A quick-start guide to frolicking down the generative AI design path
It’s taken me a few months of research, experimenting, and existential crisis recovery — but I’m finally ready to share my thoughts on how generative AI can be used to get work done faster, smarter, inspire us creatively, and deliver better end results — and what we need to be thinking about as we design new gen AI powered user experiences. Ta-da and enjoy!
1. What is generative AI?
2. How is gen AI going to change the world?
3. New design considerations for generative AI
4. Getting started with generative AI design
5. Favorite generative AI learning resources
1. What is generative AI?
Despite the mind-boggling science behind generative AI, what it can do to improve user experiences isn’t overly complex at its core — and that’s what this little article focuses on — how generative AI can be used to get work done faster, smarter, inspire us creatively, and deliver better end results.
Here are my two most important concepts to understand before you start your journey using gen AI to improve your designs — as well as designing top notch generative AI use cases and UX:
- Generative AI creates novel, new content based on huge amounts of data.
Going far beyond the scale of data sets we’ve used to train AI models up until today, this technology looks at all of the data publicly available on the internet and attempts to generate more of it — in new, novel ways that emulate humans. This includes the next numbers in a sequence, the next words in a sentence, the color of the next pixels in an image, even the next note in a song or movement in a video. - Generative AI allows users to navigate and interact with digital interfaces through natural language.
Unlike search engines or chat bots, generative AI can play the role of a persona the user wants to collaborate with, then proceed to converse through this tone of voice and subject matter insight. This new conversational and collaborative interaction model will elevate and enhance search and chatbot experiences, as well as being embedded into UI features, plug-ins, extensions, navigation — even operating systems.
There are two big reasons the world is going bonkers over this new technology:
- Generative AI models are unique in that they process more data than we’ve ever used before. Virtually ALL of the public data available on the internet. The result is, they can give you answers about almost anything. This is allowing the world to leap over the costs and challenges of scaling AI and letting us apply one model to any use case with minimal customization, resources, or maintenance.
- Generative AI isn’t just combing through data and spitting out an analysis. By looking at such vast amounts of data and applying style transfer models alongside large language models, it’s really learning to emulate what humans have created and reverberating back to us altogether new organization and stylization of our own ideas and content.
Now users can give the AI detailed instructions, such as…
- Take my photo and turn it into a Picasso style painting
- Summarize all this week’s news articles about generative AI in 500 words or less in a way a five-year-old would understand
- Create intro music for my podcast that’s soothing, classical, and include violins and cellos
- Read a problem I’ve described and recommend a solution as if you were psychologist that specializes in childhood development
…and get infinite versions of wholly new, historically and algorithmically inspired, stylized, creative content.
2. How is gen AI going to change the world?
Oye vei — great question and the one we’re all going to be asking for a long time to come. The full impact is as unforeseeable as the impact of the first home computers — but the enormity of the change will likely be along the same scale, if not grander.
The best way to start understanding and imagining what’s coming is to take a look at some of the most groundbreaking gen AI demos from 2023.
Buckle up folks — this is going to be a wild ride:
3. New design considerations for generative AI
The world is about to go on a journey similar to what we experienced back in the late 90’s with the arrival of the internet. Back then we were furiously researching and experimenting with how to make digital tools work for users.
Generative AI is going to take what we’ve learned about user experiences to heights we couldn’t have imagined before — shifting from tabs and search bars to conversational navigation and inspiring creative collaboration.
Over the next five years, every product’s end-to-end user experience will be revisited through the lens of this new technology.
Gen AI design consideration 1: Conversational collaboration
Inspire and improve, not copy and paste.
Generative AI has the potential to be a full reset on how users interact with digital products. We’re going to see a shift from transactional interactions (computer, add these numbers) to truly collaborative interactions (computer, what are some ways I could solve this problem). But the results users get out of a generative AI model are only as good as the prompts they give it.
How good are you at writing a very specific description for exactly what you’re looking for? Let’s put it into a scenario we’re all familiar with:
Imagine a client comes to you and says, “I want a design that does this and this.” You go off and do it. When you show it to them, they say, “Now that I see it, what I actually want is this and this.” You roll your eyes and think, “Why didn’t you say that in the first place?!?”
Describing to a generative AI model what you want poses the same challenges. Designing good prompts is not enough. If you are not an SME (subject matter expert) with the content the tool produces, you’ll need to find an SME. You’ll need to point it to the resources and data you want it to refer to. You’ll need to write and re-write specific and extensive prompts to get good results. And you’ll need to review and critically evaluate everything it creates.
Using generative AI tools will require users to become editors, needing to fact check before publishing. Mastering these skills is a challenging enough task that you’ll find job postings for Prompt Engineers whose entire job is to use generative AI tools.
Every day we’re seeing creative new UX approaches to these challenges — from providing pre-designed prompts to including personalities and philosophies for the AI to embody in its responses.
And then there are the tricks of the trade, like prompting a gen AI tool itself to act as a prompt engineer and write a list of prompts to solve a problem.
Designers and design minded product teams will be working hard for the next few years on making generative AI that works and is inclusive for all of us. Extensive user research and design exploration will be required in the coming years to make interaction patterns that are truly usable and accessible. Otherwise, we’ll all spend more time trying to formulate the right questions than it would take us to do the work by hand.
Gen AI design consideration 2: Explainable trust
How do we know what’s real or true anymore?
If you ask a generative AI tool a question, you need to know if the answers it’s giving you are accurate. How did it come up with that response? What data did it analyze? How trustworthy is that data for the problem you’re trying to solve?
Today this level of explainability is nearly opaque in most generative AI results. This issue is elevating and expanding concerns around AI — fake news, deep fakes, inaccurate and unethical gen AI results informing human decisions. We’re going to experience all of this and more in the next few years.
There are other questions arising that we’ll need to figure out, like when is it important to know that content was created by AI and to what extent? Do we need to watermark everything it creates before we post it? Should we require AI generated content to be customized by humans to a certain extent before publishing it?
This is only the tip of the iceberg in terms of the new challenges and future solutions we’ll need to be solving for if we want AI to evolve into something that’s trustworthy enough to be valuable for humans.
4. Getting started with generative AI design
It’s not AI that will snatch your job, but the individual leveraging AI to outpace your performance.
For me, the best way to learn is by doing — and then sharing what I’ve learned with others. Here’s what I recommend to build your own understanding of gen AI, make it part of your daily workflow, and incorporate gen AI use cases into the products you design:
- Dig in and start experimenting with gen AI tools — become a power user.
- Make note of what is and isn’t working for you, and capture + sketch your ideas for improving tools and interactions.
- Once you’ve formed a perspective, double down on your confidence by teaching others — whether that’s through writing, speaking, or just chatting with others.
Remember my liebchens, diversity is the most important aspect of creating solutions with AI. We need as many different minds, cultures, abilities, and life experiences as possible contributing to and shaping the future of this technology so it doesn’t become a reflection of our fears, but the positive visions we dream of.
Gen AI tools to start experimenting with
5. Favorite generative AI learning resources
How generative AI will impact the world
How generative AI will impact design
Expert opinions
Generative AI in action
No + low code technical explanations
