Design processes, augmented with AI
Design processes, augmented with AI
master’s thesis
project processes, AI, service design, co-creation | January - June 2023
The task:
In 2023, AI can be easily considered a “hot” topic. AI tools and apps are becoming widely available to the general public, allowing for an unseen level of creativity and content creation. But what do these developments mean for designers and user researchers? How can they benefit from it? How can this technology help them with their every work - enhance, expedite it, or just make it simpler? This is what I strived to discover in my master's thesis, titled "AI as a partner of service designers: reimagining design processes through novel cooperation with its underlying ethical considerations".
My approach:
With an analytical mind, I always start each task with the research phase. For this project, I used design thinking and a meta-synthesis approach. Both allowed for a structured approach to the process. The process started with a wide-range topic analysis through desk research and user-centred research. I involved design practitioners of various proficiency levels to discover how they work, use AI, and where the potential for using this technology in their work can be. By conducting explorative interviews, I learned a lot about designers' points of view on AI, where they would use it, as well as the ethical angle on this whole topic. Based on insights from said research, I started developing a toolkit for augmenting design processes with AI.
Image generated by AI.
RESEARCH QUESTION
solution:
In the end, I decided to develop the AI-augmented Desing Process Model - an interactive toolkit for designers consisting of use cases that have been created as the answer to the research question. Designers and researchers can use it to find different applications for AI in their work. See the framework below:
Welcome screen
The first screen is a welcome board, used for greeting users and presenting essential information about the toolkit. Besides the message and CTA button, in the top right corner, the user can find the navigation menu. This menu is present on every sub-page of the toolkit and contains four shortcut buttons - go back to the previous page, go to the home page (Double Diamond model), go to the instructions, and go to personal space.
Instructions
Next, the user is presented with the instructions set to ensure they know how to properly and responsibly use the toolkit.
Homepage
After clicking the button on the previous page, the user is taken to the toolkit’s homepage - the interactive AI-augmented Design Process. Here, the user can hover over different markers to get a quick glimpse of what each use case is for. Clicking the Learn More button will take the user to the specific use case page, where more information can be found and AI tools can be accessed. Moreover, two different use case filtering options have been developed for Toolkit 2.0 - Library and AI search. There will be described in detail later in this section.
Use case page
Users can learn more about the specific AI use example on this page. Each use case includes information about AI applications, tools that you can use to complete the task, professional capabilities that AI will augment you with, benefits from using AI in said situation, examples of AI use in academia and practice, example prompts to help you start working with the tool, as well as things to consider before using the tool. Additionally, AI apps included in the use case can be accessed directly from this page.
Library
The new addition to version 2.0 of the toolkit - the Library - aims to make it easier to group various AI applications by topic. In the future, the toolkit is meant to contain dozens of them, and having them all mapped on the AI-augmented Design Process model might confuse new users. Therefore, Library organizes them into categories - examples include analysis, generating ideas, and developing designs.
AI search
The second new addition to version 2.0 of the toolkit - the AI search - makes it easier for users to search for the right tool. In the future, the toolkit can include dozens, if not hundreds different tools, and browsing them one by one could be time- consuming. Therefore, with AI's help, users can tell the app what they need to do and the AI will showcase the most suitable tools for the tasks. The user can learn more about the specific tool by clicking on the Learn More button.
App info page
During the evaluation, test participants expressed concerns about both the specific AI apps and use instructions. This page is meant to remedy these issues. The user can read detailed information about the app, see general use instructions as well as is presented with a use warning - what to be mindful about when using it or feeding data to it.
Personal space page
The last significant addition to the 2.0 version is the so-called Personal space. Here, the user can take a moment and reflect on the use of the toolkit, as well as the work they have been doing in connection to AI. First of all, users can send feedback to the toolkit’s developers. Moreover, the user is able to express their feelings, problems, and thoughts about working with artificial designers. Additionally, it is possible to save a draft of each message. The page is accessible from the navigation menu in the top right corner of each page.
FUTURE WORKS
Even though this iteration of the toolkit is much improved compared to its first version, there is more room for further improvements and innovations:
• App verification procedure - it must be acknowledged that not all AI apps are trustworthy - or that their developers have good intentions. While testing the first version of the toolkit, one participant expressed their concern about the Synthetic Users app - app that allows replacing real user research with AI-generated ones, cutting down on time spent on user research and generally providing the money-saving benefit to any design team using it. Other apps might collect sensitive user data and sell it to third parties. Therefore, there would be a need for a mandatory verification procedure for each app before it is featured in the toolkit. What is more, specific evaluation criteria would need to be developed. Possible examples could include data handling, impact on users, and adherence to ethical guidelines.
• More apps added to the toolkit - with time, more AI applications will be developed and their uses will expand to new territories. It is crucial to keep the toolkit up-to-date for designers to fruitfully harness the full potential of AI. New developments, or even breakthroughs, will be happening regularly. This will surely revolutionize the way designers work and think. With the incorporation of new tools and apps in the toolkit, more complex problems will find solutions.
• Implementation - currently, the toolkit 2.0 exists as a click-dummy. In the future, there would be a need to program and develop a fully functional, possibly web-based app to harness its full potential. Moreover, additional user research would have to be conducted to ensure the interface is actually user-friendly, simple to use, and error-free. Furthermore, a more robust user guide would have to be developed - especially with the expanding functionality of the toolkit.
• New use cases and user-suggested content - as AI evolves and advances into new territories, novel opportunities for its utilization will be presented to the design community. In consequence, there will be a great chance to develop a vast array of novel use cases (even ones that cannot be thought of today). Here, the design community should be able to help develop - or propose - said use cases, as they have the best understanding of various user needs, (service) design requirements, and general wants. All in all, within the toolkit app, there should be an option to propose new use cases, tools, or other features users might need.
• Expanded information categories within use cases - even though evaluation participants were pleased with the amount and variety of information each use case contains, there was room for improvement. These could include questions and prompts to ask AI about, various examples of using AI for said tasks, instructional videos, and more. Furthermore, detailed variants of specific use would be created. Some interviewees complained that certain use cases were too general, resulting in problems with their effective use.
• More real-life examples of AI use - given the variety of presented information in the toolkit and the premise that it can help designers in their work, there is a need to make it more grounded. This can be achieved through, for instance, presenting specific examples of the use of AI in real-life scenarios. Two categories can be distinguished here - real-life projects and academic, peer-reviewed papers. One example of said use is presented in the toolkit 2.0. However, it is acknowledged that there is a need for more of these, also to give readers ideas of the best AI applications.