Aesthetic Imperfections: In Conversation With Gianpaolo Tucci

Can you share your journey into the intersection of AI, typography, and design? What sparked your interest in these fields?

As AI image generators have become available to the mass market, I was not interested in creating a new cyberpunk image, instead I wanted to test how results might relate to reality.

So a few years ago I worked on a movie title project for Warner Bros Italy directed by Andrea. An handmade exercise that was about kinetic typography, digitally captured and adapted. With AI I wanted to evaluate what I could have done, and see how much the results would have been similar to what I wanted to achieve.

This brought me to test AI for typography, where a new visual value and meaning is increasingly a key player in delivering stronger visual meaning — especially in the future when it will also be applied to VR/AR.

In your exploration of AI and typography, have you found that the element of randomness introduced by AI algorithms brings a unique aesthetic or philosophical dimension to design? Can you share any reflections on the acceptance of this randomness in the design community?

Randomness is a constant in algorithms. The latent space, this complex system of data (the AI brain and anti-consciousness), is where, based on an input, the AI connects the dots between contents of “her knowledge” to create, randomly, an output. In this, I’ve discovered the beauty of randomness in AI, while exploring these Aesthetic Imperfections, resembling metaphorical nature.

How to influence randomness? Defining a ludic interpretation-game between human and machine. By balancing randomness and intuition, the two faces of the same coin, creators can amplify the horizon of possibilities.

The design community does not seem to embrace AI randomness as a value, however we might say our life is based on a series of random events… AI Image generators are often perceived as “a press button to get” experience, for me, as in everything, time is always equal to value. 

In this context, spending time to refine and evolve possible new directions is a core value towards the discovery of new possible creative directions — bigger than even the final output.

You've touched on the theme of co-creation between humans and machines. Could you elaborate on how this collaborative approach is reshaping the design process, and how it might lead to unforeseen innovations in typography?

Typography is a technology, the technology of communication. The dialogue between human and machine is shaping this collaborative approach, so how can AI provide typography and lettering with additional value? Processes with AI are not linear, they are fluidic. This means that they are often unpredictable. When controlling this unpredictability — and being able to re-adapt the process — human creativity, knowledge, emotions and experience deliver weight. 

In terms of typography. Consider the aspects of accessibility and language variations — think about how type might dynamically change and adapt based on external parameters.

The visual enhancement of AI upon typography is pretty clear, continuously evolving visual messages are the perfect mix between typo and lettering, where the cognitive reading is not anymore based on the functional rules of typography but for the visual association to its meaning.

There's a notion that imperfections contribute to the aesthetics in design. How does this notion resonate with the unpredictabilities and 'glitches' inherent in AI-generated typography?

In Japan, Wabi-Sabi is the antithesis of the classical Western aesthetic of beauty equated with perfection and symmetry. It is a mindset that helps us appreciate the beauty of imperfection: exploring Wabi-Sabi and embracing flaws in art and in design.

My exploration of flaws also led me to analyse AI glitches in some of its most iconically flawed depictions such as “Human Hands” — creating provocative “handmade” typography as a result.

How do you see the role of interpretation in the dialogue between designers and AI tools? Does this interaction foster a deeper understanding and exploration of design principles?

Design principles serve as a guiding framework, embedding values into design outcomes. With the rise of AI and the co-creation era, there's a clear demand for these design principles.

These principles can change according to company mission and values or designers' PVPs. I have envisioned 9 Principles for the co-creation era in my book, looking at the AI's limitations and instead focussed them towards new opportunities.

With your extensive exploration in this domain, where do you envision the future of typography headed, especially with the continuous advancements in AI? What are some potential challenges and exciting opportunities you foresee?

We know of a few tentative GAN outputs based on typography trained models, and they are not really appealing, visually pleasing or typographically correct. It's intriguing to me that, in AI image generators tools, until now, typography hasn't been fully integrated or adequately trained.

AI image generators tools are advancing in trying to mimic reality — they are becoming too realistic. In doing so, they lose what to me was great about them in the first place: being a visual dreamer.

Fortunately, AI can't yet visually write (type) properly what the user wants, and I think that's great. I would like to reinforce this by training a model with all my typo letters and see what will come back — as calligraphy of my AI — and adding then something more.

In terms of nature, there is almost a question of acceptance. Is this what you are looking for also with regards to AI?

AI has been accepted since ages, and has always been part of algorithms of products we use daily. We are now understanding how powerful AI prediction and interpretation is. 

I don't feel it is a matter of acceptance but a matter of defining proper scenarios for a clear support over human activities more than, mistakenly talking about the fear of being replaced, without instead seeing the potential of certain automated task accomplishment by AI.

AI should be seen as a tool: an amplifier and not a replacement.

Is AI natural? It is a product of human beings — we are natural — does that mean it is also natural?

AI clearly is not natural, but is a technology that resembles what we have inside of us. 

Neural networks, connections of a complex system, learning to deliver "knowledge", is a technology that becomes everyday more biological.

While AI is artificial, it synergises with nature. The unique outcomes, their inherent imperfections, and the overall evolutionary approach of AI are not just engineered processes. Instead, they resemble the 'bricolage' activity described in 'Evolution and Tinkering' (1977) by Francois Jacob.

What is the purpose of your book? What do you want the reader to take away from it?

The book is a testament of this moment we are living as the beginning of the co-creation era. Ultimately, it’s a witness to an investigation across Impact, Balance and Time over the utilisation of AI as a co-creator. In this, it shows the need for the reimagining of Principles, Rules and Utilisation of this tool towards the definition of new possible future scenarios and Aesthetics Imperfections. 

Aesthetic Imperfections will be published by Slanted in November.