Generative processes – An interview with Daniel Wenzel

Marius Hansen Hogaas & Thondiv Ela Calip

Teknologisk bilde

‘Machine learning graffiti New York’, Daniel Wenzel (Copyright © Daniel Wenzel, 2020).

In today’s society, we see how Artificial Intelligence has in many ways improved the lives of individuals who otherwise may not be able to complete tasks without specialised equipment. To see how this technology affects the field of design, we decided to interview Daniel Wenzel, senior art director and procedure designer at DIA. He works with typography and generative processes, and besides he is also an artist and lecturer. He was kind enough to agree to answer a few questions about his experience with Artificial Intelligence and how he incorporates this technology in his design processes.

For how long have been working as a designer and started studying?

My cousin introduced me to photoshop when I was 8 years old. I started studying Communication Design at the University of Technology, Business and Design Konstanz in 2015. Halfway through university I paused my studies to work at DIA Studio in New York. To this day, I never left the studio, but I did return to Germany to continue and complete my studies in 2019.

How does your everyday look like?

I am working from home. Apart of the usual client work, one thing that stands out is that we do a lot of exploration and testing because it is important for us to build an archive that we can harvest and expand on when needed.

How many people is it at DIA?

DIA is fairly small, with only four people at its core – the two founders, Mitch Paone and Meg Donohoe, as well as Deanna Sperrazza and myself. If necessary, we collaborate with specialists to expand and complement our work.

Teknologisk bilde

Copyright © Daniel Wenzel

Why does design matter?

Everything is design. How it matters? It depends on the context.

From a capitalistic standpoint it is simple. There is measurable value in design. There are studies on brand value like when you ask a target group to put a price tag on a Mercedes, they estimate the value much higher than the same car without any brand recognition. If you strip away brand values from Apple for instance, would you still buy the same phone for the same price? But there is also no one solution. Imagine having a kebab shop that looks like an Apple Store – I wouldn’t go there. I would assume the other place next door is better and cheaper.

Design matters in many ways. It really depends on the project. If I am making a campaign for a phone, it is what it is. You are designing to sell. But you can also design for accessibility, for medicine packaging that needs to be read clearly or quickly understood so people don’t take the wrong drugs. Or UX design – why is it easier for my grandparents to use some phone brands over others.

How did you end up becoming a designer?

I was always interested in arts but I realized from a young age that I don’t like to be an artist for living. I liked problem-solving so graphic design spoke to me in that sense. Your work can be functional and aesthetic. Being an artist often means that your work is only for the ones who can afford it – for the rich. With graphic design, you can reach a broader audience – the everyday person.

Who would you say are your main inspirations and role models?

I try to avoid being inspired by other design. I don’t think it helps to spiral around within the design bubble. Design inspired by design is kind of silly in my opinion. You could train AI on the accounts of graphic design influencers to create design posts inspired by other graphic design but how would that be of any value? This is essentially what many graphic designers nowadays are doing on social media or maybe what social media is doing with them. This is not how we move forward.

Growing up, Wenzel got inspired by multidisciplinary artists like Leonardo Da Vinci.

I think what inspired me there the most, was the fact that he was a polymath – an architect, an engineer, a designer, as well as an artist. The fact that you can work in multiple fields that may then benefit from each other or evolve as they overlap.

Daniel thinks designers need to reach outside their niche bubble to gain from when developing new ideas.

Being too deep in your field is why every car looks the same. If we look into type design for example, the type designers I admire the most are non-traditional type designers. I do think that those are the ones that push the boundaries of type design. I do a lot of motion

design in my work, but I don’t consider myself as a motion designer. I never learnt motion design and maybe that’s good. I hope that as a non-traditional motion designer, I can bring a different perspective to the field. This mindset has paved the success of our studio long before me – Mitch, one of the DIA founders, has a motion design background but started working in the field of typography and branding, which was a very new approach before it became the standard today.

Teknologisk bilde

36 Days of Type (‘R’), Daniel Wenzel (Copyright © Daniel Wenzel, 2020).

What is your relation to AI?

AI can help to visualize a broader idea and to clear up your own thoughts in a way. If I simply don’t want to execute it manually, then using AI is still me feeding my idea. Right now, I don’t really see AI to work on its own. That is when things get boring. AI is not yet capable to create ideas like humans who can get inspired by music to create architecture or by pop culture to do art. But I see a lot of value in utilising it as a tool that assists to execute my ideas. Assistive AI instead of autonomous.

Why do you think AI is important to look at as a tool rather than a replacement for human labour?

There is a distinction to be made in AI between artificial narrow intelligence (weak AI) which can fulfil one specific task, and artificial general intelligence (strong AI) which is able to fulfil any intellectual task of a human being. All AI we have so far is weak AI. Strong AI does not yet exist. Therefore, we still have the advantage in inspiration and creativity. When I started experimenting with AI during my thesis in 2019, the possibilities were a lot more limited than today. I mostly got inspired by its mistakes and finding different letterforms within the training process. Let’s say that AI is trained on images of the letter ‘A’. The AI keeps training and improving failure rates. The moment it becomes “too good” at making A’s, I’m struggling to get any value from those results. But during the training it looks more like a child was to draw it. I find these much more interesting than the adult trying to be realistic and accurate. Choosing this imperfect stage and deciding that this is interesting is where the human comes into play. This is what I mean when I say I was ‘art directing’ the AI. I decided that this was interesting while the AI was still considering it as a failure.

Teknologisk bilde

Pixels and Tiles, Daniel Wenzel (Copyright © Daniel Wenzel, 2020—2021).

If you look at AI in 10 to 15 years, how would you use AI then? Because that type of technology would probably have evolved in some way then.

A path to strong AI might lead to super-intelligent AI and then we have robots with emotions living among us in society... who knows? haha. I have no idea what the future will bring. Things evolve with enormous speed. What I do know is that weak AI already can be extremely powerful. Right now, most designers who work with AI combine 10 different models to make one project. Stable Diffusion by itself doesn’t come up with the funny, weird ideas that people use it for. That’s still human creativity. It doesn’t have to involve thousands of hours of work for an idea to be good. A good idea can be simple and fast. I think if AI is there to help to execute those ideas, I’m all for it.

But do you think we would use it as a tool if it got stronger?

I believe we will continue to use AI more and more while it gets democratised. In 2019 when I did my thesis, it was very difficult for me as a designer to work with AI. I ran all the stuff on my machine for nights in a row. I even have friends that burned their computers from overusing them. It needed a lot more programming and research than today to even get started. Now there is thousands of YouTube tutorials and people are building interfaces or products like Adobe. They are also already very much integrated in our lives. If you look at Google phones, you can now mark anything in your pictures that you want to be removed… and that’s AI. With that said, I don’t think we should be afraid. Photoshop for instance, is a powerful tool without AI already. People were afraid of that when it launched. It revolutionized the field, but it didn’t get rid it. And even today, you still have people working with analogue photography. I think both can exist side by side.

Is your process based on many different programs? How do you combine different methods in your work?

I use anything and everything. I always find it interesting if people are only working in one program and think they are going to solve all problems there. I just try to make my life easy. If I can start in After Effects and move it to Cinema4D, then to Illustrator, to code, and back to After Effects again… and I know it would be the easiest way to achieve my goal that way I would do it. I know that different programs got different advantages. My process to break it down is the road of trying not to do more than what I need to. There is no right or wrong tools in design. Most of the tools I use, I use “wrong” or not as they were intended.

Do you think you would arrive to these ideas without assistive technology?

I mean, a lot of times I do. I don’t use AI too much. I would like to use it more, though. On an everyday basis, I have to say AI is not streamlined enough for me to use in fast-paced projects. If it needs a quick turn-around, AI is currently still too time-consuming, too much troubleshooting, too much trial and error. If I have a personal project or a something with no time pressure, then maybe.

Teknologisk bilde

Automatisierte Schriftgestaltung / Automated Type Design, Daniel Wenzel (Copyright © Daniel Wenzel, 2019).

Your Automated Type Design project shows a detailed and boundary-pushing type design research where you combine traditional and generated processes. You mentioned in your website that you made ‘self-created or misused programs to simulate processes which are not possible with conventional type design tools’. Can you tell us more about the process around this project and how you made these programs?

Think of it like this—You would like to use a brush to draw your letters in Glyphs. Unfortunately, that’s not possible without a plugin, however in Illustrator it is.

Teknologisk bilde

Automatisierte Schriftgestaltung / Automated Type Design, Daniel Wenzel (Copyright © Daniel Wenzel, 2019).

Illustrator is not a program intended or used for type design traditionally and it’s not convenient enough for that, but it can simulate a brush. Therefore, I can draw a skeleton in Glyphs, then bring it to illustrator to simulate a brush and then even bring it in After Effects to interpolate or deform it. That’s what I mean by ‘misusing programs’. I also wrote certain scripts for example to generate a data set for training the AI. For this, I needed a lot of pictures of A’s. I ended up writing a script for Illustrator that places a text box with the letter A in each typeface on my computer, and then exports each layer in the document as a PNG. It’s a script that utilises what Illustrator can already do and ‘misuses’ it to create a dataset. It’s all about making the most out of your programs.