🧐 Summary
1️. In recent times, pivotal research topics in human-AI collaboration have centered around ‘Human-AI Teamwork’ and ‘Automated Tools to Enhance Human Abilities’.
2️. AI can streamline every step of the design process, including planning, drafting designs, automating design systems, and performing usability analysis, even in the absence of data.
3️. The key to successful collaboration with AI lies in perceiving it not merely as a tool, but as an integral member of our team!
This year, numerous big-tech firms in the United States have downsized, affecting many UX designers and researchers. Though this may appear as a mere repercussion of economic challenges, various design publications have been highlighting the renewed significance of UX designers in these times. Moreover, there is a growing discourse about AI potentially supplanting UX designers. In this context, I will delve into Human-AI Collaboration (Do you recall the AI research field diagram from our last article?👇). We will examine the nature of human-AI collaboration, ongoing academic debates, the AI tools applicable in design processes, and how UX designers can contribute to fostering effective human-AI teamwork.
1. Is Human-AI Collaboration a Reality?
Presently, AI captivates the public imagination as an intriguing service, with generative AI services gaining notable traction this year. As AI evolves, it’s expected to more seamlessly blend into our lives, augmenting human productivity and support. Many professionals are already harnessing ChatGPT in their workflows, and we too have found it invaluable in crafting our newsletters. I’ll share more insights on this later.
The concept of human-AI collaboration is intrinsically linked to the principle of human-centered AI. This paradigm views AI not as a replacement for humans, but as a co-existing partner that bolsters human capabilities.
In Capel & Brereton (2023)[1], they suggest that evaluating human-AI collaboration hinges on two critical metrics: performance and satisfaction. Performance encompasses the combined effectiveness of both AI and humans, while satisfaction pertains to the gratification derived from AI partnership. Additionally, Pflanzer, M. et al., (2023) [2] address ethical considerations, emphasizing the necessity of ethical judgment in AI systems, especially in complex or human safety-related domains.
Both studies converge on a pivotal question: how can we maximize the synergies between humans and AI for effective collaboration? As UX designers, we are keen to explore the avenues for productive human-AI partnerships.
2. Are There AI Tools that Can Streamline Every Step of The Design Process?
This is the well-known Double Diamond model. The first diamond represents understanding the situation and defining the problem, while the second diamond involves finding and testing various solutions to create the final product.
Let’s map the AI services we’ll discuss onto this model.
👉 Step 1. Discover
Envision — A service that automatically summarizes and highlights user interviews conducted via Zoom.
👉 Step 2. Define
Userdoc AI: A service that automatically generates detailed User personas, Journeys, Software requirements, etc., based on inputs like service overview and target users.
💡When we assumed our Inspire X newsletter as a service and inputted information about it and our perceived users, the result was helpful! Though not very detailed, it gave me a good starting point to develop further as a PM.
👉 Step 3. Develop
Uizard — Turns hand-drawn wireframes or written descriptions into UI mockups, and converts screenshots into editable mockup versions.
💡When I requested a design for the InspireX newsletter app with a simple description, the tool provided a layout, though it mistakenly used red instead of the requested mint color. It seems suitable for inspiration in small-scale startups or personal projects that a UX designer isn’t available.
Visily — Also converts hand-drawn wireframes or screenshots into editable mockups.
💡When I uploaded a design created by Uizard, it was transformed into editable components. It’s not perfect, but useful when you have a good reference image.
👉 Step 4. Deliver
Quant-UX — Allows for quantitative analysis during usability testing with Figma prototypes.
VisualEyes — Predicts eye-tracking patterns using extensive existing eye movement data when a screen is inputted. Attention Insight is a similar tool.
Writing this newsletter, we experimented with numerous AI tools, feeling like having an intern by my side or my own design agency 😍. The major advantage was their ability to quickly and diversely handle the tedious groundwork. The downside? The ‘intern’ or ‘agency’ isn’t that smart yet 🤣
For now, it seems AI can’t replace human in the first diamond, the understanding and defining phase. In the future, UX designers must develop their ability to comprehend situations and define problems while embracing various AI tools to quickly propose and test numerous solutions. Like Doctor Strange in the movie ‘Endgame’, who viewed 1,400,605 futures to make a decision, UX designers need the ability to choose and refine the meaningful options among the vastly increased number of proposals. With more options to choose from, strategic thinking becomes essential.
3. How Can We Effectively Collaborate with AI?
What comes to mind when you think of collaboration? Amazing outcomes, challenging yet rewarding processes, and fantastic colleagues… We’ve all read such narratives. But the reality often involves barely reached conclusions and meetings that end in stubborn disagreements.
Would collaboration be any different if it were solely among AIs? An interesting study on human-AI collaboration was published in the 2021 issue of the Computers in Human Behavior journal [3]. The research team conducted a game where 46 teams worked together to escape a small island and then interviewed the participants to analyze how team member perceptions affect teamwork processes and outcomes.
There were three types of teams:
Type 1: Human + Human + Human
Type 2: Human + AI + AI
Type 3: Human + Human + AI
The results showed that Type 3 (Human + Human + AI) teams demonstrated more effective teamwork than Type 2 (Human + AI + AI).
But what makes them different?
For Type 3 (Human + Human + AI) teams:
- Recognizing AI as a Team Member: With human teammates in the mix, these teams collaborated actively, similar to the all-human Type 1 teams. They treated all members, regardless of being AI or human, with equal regard and positivity. In contrast, Type 2 (Human + AI + AI) teams had a negative perception of AI, leading to almost no communication, as they believed AI couldn’t respond adequately to human actions, resulting in a lack of cooperation and a focus on individual tasks.
- Developing Shared Mental Models (SMMs): Participants in Type 3 teams reported that AI seemed to base its decisions on other members’ choices, fostering a shared mental model. This allowed for better prediction of each other’s actions and intentions, facilitating more efficient collaboration.
Sounds familiar, right? It’s like the message in those posters hanging in every meeting room: respect colleagues, establish common goals, and collaborate actively.
But here’s the twist… all 46 teams in the experiment consisted solely of humans! There was no AI. Participants were made to believe that AI was part of their team. Ultimately, whether AI is perceived as a real team member significantly influenced the outcome of collaboration. Recognizing AI as a team member and fostering positive trust and shared values among members is essential for true collaboration.
Collaborating with AI might seem different, but the essence of ‘collaboration’ remains the same. As we progress to include AI in the design process, perceiving AI merely as a non-interactive tool limits the potential for true collaboration and progress. When designing AI services for human collaboration, it’s crucial not only to focus on AI’s performance but also to encourage perceiving AI as an interactable and integral team member.
Organizations are constantly striving for effective collaboration. Within the next five years, we might see the emergence of specific rules for collaboration with AI agents or guidelines on how to effectively partner with AI in each organization.
Did you think working with AI would be all about logical, rational outcomes that magically align with your expectations? The foundation of collaboration lies in recognizing your counterpart as a team member and moving forward through communication towards a common goal. The same applies when working with AI. Given AI’s intelligence, there might be an expectation to easily find answers, making it even more crucial to perceive AI as an organically functioning member of the team.
Even in AI collaborations, humans initiate the work, so the power of thought and decision-making becomes even more critical. Establishing your own design philosophy and criteria becomes paramount for logical decision-making. Let’s become UX designers who establish their own perspectives, with our articles!
Questions from Inspire X
Please share your thoughts via comments!
Q1. What skills should a UX designer strengthen to thrive while working with AI teammates? Perhaps a warm heart?…😘
Q2. If you have other AI tools that you find useful in your work, aside from those introduced in InspireX, please share with us! It’s always great to share good things😃
