All Blog Posts

Reflections on Using Generative AI for Design in June 2024

June 2024

This short blog is about my thoughts on generative AI as a tool for designers in June 2024. With AI evolving so quickly, I wanted to jot down some ideas to have something to reflect on in the future.

Once again, I found myself working on a personal project that quickly grew into something bigger. This time, I completed the project faster than expected and learned a lot along the way. Thanks to GenAI solutions.

What the project was

I've been publishing font designs for years, and I wanted to create a single website where I could showcase and share all my work. I set a few goals for the project:

Nothing too ambitious, but as a side project, I knew it would require a lot of work and time, especially for development tasks.

What I got from AI

I started with small technical proof-of-concepts (POCs) using ChatGPT and Copilot. I created the needed functions and features piece by piece.

My first prompt for the service was: “Can you create a website with a text field where users can test different typefaces? The page should have one centred text field that uses a custom font defined in the CSS, along with a slider to adjust the font size.”

“Can you create a website with a text field where users can test different typefaces? The page should have one centred text field that uses a custom font defined in the CSS, along with a slider to adjust the font size.”

This provided exactly what I needed. Piece by piece, with some iterations and manual corrections, I managed to get a fully functional website.

This was fantastic and exactly where I hoped AI could help the most. Since I'm not a developer, I knew the biggest challenge for this project would be dealing with technical problems. With AI's assistance, I was able to overcome these obstacles more easily.

Basic foundation by AI, Final Touch by Human

It's no surprise that AI was a superb assistant for content creation for the website. It was especially efficient for content-heavy parts like the Privacy Policy. The AI provided the basic structure and content, which I then finalized with a human touch.

Extra Jack of All Trades

During the project, GenAI turned out to be an invaluable jack of all trades, assisting me in various tasks. Here are some examples where GenAI provided useful help:

What were my feelings and thoughts after the project

Why Is This Difficult and This Easy?

It was wonderful to see how quickly I could create functions for the service and get alternative approaches whenever I wanted. The surprising thing was the quality of work GenAI tools provided—or the inconsistency of that quality. I might have five different working functions for complex features, yet the GenAI would get stuck when I asked it to draw an arrow using CSS. It was confusing and time-consuming, but in the end, GenAI was an awesome assistant.

Should I Also Test This Approach, and This One…?

I found myself doing tasks that would typically be categorized as high effort and low value. GenAI proved its skills in UI-level development, making many laborious UI fine-tuning tasks trivial. When iteration becomes easier, the temptation to test out different approaches increases.

Should I Change How I Value My Work?

Yes and no. GenAI makes things more efficient and offers numerous possibilities, but it can't handle broader concepts or understand context and relationships as well as a human can. Design and development bring value. The key for value lies in the methods of working. GenAI is an excellent tool, and a valued professional should know how to use these tools correctly.

The Thing I Loved the Most

I'm certain I have fewer personal projects left unfinished. Additionally, I'm freer to take on projects that involve technologies unfamiliar to me.

Looking Forward to the Future

AI development moves so quickly that there's always something new being published. While writing this post, all the solutions I used in the project received updates. Let's see if there's a need for a follow-up post in the future to check how things have changed.

For this post ChatGPT is used to proofread the test. Bing Create is used to create inspiration and reference images for the header illustration and Recraft to create vector elements for the illustration.