Future Things
Using machine learning to enhance and foster human creativity
September 2020
Project created during the Machine Learning Course held by Andreas Refsgaard and Jen Sykes at the Umeå Institute of Design
What is this about?
Artificial Intelligence is one of the most disruptive tools to our disposal. Can designers integrate it in their workflow and make good use of it fostering human qualities?
This is what this project is about, take Machine Learning algorithms and make them integral part of the design and idea generation process.
About the project
When:
September 2020
1 week project
Learning outcomes:
• Training ML models in Runway
• Javascript to interface with the trained models and generate limitless cards.
The team
My project team :
Joan Pablo Farré - IxD (UID)
Manu Revi - IxD (UID)
My role:
Deisgn Research | Conceptualisation | Graphic Design | Development
Tools used:
p5.js | Runway and Style GAN model | GPT-2
A tool to imagine with
We all know how difficult it is sometimes to break the ice at the beginning of a conversation, being this with a person or about a topic we are not confident about.
This project aimed to create an ice breaker, a tool to make that initial barrier easier to overcome and help people unleash their immagination.
How we did it
The cards are generated using a combination of three tools:
1 - Javascript web application is written using p5.js
This program was the center of the system, coordinating the collection of the data and the creation of the final output.
2 - Style GAN model
This model was trained on hundreds of images of design objects from the present and the past.
3 - GPT-2 pre-trained text model
From a series of prompt a sentence was generated and sent back to the program to complete the card.