Greg Walsh, Ph.D., the Parsons Professor of Digital Communication, Commerce, and Culture, opens a discussion challenging how designers are currently use artificial intelligence in their work. Rather than framing AI as a tool purely for efficiency, he argues that many designers need to challenge the way they are using it. He emphasizes that designers must rethink their approach, whether they are overrelying on it for answers or dismissing it entirely, he notes that we need to reimagine AI's real potential as a thought partner that helps refine ideas and deepen the design process.
There’s a big problem with the way that I see people use AI in their design work. A lot of times, people treat AI as an answering machine, meaning you ask it a question or you tell it to do something and it just gives you something back. I have seen student designers overly rely on it. I’ve seen student designers dismiss it entirely. I have seen professional designers over-rely as well as dismiss it. In reality, neither side of that spectrum is useful to understanding how AI can help in a design practice.
I use AI, and there’s some things that it’s really good at. First of all, it’s really good at helping you organize and externalize your thoughts, your ideas. It helps you get to the root of concepts and new ideas that you have but might have trouble expressing. It’s really good at generating questions that you haven’t thought to ask. I found that some of my best work has come from using Large Language Models (LLMs) to generate questions of things I haven’t thought about. One thing that I use AI for a lot is placeholder text or throwaway content such as fake names or fake data or dummy copy. It’s fantastic for generating things to just fill in the blanks when you don’t have any other content for them. Of course, a company like Grammarly would tell you that it’s great for checking and editing writing. No matter what kind of designer you are, your writing matters, and AI is great for that.
The important thing to realize when it comes to AI and design is that AI is not doing any of the creative work itself. I call AI the mediocrity engine. When you ask a large language model a question, it will produce, for the most part, good content that’s generic. It’s not going to give you any big revelations about anything. It won’t give you a good idea out of the box, and that’s just how it’s designed. Its goal is to give you the most middle answer that it can, and that’s why its output is only as good as how you interact with it.
I use AI as a team that I can’t afford to hire. I don’t have people to help me design apps for my website or generate role play scenarios for classes. AI is great for designers who are solo practitioners, or when you’re on an interdisciplinary team where only one person is doing one specific thing. The key is that it’s not about replacing designers. It’s not about replacing entry-level designers. It’s not about replacing interns. It’s about what you can do when you have no other resources available to you. The real work in using AI, especially in the design field, but really any type of knowledge work, is the skill in knowing how to ask for what you want.
I recently used AI to figure out how I can market our M.S. in Interaction Design and
Information Architecture program. Here is an example of my conversation:
Me: I have a concept for how I can market our Interaction Design and Information Architecture
program: I want to consolidate all of our design programs into one marketing message
so we can talk about graduate Design programs at UBalt instead of this program or
that program. What are some questions you can ask me to hone this idea? What are some strengths of this idea? What are some weaknesses?
What I really do is present an idea such as I want to show information in this way, and then I say to it: What are some questions you can ask me to hone this idea? What are some strengths of this idea? What are some weaknesses? What it did for me was provide seven different questions that are actually really solid questions for me to answer before I move forward. The important thing is that when it did the strengths and weaknesses, they’re only OK. I can’t say that the strengths were necessarily something I hadn’t thought of, but there were some good weaknesses in there. More importantly, it didn’t tell me if the idea was good or bad.
And the same thing goes from a design perspective. If you present an idea to AI, the best use of it is to help you elaborate on it. It’s to help you think through what can make this better. What are the things you haven’t thought about yet? Based on the idea prompt, the LLM is going to run through different types of questions, and those questions help you as the designer continue to solve the problem.
It would be great if I had a team of people sitting around to bounce these ideas off at any time of the day. AI is there any time of the day. If it’s 1 a.m. and I want to work through an idea, it’s a great partner. And it’s important to realize that the questions that I got back were what was valuable to me in rethinking my process. The answers I got, like I said, were just kind of generic.
I really want you to try this. I would love for you to take one concept you’re working on right now and put it into an LLM. Ask it to generate questions you haven’t thought of yet. Look at those questions and answer those questions. And see if where you are after having done that puts you in a better place than you were before you started.
Learn more about Center for Digital Communication, Commerce and Culture
