which are then turned into cat pictures by a conditional generative adversarial network (cGAN).

PixelPlayer is a system that learns to localize the sounds that correspond to individual image regions in videos.

Prompts are clearly formulated instructions or questions that are used to generate specific reactions or responses from an AI program. Visitors can practise their prompting skills at the Prompt Battle Training Station.

The ShadowGAN interprets mountains in everything it sees, even the human form.

Visitors are invited to act as a human version of Amazon Alexa.

Scientists at the University of California in Berkeley have created an interactive map that shows which brain areas react to hearing different words.

Can a machine “listen,” and can it “understand” what it listens?

These people do not actually exist. They were generated entirely by a machine—a specialized form of Generative Adversarial Network (GAN) called the StyleGAN.

All entries for the Prix Ars Electronica, a competition run by Ars Electronica, submitted since the Prix archive was established in 1987 are collected in this network.

This video from Experiments with Google explains how a multidimensional vector space works in machine learning.

Volumetric Data Collector is based on the idea of using a LiDAR sensor— a 3D laser sensor often used in autonomous vehicles—as an expanded sensory organ for the human body.

What a Ghost Dreams Of grapples with a new “ghost” of our time: digital surveillance in our society.