An ongoing program conclusion thesis project which investigates the potential of using machine learning as a tool in graphic design. Currently, the idea of artificial intelligence taking jobs is a heavily debated topic. We wonder how the creative industries will be affected by new AI technologies and how designers will adapt to these advancements. Through extensive research and my own design process I came to the conclusion that a machine cannot replace human creativity. The machine can be a helpful tool, but it is the designer that creates significance visually. Using the coding language Python, I trained an artificial intelligence machine based on an input of 700 different typefaces to generate new letterforms. Taking advantage of the machine error in the letterforms, I've been working on using them in compelling visual imagery and compositions. 
View the process book here.
A few examples of the wide range of different typefaces used to train the AI machine. Typefaces with quirky and unique characteristics were preferred so the machine could generate more interesting letterforms.
Below are some examples of the machine's output. The machine generates a series of PNG files with different sets of letterforms that include numerals, uppercase and lowercase letters. It's interesting to see how the AI can generate close to perfect letterforms as well as really strange letterforms filled with machine errors.
These are some of the different ways that the machine can generate the same letter, letter "A" in this case. In some cases the AI doesn't perfectly generate a legible letter, but the faulty letters are quite interesting.
Below are various different generated "A"s in motion. The GIFs show how the machine generates a series of letters that slightly morph into each other. The morphing letterforms in motion create visually appealing textures and liveliness.
The letterforms were optimized while still maintaining the irregularities and textures. The optimization would allow for larger scale compositions and screen printing. 
The following are pieces intended to be exhibited in GradEx together as an installation. The intent of the pieces are to create a juxtaposition between what is machine made, which has no conceptual meaning, and what is made by the designer, who is capable of real creativity and creating meaning with art. 
A video installation intended for GradEx that loops endlessly. The letterforms were meticulously typeset by hand to say "imbue sense" and animated to morph into each other. The idea is to contrast the meaningless results of the machine that can be endlessly generated with meaningful iterative design process.
Tabloid risograph prints that would work as takeaways at the exhibition. They feature compositional explorations of the letterforms' textures and with the letterforms typeset into different meaningful quotes.
This is a comparison of the scale between the tabloid prints and a 10 ft long screen print. The tabloid prints can be tiled all over a wall, while the screen print would run from the ceiling to the ground. 
A larger view of a digital mockup of the final large scale screen print. The reason why this poster would be screen printed is to further the narrative of contrasting between machine and human. Screen printing is a more intimate, hands on approach to printing which creates unique visual outputs when compared to digital printing. You can see the errors of the machine in the type, and you can see the humanistic qualities of printing the poster by hand.
Back to Top