Purpose

Create an installation for Universeum in Göteborg to introduce and inspire younger generation, specifically girls, to delve into Artificial Intelligence and the technology that makes it possible. In this case Neural Networks.

Concept

The installation was made in a pinball-like table and was divided in three stages.

First Stage- Guessing game

In this stage, the screen shows one picture of either a chihuahua or a muffin and the user has to press the button to identify what was shown. There are 10 pictures that are shown at a fast pace. Once all of the pictures are shown the number of correct answers is shown and the second stage starts.

Second Stage- Explanation Video

The second stage is made as a transition between the active stages. The video (in Swedish considering the context) shown in this stage roughly explains, and connects with everyday life neural networks. A female engineer from Ericsson also participated in the video. This is with the intention of making a subtle demonstration of women working in technology areas “if you see it, you can be it”. As it is more imaginable to see yourself as someone you can relate or identify with.

Third Stage- “Teaching” game

In this last stage, the user has to fiddle with three nodes of the “neural network” that represent different attributes so the machine can recognize if the picture is either a chihuahua or a muffin. In order to win, the user has to find the perfect balance between the values of the attributes so the machine identifies correctly all the pictures.

How it works

This installation uses a Mac mini that runs the code for the pictures, video and the LED sequences. All the LED are RGB. For the first stage the strips use pink for the muffin button and blue for the chihuahua button (pattern that we kept for the whole installation). During the second stage the LEDs are off to not distract the user while the video is playing. In the last stage, the strips are white with different levels of brightness in each path. The LED rings have blue when turning to the right and pink when turning to the left which corresponds with the mapping of the buttons this is controlled with a potentiometer and code.

We were a team of five people. I participated in the ideation, definition, and the interaction and visual parts of the development. After we decided upon the theme of neural networks, the pinball-like table, and how the general interaction should be divided in three stages, the team was divided in sub-groups where the programmers (2) focused in making the LED strips behave how it was decided, with sequences and dimming as well as connecting them to the buttons and knobs. Another group, which I was part of, was the design group where I made sure the aesthetic was consistent all through-out the experience, as well as created the explanation video coordinating the recording session with another team member and an engineer from Ericsson. One person was in charge of building the physical table where I participated for a few parts. And lastly the last person was in charge of producing a digital prototype for the interaction of the last stage.

Evaluation

There was some instances of user testing of the stages, more so of the 3rd stage as we felt this was the core part of the installation. We used a tablet to test a prototype before involving the LEDs and the potentiometers.

user test.jpg

Expo at Universeum

The installation was exhibited in Universeum the 26th and 27th of October 2018 for the general public. Here the target group, girls from different ages were able to interact with the table.

We were invited to exhibit the installation again for the inauguration of “The HUB” at Universeum the 8th of February 2019.

IMG_5511.jpg