Tuesday 16 April 2019

Art Hackathon - "Artificial Emotional Intelligence"

This month's London meetup was not a talk or tutorial but a hackathon.

The theme, to be interpreted freely, was "Artificial Emotional Intelligence".


In setting the theme I had hoped some would consider that, although AI has made huge leaps forward, they have been in the ability to repeat a task, rather than in the ability to interact with humans at an emotional level.


Art Hackathons

The normal format for our meetups is a talk or a hands-on tutorial, but there has always been demand for a session where members just code to create.

I'm not particularly experienced in organising hackathons (- should we use post-its? -) but we've run three before:

  • "Future Pangs" - [blog]
  • "I Want To Believe" - [blog]
  • "In The News" - [blog]


They seem to be fairly well received.


The key features are they everyone works on openprocessing.org so there is a common programming language (javascript and p5js) and an easy ability to share their art and code through a simple we URL.

Some people want to work together with others, some along, and depending on the night, there are a number of more experienced creative coders willing to help others. This is one area I need to organse better.

There are always newcomers and first-time coders. In addition to in-person help, onlien tutorials are helpful too. I'm pleased my own guide has been well received - ignore that it is labelled "for kids":



A unique feature of our art hackathons is setting the mood! We turn down the lights and project a nice ambient nature video with the beautiful but unobtrusive classical music - Gymnopedies 3, Satie.

Creative bliss!


Selected Works

I was again really taken by the broad diversity of interpretations and approaches to the theme. The following are selected works.


Peepl
In the work, the artist has the idea of small bodies moving around the space driven by their own speed and direction. However, when two such "people" come close to each other, they pause from their high speed dash and remain close to each other for a short while - as if having a conversation.


This is an interesting take on particle systems, one I've not seen before.


Colour Mind
In this work, the artists working as a team, developed the idea of taking a canonical list of emotions and using a hash function to convert them into colours.


The work itself presents itself as three buttons which the viewer is impelled to click, resulting in a series of circles, coloured and labelled with one the emotions.


The artists insisted that there was no intended correlation between the colour chosen by the hash function and the meaning of the chosen emotion word. The reason was to place meaning, interpretation, and reaction firmly in the space of the viewer, not the code.



Hamlet's Emotions
In this unique work, the artist has taken the text of Shakespeare's Hamlet and as his code reads it line by line, it picks out key emotive words, and visualises them. The words fall into three categories, broadly strong negative, intermediate, and strong positive.

It is interesting to see this chart grow as the code reads the text.


I particularly liked the graphs at the bottom, which looks like cpu or load monitors, evoking the feeling of a monitor, perhaps a medical monitor, keeping an eye on the emotions in Hamlet as the tale proceeds.



Dots
In this work, the artist started from the particles theme which looks fairly basic but as the particles interact they change and flash.


The code is online, but to run it effectively you should delete the debugging print statement first. I hope the artist continues to work on this code because the ideas are interesting and I think the fruits could be even more interesting.



Cellular Growth
This work is quite striking in how it is presented, growing out from the centre of the canvas.

The patterns formed by the colours seem natural, not entirely random, with a definite organic structure.


Looking through the code, I think I'm right that the pattern grows based on cell creation determined by its neighbours. It reminds me of code I first saw in the 1980s (!) which seemed to magically create detail based on a matrix of neighbouring cells, but following a formula that was sufficiently organic and not-so-random to make it feel natural.

Ben later explained that he wanted to show generations learning how best to communicate with their neighbours. He got as far as cells reproducing and their children taking traits from the parent(s). A cell's traits represent their emotional disposition, shown here by the red/blue colour. His intention was to use a genetic algorithm that learns an optimal way for cells to communicate, and the visualisation to show this learning over generations.

I can see echoes of the artist's previously presented work on grids and game theory!

You can try the code yourself:




(Miro-Like) Shapes
This work I is particularly effective. One one level it simply draws some simple shapes - circles and triangles, and some curling curves.


But the artists careful and thoughtful constraints on the colour palette, on the sizes and orientation of the shapes, the amount of curl and varying thickness of the wavy lines - all combines to create a powerful composition, very reminiscent of the works of Joan Miro.

The code produces different compositions every time it is run - and almost all of them seem to work.


Simple - yet sophisticated!

For me, I the algorithm that draws the swirls is particularly effective and I can't wait to explore the code to see how it works...



Thoughts

The art hackathons are always popular, and the show-n-tells at the end are, in my experience, the best part of the evening. It is inspiring and exciting to see the ideas that others have come up with, even if they haven't managed to finish them in the session itself.

I still feel I need to organise the hackathons better, and would love suggestions. At times they can feel like nobody knows what supposed to be happening. Maybe I can project a schedule of what's happening, and also make very clear that finished work is far from expected.

A useful feedback was to extend the coding time - one hour isn't enough!



I haven't run a creative coding hackathon in Cornwall yet, so that will happen at the next available date.

No comments:

Post a Comment