Like almost everyone, I’ve been playing with artificial intelligence to see what it’s all about. My focus has been to see how it might impact teaching and learning and the processes I use as a designer. As you probably know, design is grounded by a human-centered engagement process, so I was interested in the human-artificial dynamic and how that would play out and impact not only the processes that I use but the outcomes that result.

To provide some context, I do have a ChatGPT-4 account that enables me to use plugins, which has been interesting. I have Canva, Smart Slides, Diagr.am, and ChatWithPDF installed. And, like any tool that is evolving, new things appear - today, I have the capacity to create imagery through DALL E 3 directly in the ChatGPT-4 interface (see the embedded image directly below; the prompts were about student collaboration in high schools). I also have ChatDoc and Grammarly accounts and various other resources with AI assistants and capabilities built into them (Canva, Rev). Finally, I can access Browse with Bing directly in the ChatGPT-4 interface, although I have yet to figure that out.

Image creation through DALL-E 3 plugin for ChatGPT-4. Note the prompt, which automatically triggers the plugin to respond to the prompt. I asked for six images, it gave me four.

As I mentioned, I was quite interested in seeing how AI could impact my design work. Here are some initial thoughts on the intersection of AI and Design:

Data analysis: During Discovery, I collect a lot of information in various formats. I have learned that ChatGPT-4 is very good at processing large amounts of data, and it saves me enormous amounts of processing time. For example, I inputted definitions of collaboration from a stakeholder's meeting and asked ChatGPT-4 to develop a set of norms based on their definitions. It was remarkably good. In another instance, I used ChatGPT-4 to develop a method card deck on organization growth and development, where I inputted all of my general observation notes from the engagement and asked it to develop a deck of method cards composed of 6 different distinct types of cards, that could be played to develop strategic conversations about the organization. And in about 20 seconds, it created the card deck organization and the card types, which were both terrific. I then asked it to develop ten games that could be played using that deck, and it provided a great start to the game types. The ability of this tool to process large amounts of data and to provide insights, patterns, and trends has been truly useful and somewhat surprising. It is also possible to process information through a particular lens (administrator, community member, HR Director, etc.) to develop multiple perspectives on the same data set. Impressive. And again, very useful.
It’s a start. As you know, ChatGPT-4 responds to your prompts and gives you a response. In my case, I’ve used that to jumpstart my thinking and as a starting point for a design response. I generally add project-specific nuances and my language (it can be wordy) to the document ChatGPT-4 has created, should I choose to use it. I have found that using this tool in my work is best described as a partnership and a negotiation. As you get a response, see what it is and offer an additional prompt to continue the dialog as needed to improve the response.
Prompts are important. I often begin by having it assume a role (such as teacher, blog writer, game designer), and it has been interesting to see the difference when I don’t do that. It's significant.
It sees things that I don’t. I’ve been impressed with the outputs of prompts and ChatGPT-4’s ability to return really creative responses. I have found myself saying: “Wow, that’s a pretty interesting insight, I wouldn’t have thought of that…” Or, it can return really interesting language in how it frames its response that I can further develop. To get an idea of this, check out bullet points 1-8 in ChatGPT-4’s version of this post.
As a research assistant. I have a ChatDoc account that lets me upload PDFs (think learning spaces research PDFs) into a folder. I can then query the documents collectively to ask for major findings from all, the most significant impacts, etc. It will process your prompt using the PDFs as a data source and then return a response with hyperlinked references to which document(s) were used in the response or part of the response. It is amazingly helpful, but you still have to read the research.
I use my Rev account to record and transcribe Zoom meetings. After the transcript is processed, I can use the onboard AI tool to question the transcript. That’s a different way to look at an interaction, but the AI tools in ChatDocs and Rev are the interfaces that enable me to have a conversation with…documents.
It's too easy sometimes. Yeah, it can write a pretty good blog post. The temptation is there just to pump out the content and distribute it. It does write good copy for newsletters, websites, and other documents. And, using Plugins, you can do some interesting things like ask it to build slidedecks (see below). Again, this starting point saves me a lot of time.
You can get too dependent. With that said, the temptation is to rely too much on AI. I’ve found myself starting there when I probably should engage in my own analysis first. Or should I? I’m still developing the workflow, and I think this might be the most important thing to consider among all things to consider, especially with students and education.
It is not perfect. No, but it's very good. And like all technologies, it will improve and have a larger impact.

Powerpoint file created through the use of the SmartSlides plugin for ChatGPT-4. Prompt: Develop a slidedeck with 5 slides on student collaboration where 1/2 of each slide has an image that features student collaboration. It’s not bad for 30 seconds, and its ready for modification.

Mostly, I’m looking for interesting connections, phrases, insights, and ideas- things that extend my imagination. And ChatGPT-4 is very good at that if you allow it to become your partner. The data analysis part, as it relates to the design process, is profound and a massive time saver that enables me to process and see data in ways never before possible. Using AI to jumpstart my work is also critical. It would have taken me a long time to develop those game cards, and ChatGPT-4 did it in 25 seconds. That’s something you can’t ignore.

With regards to design, the role that humans play in understanding a situation with deep empathy, asking the right questions in the moment, seeing the visceral visual responses, and seeing all the human interactions required to do design distinctly defines the human role in the process. Use AI to help advance your craft, not to replace it, and to be that really smart colleague that you’ve always wanted.

NOTE: you can read the ChatGPT-4 version of this post here. I asked it to improve what I wrote. See the post on this topic written by ChatGPT-4.