FULLTEXT01

Linköping University - Department of Management and Engineering
Master’s thesis, 120 credits - Master’s programme

Spring 2023 - ISRN: LIU-IEI-TEK-A–23/04601–SE
Enhancing User Experience

Design workflow with Artificial
Intelligence tools
Autor:
Arezou Mortazavi
Supervisors:
Torbjörn Andersson
Tomas Jankauskas
Examiner:
Renee Wever
Linköping University
SE-581 83 Linköping,
Sweden
+46 013 28 10 00,
www.liu.se
Abstract
This thesis explores the intersection of Artificial Intelligence (AI) and User Experience (UX)
design, aiming to leverage AI capabilities to enhance the overall user experience design
process. The research included user interviews, data collection, and the usage of AI tools
to enhance efficiency. The collaboration with a company provided insightful information
and directed the research toward investigating AI capabilities in UX design. The study
clarifies the advantages of AI design tools, their impact on the UX design procedure, and the
possibility of further developments. It ends by introducing the AIUX app concept, which
provides a streamlined and collaborative approach to UX design. The results highlight
the significance of a balanced approach, where AI enhances rather than replaces human
creativity and invention. As AI technology develops quickly and is increasingly included in
design tools, this thesis offers a starting point for further study.
2
Acknowledgments
I would like to express my sincere gratitude to all the participants who contributed their
invaluable insights during the course of this project. Additionally, I extend my heartfelt
appreciation to my supervisors, Tomas and Torbjörn, for their invaluable guidance and
constructive feedback. I am also thankful to Renee, my examiner, for his expertise and
helpful comments. Lastly, I would like to extend a special word of thanks to Amin for his
consistent support throughout my master’s study.
3
Contents
List of Acronyms 3
1 Introduction 8
1.1 Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Aim and goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Toyota material handlling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 User experience design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Research question . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.7 Deliverables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.8 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Literature Review 10
2.1 User experience design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Creativity in Artificial intelligence . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Artificial intelligence in design practice . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 Text to image generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Method 14
3.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.1 Research through design . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.2 Design framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.3 Secondary research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.4 Web based survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.5 Interview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.6 Focus group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.7 Workshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4 Research process 17
4.1 Discover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1.1 Exploring current AI tools in design . . . . . . . . . . . . . . . . . . . 17
4.1.2 User experience design process . . . . . . . . . . . . . . . . . . . . . . 31
4.1.3 Focus group session with design students . . . . . . . . . . . . . . . . 32
4.1.4 Semi-stractured interviews . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Define . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2.1 Web based survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2.2 Exploring Designers’ Attitudes towards Artificial Intelligence support
tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2.3 Workshop with design students . . . . . . . . . . . . . . . . . . . . . . 42
4.2.4 Persona . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4
5 Ideation 47
6 Prototype 50
6.1 Low fidelity prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
6.2 Figma prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
7 Evaluation 63
7.0.1 Usability test with design students . . . . . . . . . . . . . . . . . . . . 63
7.0.2 Expert evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
8 Discussion 66
8.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
8.2 The Influence of AI on Design Education . . . . . . . . . . . . . . . . . . . . 66
8.3 Ethical concerns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
8.4 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
9 Conclusion 69
9.1 Knowledge contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
10 Future works 70
10.1 Collaboration of designers and data scientists . . . . . . . . . . . . . . . . . . 70
5
List of Figures
2.1 Generated image by Midjourney . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1 Framework for innovation [1] adapted from Double diamond by design council [2] 15
4.1 List of text to image generators . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.2 generated image by Gencraft, prompt: a futuristic design studio, innovative,
working with AI tools::5, creative tools on the desk::2, glass walls, with natural
light, cozy, green::3, warm, eloquent –no allwhite . . . . . . . . . . . . . . . . 18
4.3 generated image by Dall.E2, prompt: a futuristic design studio, innovative,
4.4 generated image by Bluewillow, prompt: a futuristic design studio, innovative,
4.5 generated image by Dreamstudio, prompt: a futuristic design studio, innova-
tive, working with AI tools::5, creative tools on the desk::2, glass walls, with
natural light, cozy, green::3, warm, eloquent –no allwhite . . . . . . . . . . . . 21
4.6 generated image by Playground, prompt: a futuristic design studio, innova-
4.7 generated image by Midjourney, prompt: a futuristic design studio, innova-
4.8 List of text to 3D generators . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.9 List of AI-powered design tools . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.10 List of AI-powered tools that support text creation . . . . . . . . . . . . . . . 26
4.11 List of AI-powered figma plugins . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.12 An example of QoQo and Cluify output that are AI-powered Figma plugins 28
4.13 An example of Dreamer output that is an AI-powered Figma plugin . . . . . 29
4.14 UX design process [3] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.15 Focus group with design students . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.16 AI tools and Figma plugins can be used in different stages of UX design process 33
4.17 Design team’s answers to the web-based survey . . . . . . . . . . . . . . . . . 39
4.18 Designers’ respondents to some questions in the Pfeiffer’s study [4] . . . . . . 41
4.19 Picture of the workshop with design students . . . . . . . . . . . . . . . . . . 43
4.20 Original Persona template provided by Fabrie . . . . . . . . . . . . . . . . . . 45
4.21 Customized persona for this project, using Fabrie template and Midjourney
for picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.1 Mind mapping after adjusting Fabrie’s version . . . . . . . . . . . . . . . . . . 48

5.2 Brainstorming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6
6.1 Prototyping a feature that suggests different AI tools to designers in each
stage in JSFiddle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.2 Prototyping Idea/image generator in JSFiddle . . . . . . . . . . . . . . . . . . 52
6.3 Participants testing wireframe sketches and Low fidelity prototypes . . . . . . 53
6.4 Userflow for the AI-powered UX design assistant application . . . . . . . . . . 55
6.5 Figma prototype, the landing page, commercial . . . . . . . . . . . . . . . . . 56
6.6 Figma prototype, Login page . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.7 Figma prototype, dashboard page, creating new board . . . . . . . . . . . . . 57
6.8 Figma prototype, data entry process, specifying the current stage of your
project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.9 Figma prototype, data entry process, discovery stage . . . . . . . . . . . . . . 58
6.10 Figma prototype, data entry process, define stage . . . . . . . . . . . . . . . . 58
6.11 Figma prototype, a holistic view of the board . . . . . . . . . . . . . . . . . . 59
6.12 Figma prototype, Discover stage on the board, AI suggest user to add another
competitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.13 Figma prototype, AI suggestion accepted, Right clicking for more options AI
can do, picking interview questions . . . . . . . . . . . . . . . . . . . . . . . . 60
6.14 Figma prototype, interview questions generated by AI is added to the board . 60
6.15 Figma prototype, define stage, AI suggests a persona for your project . . . . 61
6.16 Figma prototype, user accepted that persona by AI be added to the board . . 61
6.17 Figma prototype, Ideation phase, AI suggests user to start by benchmarking 62
6.18 Figma prototype, Dashboard page, introducing all the boards user have . . . 62
7.1 Expert evaluation of the Figma prototype . . . . . . . . . . . . . . . . . . . . 64

7.2 Usability testing of the Figma prototype with design students . . . . . . . . . 65
10.1 The gaps between design thinking and AI science process [5] . . . . . . . . . . 70
10.2 AI by design approach which is a combination of design thinking and data
science processes [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7
1 Introduction
The core concepts of this thesis will be covered in this chapter.
1.1 Context
This thesis is conducted in the master’s design program at Linköping University and Toyota
Material Handling, located in Mjölby, is collaborating on this project.
1.2 Background
In recent years, technological advancement, and the rise of Artificial intelligence (AI) have
revolutionized many industries. The field of AI has transformed from a theoretical concept
to a practical application and is now deeply integrated into many parts of society. [6] The
developments that happened in early 2020 are the fourth wave of the technological revolution
which focuses on autonomy, generative and creative applications, where AI services act on
behalf of humans. [7] For instance, using machine learning algorithms that analyze user data,
companies can recommend new products based on the user’s preference and even assist them
in chatbots.
1.3 Aim and goals

The purpose of this thesis is to explore the impacts of AI-powered tools on the user experience
design process. To achieve this aim, the thesis will begin by providing an overview of the
current state of the art in AI tools, their capabilities, and their limitations. The thesis will
then examine the ways in which AI tools have been used in design, and provide insights into
how designers can best leverage these tools in their design process.
In this research, I will delve into the world of AI image generation technology to gain a
comprehensive understanding of how it works. Through this exploration, I will be able
to realize the potential of AI image generation technology and how it can be utilized in
various design applications. Ultimately, this will contribute to a better understanding of
the capabilities and limitations of AI in the field of design. Incorporating AI into the design
process can provide significant benefits in terms of efficiency, automation, and insights. By
comparing the design process with and without the use of AI, I can identify the advantages
and disadvantages of utilizing AI in design.
1.4 Toyota material handlling

Toyota is widely recognized as one of the world’s largest and most influential vehicle manu-
facturers producing and selling automobiles and auto parts. Its three core business sectors
are the development, manufacture, and sale of passenger cars, minivans, and trucks, as well
as related accessories and services. Toyota also participates in the creation of intelligent
transportation systems. Lease agreements are among the financing solutions that the Fi-
nancial Services sector provides to Toyota dealers and customers. The All Others section
includes housing, telecommunications, and other business activities in addition to its auto-
motive operations. Kiichiro Toyoda began the business in 1937, and it is based in Toyota,
Japan.
8
Toyota has been recognized for its leadership in designing and selling hybrid electric vehicles,
such as the 1997 Toyota Prius. Today, the company sells over 40 hybrid car models globally.
[8]
Toyota Material Handling is a world leader in the materials handling business, with roots
in Japan. Approximately 4,000 people work in Sweden to design, manufacture, sell, and
service material handling products. Manual hand trucks to innovative energy solutions and
autonomous self-driving vehicles weighing up to 1.5 tonnes are among them. Toyota Material
Handling has offices in Sweden’s Mjölby, Gothenburg, and Stockholm. [9]
1.5 User experience design

User experience (UX) design is the process by which design teams create products that
provide meaningful and relevant experiences to users. User experience design involves the
entire acquisition and integration process of a product, including branding, design, usability,
and function. [10] Creating an experience entails not just making the software simple to use,
but also creating extra product-related experiences such as marketing campaigns, packaging,
and after-sales support. Ultimately, UX design is concerned with delivering solutions to
issues and needs. [11]
1.6 Research question

The use of AI in different industries has had both positive and negative effects, which has
raised curiosity about its potential impacts on the field of design. In this thesis, I aim to
explore the answers to the following questions related to the impact of AI on design.
-How can AI design tools influence the User Experience design process?
1.7 Deliverables
The project’s deliverables include:
-A Figma prototype illustrating the idea
-A project report that describes the research methodology, the process, and the outcomes.
1.8 Delimitations
The following limits have been specified in order to more precisely define the scope of this
project:
-The thesis will not be a functional application; rather, it will showcase the idea through a
Figma prototype.
-It is crucial to note that the field of artificial intelligence is continually growing, and this
research reflects the most up-to-date information available as of the end of April 2023.
9
2 Literature Review
In this chapter, the essential concepts presented in this thesis will be overviewed.
2.1 User experience design

User experience is a multidisciplinary discipline. UX designers come from a variety of
backgrounds, including graphic design, programming, psychology, and interaction design.
Designing for humans entails working with a broader scope in terms of accessibility and
supporting various possible users’ physical constraints, such as reading small text. [11]
While the typical activities of a UX designer may vary depending on the company they
work for or their project domain. However, they frequently include user research, persona
creation, building wireframes and interactive prototypes, and testing concepts. The majority
of UX designers employ some kind of user-centered work method, and they keep applying
their best judgment to the problem at hand until it is fully and effectively solved. [3]
Figure 2.1: Generated image by Midjourney
10
2.2 Creativity in Artificial intelligence
The implementation of artificial intelligence in creative practices has advanced recently.
They have achieved success in generating images, music, videos, or stories. However, since
creativity emerges from the complex interplay of conscious and unconscious cognitive pro-
cesses, the creativity of AI still has some limitations and cannot replace the human mind.
In addition, humans have a sense of purpose, meaning, sympathy, and emotional connection
that allows them to design products or services that inspire, provokes thought, and resonates
deeply with audiences. They may become far more capable and nuanced in the future or get
better at mimicking the style and form of human creativity, but human creativity will always
be needed to craft a vision, make depth meaningful connections, and design experiences that
touch hearts and minds. [12]
2.3 Artificial intelligence in design practice

According to Aceto, Persico, and Antonio [13] and Lember, Brandsen, and Piret [14], digital
technology enables access to new data forms via smart devices, provides novel communica-
tion and information exchange methods, enables innovative data collection, analysis, and
decision-making via AI-based programs, cloud computing, big data analytics, and machine
learning, etc. The incorporation of technological tools in design practice has become in-
creasingly important in the modern era. Designers who leverage these tools are likely to
be more productive, efficient, and creative. As technology continues to evolve, it is crucial
for designers to stay up to date on the latest advancements to remain competitive in their
respective fields.
Professionals in the creative business frequently experience issues with technology that does
not operate as planned, resulting in project delays and inefficiencies. Any technology that
can minimize effort and handle tiresome chores, such as image search or learning new fea-
tures, is greatly sought after to boost productivity. The creative process includes several
iteration, adaptation, and process management phases, which are sometimes considered as
the most time-consuming aspects of the project. These jobs frequently require manual op-
erations that can be streamlined with the help of AI and machine learning, which creative
professionals regard as an area for major progress. To summarize, there is an increasing
expectation that technology will benefit the creative business by simplifying complex pro-
cedures and eliminating human work. [4]
AI systems today have narrow, specialized capabilities such as generating variations and
combinations within a limited scope. They can assist and enhance human creativity by
suggesting new ideas, patterns, text contents and providing sources of inspiration and stim-
ulation. With AI, designers can automate some types of tasks, allowing them to focus on
more complex aspects of the design. It is essential to recognize that the use of AI in de-
sign also has its disadvantages. For instance, AI tools may not always reflect the designer’s
intention, leading to a loss of control or distraction.
Gaining an understanding of machine learning can aid us in navigating the complexities,

and capabilities of this technology.
11
2.4 Machine Learning
Machine learning is one of the approaches and subfields of artificial intelligence which can
be used in numerous fields. Machine learning has the ability to use computational methods
to learn from the analysis. These algorithms make use of data sets to identify and discover
patterns that can be used to develop and improve robots’ capacity for autonomous decision-
making.
Machines may form neural connections on several hierarchical data levels, similar to how the
human brain does. These networks have given rise to deep learning, a new branch of artificial
intelligence. [6] Machine learning offers advantages, but it also has substantial drawbacks,
such as the need for a large amount of data—particularly high-quality training data. [15].
2.5 Text to image generators

One area where AI is having a significant impact is in the creation and manipulation of
images. AI-powered image tools, such as generative adversarial networks (GANs), style
transfer algorithms, and object detection models, have made it easier and faster to create
visually appealing and realistic images. The use of AI in UI/ UX design could have a lot of
benefits and drawbacks.
It not only reduces the amount of time and effort needed to be spent to create images, but it
has also enabled designers to explore new creative possibilities and generate realistic images
which is time-consuming or impossible to create manually.
Dall-E 2, Midjourney, Stable Diffusion and other text to image software are rapidly de-
veloping, and every month provide new features. Nowadays, it simply works by writing a
text as an input, and the output will be an image in a dictated style. Although the inputs
are usually simple, the outputs can be extraordinary, and even win the art competitions
award. [16]
Different models are used to generate images based on text prompts. The Generative Adver-
sarial Networks-based model (GAN) is one of the elementary model’s examples and DALLE-
2 is a newer model that benefits from contrastive Learning Image Pretraining (CLIP) and
diffusion models. Even though errors may occur in some cases such as generating graphics
with text or photorealistic human faces, it still creates aesthetic, accurate, and relatively
creative pictures. [17]
At a more specific level, the utilization of diffusion-based techniques in text-to-image gen-

erators has garnered significant attention and interest across diverse fields. The process of
diffusion involves providing artificial intelligence (AI) with a reference object, such as an
image, along with contextual information that humans associate with it (such as ”this is a
black cat”). Subsequently, the diffusion algorithm introduces random alterations or ”noise”
to the image gradually, until the original image becomes indistinguishable. [16]
Image generators are AI models that are trained to create images from scratch, based on
patterns and data they have learned from a dataset. To create an image, an image generator
starts with a random set of values, and then iteratively adjusts these values based on the
patterns it has learned from the dataset. The generator can also incorporate feedback from
a user or a target image to further refine the generated image.
The process of creating an image using an image generator involves several steps, including
encoding, synthesis, and decoding. During the encoding phase, the generator maps an input
signal, such as a random set of values, to a higher-dimensional space where it can better
12
capture the features and patterns of the target dataset. In the synthesis phase, the generator
uses this encoded information to create a new image. Finally, during the decoding phase,
the generator maps the synthesized image back to its original, lower-dimensional form.
Image generators, also known as generative models, are AI algorithms that create new dig-
ital images based on patterns and features that they learn from existing images. These
algorithms consist of two main components: a generator and a discriminator. The generator
is responsible for creating new images based on random noise input that is fed into the algo-
rithm. The discriminator, on the other hand, evaluates the generated image and compares
it to the original image to determine whether it is real or fake.
The generator is trained through a process called backpropagation, where the algorithm
adjusts its internal parameters based on the feedback received from the discriminator. As
the generator becomes more skilled at creating realistic images, the discriminator becomes
better at detecting fake ones.
Over time, the generator learns to create increasingly realistic images by mimicking the
patterns and features of the original images. These generated images can then be used for
a variety of purposes, such as creating new visual designs, improving image editing tools, or
even generating realistic synthetic images for scientific research.
Given that text-to-image applications rely heavily on datasets containing large numbers of
images and associated text for their model development, it is pertinent to inquire about
the origins of these images. While the origins of such images may be unclear for certain
programs, in several cases, known sources of these images raise ethical concerns. [16]
13
3 Method
This section outlines the methodology employed in this thesis and the methods that were
used.
3.1 Methodology
The methodology employed in this research was Research Through Design (RTD), which
involved conducting semi-structured interviews, focus group sessions, and workshops to ad-
dress the research question. Artificial intelligence was studied, and existing tools that sup-
port design practices were experimented with. Throughout the research process, ideas were
prototyped, tested, and iteratively developed to improve their quality.
3.1.1 Research through design

Research through Design (RtD) is an approach to research that involves using design actions
as part of the knowledge generation process. This means that investigating, iterating, and
evaluating concepts are integrated into the research process, leading to knowledge contribu-
tion. [18]
3.1.2 Design framework

The Double Diamond design process was introduced by the Design Council in 2004 and has
since been widely adopted. This framework provides a structured approach for designers to
follow and consists of two diamonds, each representing a different phase of thought: divergent
thinking (exploring the issue extensively) and convergent thinking (focusing on a particular
approach). The design process is broken down into four distinct phases, which are organized
within the two diamonds: Discover, which involves investigating and determining user needs
(divergent thinking), Define, which involves connecting findings to the problem (convergent
thinking), Develop, which involves generating potential solutions (divergent thinking), and
Deliver, which involves selecting the best solution and putting it into action (convergent
thinking). [1]
The Design Council introduced an updated innovation framework in 2019, which replaced
its previous Double Diamond model. The new framework features four key concepts that are
crucial to consider during the design process. Unlike the linear nature of the old model, this
updated version outlines 12 potential iterations and highlights the importance of a human-
centered approach to design. Additionally, the framework underscores the value of using
visual communication techniques and prioritizing cooperation and co-creation throughout
the design process. [2]
14
Figure 3.1: Framework for innovation [1] adapted from Double diamond by design council [2]
3.1.3 Secondary research

To conduct this research, sources such as academic journals, trade publications, market
research reports, and online resources are reviewed and studied to gain a comprehensive
understanding of the field.
3.1.4 Web based survey

Online surveys are one approach to gathering data online which has become increasingly
popular due to their various benefits, including time efficiency, cost-effectiveness, and the
ability to reach a larger and more diverse pool of respondents globally. These surveys can be
distributed via email, social media, or embedded into websites, making them easily accessible
to a wide audience. They are particularly useful for collecting specific data from a targeted
group of people, such as their preferences, opinions, behavior, or factual information. [19]
To achieve this project’s goal, I conducted a web-based survey to gather additional informa-
tion from other members of Toyota’s design team about their attitudes toward AI tools and
their impacts on UX design. Although the number of responses to the web-based survey
was minimal, the insights collected from the survey were important in giving me a deeper
knowledge of Toyota’s design team’s perspectives on AI technologies and their impact on UX
design. The responses assisted me in identifying potential areas of concern and skepticism,
as well as opportunities and potential benefits that might be realized by incorporating AI
into the design process. Furthermore, survey respondents have emphasized the need for ad-
ditional exploration and research into some critical areas connected to AI in UX design. The
extent to which AI tools can increase the effectiveness and efficiency of the design process is
one of these topics, as are ethical considerations, and the possible effects of AI on creativity
and innovation in the design process.
15
3.1.5 Interview
Interviews are a useful means of gathering diverse information on a particular topic, including
the interviewee’s personal opinions, perceptions, and attitudes, as well as their background
knowledge or specialized expertise on the subject. It is not uncommon for interviews to
encompass both types of information. The flexibility of this method and the level of control
that the interviewer can exert over the respondent’s answers may vary depending on the
situation. [20]
Semi-structured interviews follow a guide with predetermined questions, but the interviewer
may ask follow-up questions to gather additional information based on the interviewee’s
responses. This approach allows for an in-depth comprehension of the data gathered. [20]
To accomplish the goal of this project, I conducted a series of semi-structured interviews

with four individuals, each from a different professional background. The first interviewee
was a Senior UX designer employed by Toyota. The second interviewee was a junior visual
designer who actively utilizes AI tools in his work. The third interviewee was a deep learning
expert. The final interviewee was a mid-senior product designer who has been actively and
curiously working with image generators for around one and half a years.
3.1.6 Focus group

Focus groups are a popular qualitative research technique in the field of design for collecting
information from a group of people. A focus group is a ”semi-structured group interview
process that allows for a more interactive and exploratory discussion on a particular topic.”
Focus group participants are often chosen based on particular criteria and requested to
discuss their thoughts, impressions, and experiences on the research issue. The approach
can offer a plethora of data on how customers engage with a product or service, as well
as insights into their requirements and preferences, making it particularly helpful in design
research. Focus groups allow designers to better understand their target audience. [21]
To achieve this project’s goal, we conducted a focus group study with two design students
in the first year of their master’s to draw the usual design process and figure out their
mindset about AI tools. The feedback we received from the design students allowed us to
make adjustments where necessary and ensure that it was tailored to meet the needs and
expectations of the target audience. Overall, this was a valuable experience that helped us
to identify any potential issues and refine our approach.
3.1.7 Workshop
Workshops are a design study technique that can be used to gather information and partic-
ipant insights. Workshops have been a popular way of doing design research because they
may give both researchers and participants a collaborative and participatory setting. It can
be organized in a variety of ways and can include a variety of tasks like idea generation,
prototyping, and testing. The objective is to involve participants in the design process and
to establish a forum for honest and helpful discussion. [22]
16
4 Research process
4.1 Discover
During the discovery phase, I examined the available AI tools that support design practices,
made a list of them all, and categorized them according to the design practices that each tool
mostly supports. Meanwhile, I delved into researching the UX design process by running a
workshop with design students and doing secondary research to explore how AI tools could
be integrated into the process and what their impact might be.
4.1.1 Exploring current AI tools in design

I ran several experiments in numerous ways of prompting to identify the differences between
various image generators and design tools. The purpose was to determine their distinguishing
features, strengths, and limitations. Some of the tools I utilized in my research are listed in
the table. I’ve also added a number of images that were produced from a single prompt in
different models.
Figure 4.1: List of text to image generators
17
Figure 4.2: generated image by Gencraft, prompt: a futuristic design studio, innovative,
working with AI tools::5, creative tools on the desk::2, glass walls, with natural light, cozy,
green::3, warm, eloquent –no allwhite
18
Figure 4.3: generated image by Dall.E2, prompt: a futuristic design studio, innovative,
19
Figure 4.4: generated image by Bluewillow, prompt: a futuristic design studio, innovative,
20
Figure 4.5: generated image by Dreamstudio, prompt: a futuristic design studio, innovative,
21
Figure 4.6: generated image by Playground, prompt: a futuristic design studio, innovative,
22
Figure 4.7: generated image by Midjourney, prompt: a futuristic design studio, innovative,
23
Figure 4.8: List of text to 3D generators
24
Figure 4.9: List of AI-powered design tools
25
Figure 4.10: List of AI-powered tools that support text creation
26
Figure 4.11: List of AI-powered figma plugins
27
Figure 4.12: An example of QoQo and Cluify output that are AI-powered Figma plugins
28
Figure 4.13: An example of Dreamer output that is an AI-powered Figma plugin
29
Current abilities of AI Some of the specific ways AI is currently being used in UI and
UX design include:
• Generating low-fidelity prototypes
• Creating variations of a design for A/B testing
• Suggesting new fonts, colors, and other stylistic elements that fit within a brand’s
guidelines
• Automating the production of assets like icons and illustrations
• Predicting the heatmap of the user’s interaction
• Analyzing how users interact with an interface and identifying potential UX

issues
• Producing inspiration for designers to build upon
• Translating hand-drawn wireframes into polished user interfaces
Future possibilities of AI Looking ahead, some of the ways AI may further integrate
with and enhance the design process include:
• Generating interactive prototypes that designers can build upon
• Producing custom design assets like illustrations, icons, and imagery on-demand
to match a designer’s needs
• Automating the production of design documentation like style guides, pattern

libraries, and design systems
• Suggesting completely new design concepts and ideas that designers would never
have thought of on their own
• Continuously analyzing how users interact with live interfaces and providing
recommendations to improve their experiences
• Generating personalized design feedback for individual designers based on their

skills, strengths, and weaknesses
• Collaborating with designers in real-time through conversational interfaces to

brainstorm, get input, and refine designs
The future of AI in design is incredibly promising but will depend on how well designers
can collaborate with these new tools. By developing a symbiotic relationship with AI sys-
tems instead of viewing them as a threat, designers can utilize them to push their creative
capabilities even further and design experiences that are both highly functional and deeply
human. The age of human and AI collaboration in design is here, and the possibilities are
as endless as the imagination.
30
Figure 4.14: UX design process [3]
4.1.2 User experience design process

The infographic depicts a traditional user experience process that adheres to the Design
Thinking methodology. However, because each project is unique, UX methods and tools
must be tailored accordingly. [3]
To further our research, I have planned to map the current design process used by two of
the design students. Additionally, I intend to experiment with several AI-powered tools
with them, in order to gain a better understanding of their potential impact on the team’s
workflow. Through this process, I hope to identify areas where AI tools could be beneficial
and investigate the potential limitations of AI. By testing AI tools with the design team,
I hope to gain a more practical understanding of the user’s insights as well. Overall, my
planned experimentation with AI-powered tools will provide us with valuable insights into
their potential benefits and limitations when used in the design process.
31
Figure 4.15: Focus group with design students
4.1.3 Focus group session with design students

To begin, I asked the participants to draw their own design process on paper, based on one
of their recent projects. By drawing their process, the participants were able to visualize
the steps involved in the process and gain a deeper understanding of their own design
methodology. This activity also provided a starting point for the focus group study and
helped to start a discussion among the participants about their workflow.
32
Figure 4.16: AI tools and Figma plugins can be used in different stages of UX design process
When participants completed their drawings, we discussed their workflow, and I introduced
some AI tools that could be used in each step of the design process based on my exper-
imentation with these tools and their features. The purpose of this activity was to help
them understand how AI can enhance their design workflow. By discussing the various AI
tools available, the participants were able to gain a better understanding of how AI can be
integrated into their own design processes. This activity also provided an opportunity for
the participants to ask questions and share their thoughts and concerns.
Following the introduction of AI tools that are useful in each stage of the design process, I
asked participants to share their opinions about the tools and then invited them to try out
some of the tools themselves.
By trying out some of the tools such as ChatGPT, Dall e, and Midjourney, the participants
were able to see how AI could be used to generate images and answer some design-related
questions. This activity also provided an opportunity for the participants to share their
thoughts and feedback about the tools and discuss their potential applications in their own
design work.
Insights from participants Participants were amazed by the abilities of the AI tools that
were introduced. This suggests that they are receptive to exploring new technologies and
are open to evolving their work processes. As a moderator, it was important to cultivate this
enthusiasm and highlight the full range of capabilities that AI tools can offer. By showcasing
how AI can streamline repetitive tasks, inspire ideas, and optimize workflows, participants
were able to recognize the potential of AI to enhance their productivity and efficiency. It’s
understandable that some participants might be concerned about the impact of AI on their
job security. As a moderator, it is important to address these concerns and help participants
understand the potential benefits and limitations of AI in the design industry. One way to
address these concerns is to emphasize that AI tools are not intended to replace human
designers, but rather to augment their skills and ease their workflow. It’s important to
highlight that AI can help designers work more efficiently, allowing them to focus on higher-
level tasks that require human creativity, empathy, and critical thinking. Ultimately, it’s
important to acknowledge and validate the concerns of participants while also providing a
balanced perspective on the potential impact of AI in the design industry.
33
4.1.4 Semi-stractured interviews
All interviews were conducted online and lasted approximately 40 to 60 minutes. The
audio was recorded and transcribed afterward. All interviewees were asked identical ques-
tions, which included a brief introduction to themselves, their professional backgrounds, and
inquiries about their present occupations. Additionally, I adjusted my questions to each in-
terviewee’s occupation and familiarity with AI in order to elicit information and viewpoints
relevant to my study objectives.
34
Interview A
• Academic background: Media and communication science
• Profession: Senior UX designer
• Location: Gothenburg, Sweden
• How does the design process look like in your team? Firstly, we try to understand
the need for a product and do interviews to research, then based on that do
designs. After designing a prototype, we test them with our users and after some
evaluations, we release it. So I would say that’s like a normal procedure.
• What software do you use? We use only Figma.
• Have you faced any issue that you think would be easier if it was supported by
AI? We have a design library right now but it would be nice if Figma could go
through the designs that you’ve done, and then make a design library out of that
automatically.
• Have you ever used AI tools for professional purposes? No I have not.
• Have you used AI tools for other purposes? Yes, I have tried ChatGPT and Dalle
once or twice, I was advised to test these tools by a buddy who is involved in 3D
concept modeling.
• Do you think AI can be used in a professional way in UX design? Yes, of course.

AI may be highly helpful for things like benchmarking, recommending color
schemes, and other UI design elements. I once requested a complimentary color
from ChatGPT, and he suggested a very good answer to me, he said: ”There is
not a perfect color, you need to think about your context. I also think AI can
speed up and simplify processes like crafting user stories or feature descriptions.
The outcome may need some manual changes but still make your work faster
and easier.
• Which phase of the design process do you think artificial intelligence will be
most useful? having AI help in the final stages of the design process would
be advantageous. It might study the design and make recommendations, for
example, or it might examine user behavior patterns. For jobs like interviewing,
documentation, and data analysis, AI is well suited. I guess it can do these kinds
of tasks which are very time-consuming for humans to do.
• If you wanted to use AI professionally, where do you think is the stage you would
use it? I would use it for inspiration, or if I want to make any illustration.
• What are the positive and negative impacts of AI tools in your opinion? In the
short term, it is very positive but in the long term, it may change our way of
working or cause job losses. Some tasks that are currently considered tedious
could become automated, leaving only more enjoyable tasks for us to handle.
35
Interview B
• Academic background: Digital creative
• Profession: Visual designer
• Location: Stockholm, Sweden
• Have you used AI tools professionally? Yes, I frequently utilize them for inspi-
ration and as a foundation for my illustrations.
• What kind of AI tools do you use? I’ve tried several image generators, but
Midjourney is my favorite. It is, in my opinion, the best of all the tools I have
used so far.
• How long have you been working with them? Around eight months
• Do you believe that the quality of the generated images is high enough to be
considered as the final work for your job? Yes, Midjourney can produce extremely
high-quality photos if the quality level is specified in your prompt. Recently, I
showed one of the generated images for a project to my boss. He was amazed by
the quality and requested minimal changes to that. Indicating that the image
was of sufficient quality to be used as a finished product in my project.
• Have not you faced any issues with your team creating some of your design works
with Midjourney?
Fortunately, my company is very open and receptive to it because they see the
efficiency it brings to our work and the high quality of the images produced.
However, we have used AI as an inspiration source until now. For instance, if
the generated image by AI was perfect, we still have to edit and use it just for
the foundation of our work. Probably this will change in the future.
• In which step of the design process do you believe AI can be most useful?
In my experience, AI can be a valuable source of inspiration and aid in designing

storyboards or illustrating your ideas. Depending on the project, I have found
various AI tools useful in all stages of the design process to fulfill different needs.
36
Interview C
• Academic background: Design for play
• Profession: Creative designer
• Location: Arhus, Denmark
• Have you used AI tools professionally? Yes
• What kind of AI tools do you use? Please identify a couple of them.
Midjourney - Dall e- Stylegan
• How long have you been working with them? one year now
• What kinds of changes have they experienced up to this point?
Midjourney is a relatively new tool that is constantly improving its image gener-
ation capabilities, user interface, and other features. As it continues to develop,
is becoming more powerful and useful for product designers. As an example of its
development, in addition to generating images from text, Midjourney’s mother
company, Runway ML, has also developed Runway Gen 2, which can generate
video from the text. These developments demonstrate the immense potential
of AI-powered design tools and highlight the ways in which they are rapidly
changing the design landscape.
• Which design field do you believe they are most powerful in?
In my opinion, Midjourney is most powerful in the product design field. It can

help designers quickly generate and visualize images of their ideas without the
need for extensive 3D modeling skills. Product design often involves a lot of trial
and error and iteration, and Midjourney can speed up this process significantly
by allowing designers to quickly create and test out different design concepts.
Additionally, it can help designers communicate their ideas more effectively to
clients and stakeholders, which can improve the overall design process.
• In which step of the design process do you believe AI can be most useful? From
a product designer’s perspective, AI can be useful in different steps of the design
process, but in the case of Midjourney, it is most useful during the ideation
phase. This is because it allows designers to quickly generate and visualize
their ideas, which can help them explore different design concepts and iterate
quickly. By using Midjourney during the ideation phase, designers can rapidly
test out different design concepts and refine their ideas before moving on to the
prototyping or production phases. This can save a significant amount of time
and money, as it can help catch design flaws or issues early on in the process.
37
4.2 Define
During this phase, I conducted a web-based survey, reviewed literature related to users’
attitudes toward AI, and organized another workshop to validate the results from the survey
and literature review. As a result, a persona was developed to better understand and define
the user.
4.2.1 Web based survey

The design team at Toyota Material Handling in Sweden was emailed a web-based survey
that was created through Google Forms. The survey questions were aimed to elicit infor-
mation about the design team’s AI familiarity and experience, specifically whether they
have utilized any AI-powered tools as a design supporting tool or not. The goal of these
questions was to acquire a better idea of the team’s familiarity with AI as well as their
level of willingness to utilize this technology in their professional activities. Unfortunately,
the number of participants who offered their opinions was low, with only three individuals.
However, it offered insightful observations and ideas that were helpful for pursuing more
in-depth research. Despite the small survey size, the responses obtained can still contribute
to the overall understanding of the topic and assist in guiding the direction of the project.
According to the results of my survey, participants are interested in learning more about
AI tools and the benefits they may give. However, it seems that they are not yet using
them professionally. While there is a willingness to investigate the possibilities given by AI,
there may be some hesitation or lack of familiarity with these technologies that prevent their
acceptance in a professional setting.
38
Figure 4.17: Design team’s answers to the web-based survey
39
After reviewing the responses to the web-based survey, I delved into an extensive literature
review that focused on exploring designers’ attitudes towards the design support AI tools.
The review was highly informative and yielded several insightful studies that shed light on
the topic. These studies provided a more in-depth understanding of the designers’ attitudes
toward AI and their impacts on design [4]. Overall, the literature review proved to be an
excellent resource for validating the results of the web-based survey and provided a valuable
perspective on the topic.
4.2.2 Exploring Designers’ Attitudes towards Artificial Intelligence

support tools
According to a study, designers are willing to utilize creativity support tools, particularly
for tasks that do not require a high level of creativity. They have expressed a positive
attitude towards the potential of AI tools as collaborators in design projects. Although
there may be a need for a shift in perceptions before designers fully embrace the idea of AI
tools serving as co-creators, they seem open to accepting them in supportive roles such as
assistant, researcher, or facilitator. Such positions would reinforce, rather than diminish,
their role as a creative. [23]
In another study by Pfeiffer [4], respondents were asked about their thoughts on an AI-
based creative assistant that would offer tailored assistance based on machine learning in
this research segment. While some found the idea strange or unsettling, most were willing
to try if two conditions were met. First and foremost, the assistant must outperform current
voice assistants such as Siri or Alexa. Second, this must be on their terms and under their
authority. Respondents stated a preference for call-in assistance rather than an always-on
technology that could disrupt their job. Overall, AI-based creative supporters have potential,
but they must fit certain requirements and be managed by the user.
40
Figure 4.18: Designers’ respondents to some questions in the Pfeiffer’s study [4]
41
According to a study [23], generally, designers are willing to use technologies that foster
creativity, especially for projects that don’t call for a lot of it. As possible collaborators in
design initiatives, AI technologies could potentially play a good and useful role in their opin-
ion. However, how these intelligent creative supporting tools are described and positioned
may play a significant role in their uptake.
They may be more open to employing these tools as assistants, collaborators, researchers,
or facilitators, roles that complement their own creativity. This suggests that designers are
interested in adopting AI to improve their work and are open to experimenting with new
ways of working while maintaining their creative involvement in the process.
Designers frequently identify fixation and distraction as two potential obstacles when dis-
cussing the obstacles to creativity. This demonstrates the difficulty in defining creativity
and locating chances to encourage it.
Ideation is a critical stage in the design process that demands high creativity. To explore
the potential of AI tools in the Ideation phase, I organized a workshop with design students.
4.2.3 Workshop with design students

This workshop was conducted with two participants who were master’s students in Design.
The concentration of the workshop was on the ideation stage of the design process. The
goal was to assess the benefits and drawbacks of employing AI technologies for ideating
approaches such as brainstorming or mind-mapping. Participants were first asked to describe
their experiences with traditional ideation approaches, as well as any difficulties they had
faced. After that, the participants were invited to use Eilla and ChatGPT which are AI
tools that support language processing and text content creation. They produced a list of
ideas with these tools and compare the results to their previous experiences using traditional
approaches. Afterward, they were invited to examine the advantages and disadvantages of
each tool and share their thoughts with the rest of the group.
Participants became aware of the possible limits of AI technologies in the ideation process.
They discovered that a potential limitation is that many of the ideas generated by AI already
exist, rather than being fresh ideas. They recognize that the automated idea-generation
process relies on existing data and as a result, they give us outputs based on the existing
concepts.
By the end of the workshop, participants had gained a deeper understanding of the benefits
and limitations of AI tools in the ideation phase of the design process. They could determine
which strategy best suited their individual or team needs. They believe these tools were
ideal for speeding up the benchmarking process and offering a brief overview of the chosen
subject. Additionally, the automated idea-generating tools offered a different viewpoint that
would not have been possible with more conventional techniques.
While idea generation using AI tools can help generate more ideas, participants noted that it
is important to consider the originality and uniqueness of ideas. They suggest that it would
be most effective to combine AI-generated ideas with traditional idea-generation methods
to provide a more diverse range of ideas while maintaining an element of creativity and
originality with less time.
Overall, participants agreed that using AI tools to generate ideas can be beneficial to op-
timize their workflow and reduce the amount of time and effort required for traditional
methods, but it is important to keep in mind its limitations and carefully consider how and
when to incorporate these tools into the design process.
42
Figure 4.19: Picture of the workshop with design students
43
Furthermore, the participants also experimented with image generators to visualize a few
ideas generated by automated idea-generating tools. However, they found that this tool
was not as helpful as they had expected. While the images produced by the generator were
visually interesting, they did not accurately represent the ideas in a meaningful way, making
it difficult to fully understand or evaluate the ideas.
Despite the challenges they encountered with the image generators, they still recognized the
potential benefits of using image generator tools in the design process. They suggested that
further exploration and experimentation with different AI tools and methods may lead to
more effective and efficient ways of generating and visualizing ideas.
44
Figure 4.20: Original Persona template provided by Fabrie
4.2.4 Persona
Afterward, I created a persona. This persona helps me evaluate the user’s expectations and
preferences when developing ideas at the next step. This project’s persona portrays a UX
designer who is curious and creative and wants to maximize her productivity and creativity
while saving time. She is open to testing new tools but tries to optimize her process with
user-friendly and uncomplicated software.
Since the goal of this project is to promote the use of AI in the design workflow, I have
personally experimented with several AI-powered tools in this project as well. I used the
Persona template provided by Fabrie, one of the AI tools in design. However, I customized
it to make it more suitable for my project’s needs. you can see the differences comparing
Figures 4.21 and 4.22.
In addition to the persona, I took into account Toyota’s requirements for this project, which
were focused on two key factors: feasibility and utilizing the advantages of AI.
45
Figure 4.21: Customized persona for this project, using Fabrie template and Midjourney for
picture
46
5 Ideation
Throughout this stage, I began developing concepts based on the information I obtained
previously. After considering all of the insights and literature reviews, I decided to cre-
ate an AI design assistant specifically for UX designers and began the mind mapping and
brainstorming process.
To investigate the feasibility of having an AI assistant for UX design practices, I integrated

ChatGPT and Fabrie into my real project.
First, I asked ChatGPT to create a target audience and list their pain points, provide me an
information architecture, a user flow, a copy of the hero section, and a FAQ section based
on common questions the target audience might have. I used ChatGPT as my UX research
assistant and compared all the data I gained with ChatGPT responses.
Although ChatGPT is not connected to the internet to search through all the updated data
that exist there, it performed well, and the responses were reasonable to a great extent.
I discovered that ChatGPT may actually serve as an assistant in the UX design process,
offering useful recommendations especially if you give basic information about your project
to that.
Meanwhile, I tested the mind-mapping tool given by Fabrie to see how well it understood
the context and generated ideas. It worked nicely, although some additional adjustments
were required to customize it for my specific project.
Taking into account the capabilities of AI tools and the desires of designers during their
workflow, I came up with some ideas for AI assistant features.
47
Figure 5.1: Mind mapping after adjusting Fabrie’s version
48
Figure 5.2: Brainstorming
49
6 Prototype
6.1 Low fidelity prototype

To prototype some of the ideas, I requested ChatGPT to provide me with codes for two
services: one that suggests AI-powered tools to help designers in each stage of the design
process and another that generates ideas based on user prompts and generates an image
for the chosen idea at the same platform. I then inputted these codes into JSFiddle to
create visual demonstrations of the ideas, which can be viewed in Figure 6.1 and Figure
6.2, respectively. These were some of the initial ideas that I was thinking to combine with
a project planning application, however, after speaking with experts and testing I realized
that it doesn’t use all the capabilities that AI can offer, so I led to design an AI-powered
assistant specifically for UX designers.
Afterward, I draw a wireframe version of an assistant application on paper and tested a

combination of digital and physical prototypes with design students who had participated
in previous workshops.
50
Figure 6.1: Prototyping a feature that suggests different AI tools to designers in each stage
in JSFiddle
51
Figure 6.2: Prototyping Idea/image generator in JSFiddle
52
Figure 6.3: Participants testing wireframe sketches and Low fidelity prototypes
53
Although the workshop with design students resulted in insightful information, I was keen to
get more qualified input from a seasoned UX designer working in this field. So, I conducted
an online interview with an experienced UX designer to receive feedback on this project’s
progress and to determine how to develop the prototype in a way that better meets the
user’s needs. The interview proved to be incredibly instructive.
Interview D
• Academic background: Industrial design
• Profession: UX designer
• Location: Tehran, Iran
I described the process of the project briefly to the interviewee and asked for his
opinion;
• What is your opinion about having an AI assistant on your team?
It would be great. Nowadays, we use ChatGPT to answer many of our questions,

but it is sometimes not accurate, since it requires back-and-forth communication
to get the best output.
• What is your opinion of my prototypes? Do you think you will use them?
Certainly, I appreciate the capabilities of AI, and having such a tool in our work
makes our work much easier. However, I believe that a more consistent tool
that is specifically designed to tailor to the mindset of UX designers would be
preferable.
• How does a user experience design process look across various projects?
I would say that for the most part, the process is similar across all projects,
although there may be some differences in the details or methods used in each
stage, depending on the context. Generally speaking, we start with research in
order to gain an understanding of the context, the problems faced by users, and
any other relevant data. The next stage involves defining the project’s scope,
taking into account stakeholder requirements and our target audience. We also
identify the main problems that we want to address and the values that we hope
to bring to the product. After that, we usually move on to benchmarking or
brainstorming sessions and then create a prototype. This process is repeated
multiple times to continuously improve the product until we arrive at the best
version possible.
• What challenges do you typically encounter during this process?
One issue that arises during the design process is that certain tasks can be tedious
and designers may not be willing to perform them, such as documenting data.
Another challenge is prioritizing design over user needs, which highlights the
importance of having someone check the product regularly to ensure it meets
user expectations.
• In which stage of the UX design process do you think AI can be helpful?
I believe AI can be helpful in all stages of the design process. We need a more
user-friendly and specific tool for designers to streamline the design process.
54
Figure 6.4: Userflow for the AI-powered UX design assistant application
6.2 Figma prototype

After testing my low-fi prototypes with design students, and speaking to a UX designer
about my idea, I developed my prototype with a User flow and a high-fidelity prototype
version using Figma.
Below are some pages of the prototyped application:
55
Figure 6.5: Figma prototype, the landing page, commercial
Figure 6.6: Figma prototype, Login page
56
Figure 6.7: Figma prototype, dashboard page, creating new board
Figure 6.8: Figma prototype, data entry process, specifying the current stage of your project
57
Figure 6.9: Figma prototype, data entry process, discovery stage
Figure 6.10: Figma prototype, data entry process, define stage
58
Figure 6.11: Figma prototype, a holistic view of the board
Figure 6.12: Figma prototype, Discover stage on the board, AI suggest user to add another
competitor
59
Figure 6.13: Figma prototype, AI suggestion accepted, Right clicking for more options AI
can do, picking interview questions
Figure 6.14: Figma prototype, interview questions generated by AI is added to the board
60
Figure 6.15: Figma prototype, define stage, AI suggests a persona for your project
Figure 6.16: Figma prototype, user accepted that persona by AI be added to the board
61
Figure 6.17: Figma prototype, Ideation phase, AI suggests user to start by benchmarking
Figure 6.18: Figma prototype, Dashboard page, introducing all the boards user have
62
7 Evaluation
To evaluate the Figma prototype, I conducted a usability test with design students and
forwarded the Figma prototype URL to two expert UX designers.
7.0.1 Usability test with design students

During the usability test, participants tried the prototype from the beginning to the end
and then provided their comments and insights about it. I took note of the participant’s
feedback to consider later.
Inshights from the participants
Participants indicated their satisfaction with the platform’s data entry capabilities and their
appreciation of the suggestions made along the way by artificial intelligence. They recog-
nized the value of using this application as it enables them to organize all the project data
in one place, with the added benefit of having an assistant constantly reviewing the informa-
tion and alerting them to any inconsistencies or gaps. They believed that having dedicated
software specifically designed for the field of design is preferable. This gives them a sense
of professionalism and assurance that the tool is tailored to their specific needs and require-
ments. They also mentioned that the capabilities of AI integrated into this application can
significantly speed up their design process and enhance collaboration with their teammates.
By enabling them to work together on the same board and pursue their design process col-
lectively, facilitates seamless teamwork and improves overall efficiency. The capability to
generate images is the feature they would want to see added to this program.
7.0.2 Expert evaluation

During the online tests with expert UX designers, participants provided comments and
feedback on the work within the Figma platform. While some of the mentioned problems
can be addressed in the final product, I made adjustments to the prototype based on their
feedback, resulting in an improved user experience.
Insights from the experts
UX experts mentioned some issues in the app, which have now been resolved. One of the
improvements made is the ability to skip certain questions if the user is unsure or does not
wish to answer at that moment, allowing them to proceed to the next step without filling
in that specific question. Another issue raised by UX experts was the inability to view the
entire board and see all the information added to it at a glance after answering the questions.
This has been addressed by implementing a feature that allows users to have an overview
of the entire board, providing them with a comprehensive view of the gathered information
in a single glance.
The UX experts identified several areas for further development in the future product. One of
the key points is creating a responsive application that can be accessed and used seamlessly
on mobile phones and tablets. Additionally, they suggested streamlining the process of
entering data, particularly for users who prefer to start with a blank board and complete
it in their own way. Another important aspect highlighted by the experts is ensuring that
all the key elements of the application have interactive functionality, providing a complete
and engaging user experience. These suggestions will be valuable considerations for future
iterations of the product.
63
Figure 7.1: Expert evaluation of the Figma prototype
64
Figure 7.2: Usability testing of the Figma prototype with design students
65
8 Discussion
8.1 Results
The research findings indicate that AI offers significant capabilities that enhance the work-
flow efficiency of UX designers across all stages of the design process. In the discovery
phase, AI can provide interview questions tailored to the defined target audience, facilitat-
ing data collection. During the define phase, AI can process data, identify competitors,
highlight potential technical issues, and even generate personas. In the ideation stage, AI
tools accelerate brainstorming and visualization through text and image generators. Creat-
ing prototypes becomes more streamlined as AI can be prompted with code or step-by-step
guidelines. Lastly, during the evaluation phase, AI can provide valuable feedback on designs,
predict heatmaps, and analyze usability test data to offer development solutions.
Based on my research, I recommend the development of a dedicated application that caters

specifically to the needs and workflow of UX designers, incorporating AI capabilities through-
out the design process. This tailored application would leverage AI in various practices across
different stages, allowing designers to maximize its benefits. By aligning the application with
the unique requirements of UX designers, it would provide a more seamless and efficient ex-
perience, enhancing their productivity and enabling them to leverage AI effectively.
8.2 The Influence of AI on Design Education

The teaching of design is expected to be significantly impacted by the growing changes in
the professional field of design, which have been driven by technological advances and the
integration of AI. Design instructors will need to modify their courses to ensure students
are given the essential information and skills as AI becomes more widely available in design
tools and processes.
Design educators will be vital in educating students for a future in which AI is an integral
part of the design profession. design programs should encourage students’ ongoing learning
and flexibility. By embracing these developments, design education can prepare students to
gain the benefits of AI while keeping conscious of the field’s ethical, social, and cultural con-
sequences. Furthermore, in the face of AI integration, design instructors must emphasize the
need for critical thinking and human-centered design concepts. While artificial intelligence
can automate certain jobs and provide significant insights, it is critical to maintain a user-
centric approach and ensure that AI tools are used as promoters rather than replacements
for human creativity and empathy.
in conclusion, the evolving professional area of design, as affected by AI breakthroughs,

demands changes in design education. Design education could equip students with the
skills and mentality needed for success in the growing design landscape by incorporating AI-
related topics, facilitating interdisciplinary cooperation, emphasizing human-centered design
principles, and supporting continuous learning.
8.3 Ethical concerns

Data management, processing, and computing are very necessary due to the importance
of information and data in many industries, which are driven by the quick development of
digital technology. Digitalization processes have the potential to alter society, but they must
be carried out correctly. Because of our reliance on digital information and communication
66
technology, we may face new security and privacy threats. New human-machine interfaces
demand ethical and anthropological issues. These changes may also have an influence on em-
ployment, inclusion, and distributive justice, potentially leading to societal unrest. To tackle
these challenges, interdisciplinary approaches that consider social, political, and economic
considerations, as well as technology assessment, systems analysis, innovation research, and
ethics, are required. [24]
An analysis of 12 million images found that, of the 2.3 billion total images used by the
system, nearly half came from 100 domains, including well-known websites like Pinterest,
WordPress-hosted sites, Flickr, DeviantArt, and other photo art and blogging platforms. [16]
These websites rely on user contributions rather than content from a centralized commercial
source like Getty Images or newspaper archives. Even if AI doesn’t directly reproduce a
user’s work, it may still produce photos with outstanding creative aspects, minute details,
and vibrant liveliness that would otherwise be impossible to create without having access to
the user’s work. As a result, assuming users were even aware of the terms, it is likely that they
did not intend for AI to be used in this situation when they set the copyright conditions
for their uploaded photographs. The use of websites like Pinterest, where many of the
images posted may have come from somewhere else, and users’ actions further exacerbate
the situation. [25]
A new tool has been created to assist users in determining whether an AI picture generator
has copied their creative work without their consent, according to a recent article on the Ars
Technica blog. The program is made to look for similarities between AI-generated images
in a database and the original artwork by comparing the two. In circumstances when AI-
generated images have been exploited to copy or alter original works, this can aid artists and
creators in preventing unlawful usage of their creations. It is anticipated that the tool will be
a useful resource for artists and creators who are concerned about the possible infringement
of their intellectual property by AI algorithms [26].
Artists have expressed diverse reactions upon discovering their work on such platforms, and
it is anticipated that future legal actions may arise to address questions around the legality
of these methods and determine ownership of copyrights for artworks produced by these
tools. [27].
This is further complicated by the fact that, unlike Stable Diffusion, other software tech-
nologies do not provide as much transparency regarding the origin of the images utilized to
train their AI tools.
8.4 Method
On the one hand, my approach to this project was advantageous. I was able to iterate and
enhance my designs thanks to a combination of research, brainstorming, prototyping, testing,
and feedback. Including design students, interviewing AI users for design purposes, and UX
experts during interviews resulted in the development of insightful and varied viewpoints.
This method enabled me to produce a prototype that was in line with user requirements.
Since this project aims to understand AI and its impacts on the UX process, using AI tools
like ChatGPT and Fabrie was highly beneficial since they not only helped me better grasp
how AI works in a UX design project but also made the project efficient. I was able to
learn how AI may help with various areas of the design process and increase productivity
by utilizing these technologies. Overall, my strategy enabled an extensive analysis of AI’s
influence on UX design and offered insightful information for further use.
On the other hand, It is worthwhile to mention that due to my lack of strong background
in AI, I had to spend extra time investigating and learning about it during this research.
67
As a result, my experience leads me to recommend that designers who are interested in
this field team up with data scientists or other pertinent professionals to streamline their
procedures. However, this approach may also present some challenges due to the differences
in processes and mindsets. Consequently, a combined process that covers all aspects can be
a valuable solution for both sides that is also resulting in a higher quality outcome. This is
an interesting topic for further research and has been added to the future work chapter.
68
9 Conclusion
In conclusion, this study looked into the incorporation of AI technologies into the UX design
process, with a focus on the potential impacts on efficiency. By utilizing AI tools such as
Midjourney, ChatGPT, Fabrie, and Figma plugins, designers can potentially improve their
productivity, foster creativity, and streamline different stages of the design process. However,
a more tailored application designed specifically for their needs could be even more beneficial,
as it would align with their thought processes and reduce the need for additional work.
Throughout the investigation, it became clear that AI has the potential to function as a
beneficial design assistant, making suggestions, automating some tasks, and providing real-
time feedback. Designers can utilize AI technologies to gather insights, create personas,
establish information architectures, design user flows, and even generate content such as
hero sections and FAQ sections. By using AI in these areas, designers may be able to save
time and effort and concentrate on more strategic and creative aspects of their job.
However, there might be difficulties in incorporating AI into the design process since de-
signers must embrace new technologies by transforming to AI-powered design-supporting
software, learn the fundamentals of AI, and make sure the AI tools they use align with their
specific needs and design mindset. To guarantee the integrity and human-centered approach
of the design process, it is essential to strike a balance between AI support and maintaining
the designer’s knowledge and creative input.
9.1 Knowledge contribution

The exploration of the integration of AI in the UX design process is the study’s knowledge
contribution. This study contributes to a deeper understanding of how AI can enhance the
user experience design process by describing the AIUX app concept, demonstrating the flow
and features of the app, and highlighting the possible benefits and challenges involved in
employing AI in UX design. The discussion highlights AI’s transformative potential in UX
design, emphasizing its ability to supplement designers’ capabilities, enhance efficiency, and
promote user-centric design methods.
The study recognizes AI technology’s rapid growth and its potential to revolutionize the
design landscape, underlining the importance of ongoing research and discovery in this in-
teresting sector. Furthermore, it emphasizes the value of human creativity, engagement, and
invention in conjunction with AI capabilities, underlining the importance of a collaborative
approach between designers and AI systems.
69
10 Future works
10.1 Collaboration of designers and data scientists

Working on this project has motivated me to read more about the collaboration between
designers and data scientists. I have included the findings in this chapter as a suggestion
for future work.
According to Zempekakis and Westra (2019) [5], the collaboration between data scientists
and designers can considerably improve the efficiency and success of the design process
by eliminating blind spots and decreasing miscommunication. This collaborative approach
can also help to mitigate the issues that frequently develop during the design phase. Data
scientists may be more focused on problem resolution without fully considering the needs of
end users, resulting in solutions that may not meet consumer wants. Designers, on the other
hand, may have high expectations for AI capabilities while lacking the technical skills to
analyze the viability and implementation of their proposed solutions. Furthermore, the need
of developing feedback loops to include fresh data over time is sometimes underestimated,
and designers may feel excluded in the final stages of presenting the solution. As a result, a
collaborative strategy that incorporates the skills of both designers and data scientists can
solve these difficulties and lead to the development of more human-centered and effective
AI-powered solutions (Zempekakis Westra, 2019). [5]
Figure 10.1: The gaps between design thinking and AI science process [5]
70
Figure 10.2: AI by design approach which is a combination of design thinking and data
science processes [5]
71
Bibliography
[1] Martyna J. Dietary Choices and Their Impact on the Environment Building a Concept
of an Interactive Virtual Reality Experience to Initiate Reflection [Master’s thesis].
SE-581 83 Linköping, Sweden: Linköping University; 2022.
[2] design council. Framework for Innovation: Design Council’s evolved Double Diamond;
2019.
[3] Propulsé F. UX DESIGN PROCESS; 2020. Available from: http://uxuidesigner.

io/.
[4] Pfeiffer A. Creativity and technology in the age of AI. 2018.
[5] Zempekakis WS I. AI-by-Design A 6-Step Approach for Building Human-Centred AI

Solutions. Creative Commons Attribution-ShareAlike 4.0 International License., Koos
service design OLX group; 2022.
[6] Helm JM, Swiergosz AM, Haeberle HS, Karnuta JM, Schaffer JL, Krebs VE, et al.
Machine Learning and Artificial Intelligence: Definitions, Applications, and Future
Directions. Current Reviews in Musculoskeletal Medicine. 2020;13(1):69-76.
[7] Wärnestål P. Designing AI-powered services. Studentlitteratur AB.; 2022.
[8] Wikipedia. Toyota; 2023. [Online; accessed (08)-May-2023]. Available from:

\url{https://en.wikipedia.org/wiki/Toyota#History}.
[9] forklifts se T. Welcome to the Toyota Material Handling career page;. [Online; accessed
(08)-May-2023]. Available from: \url{https://toyota-forklifts.se/career-en/}.
[10] Foundation ID. What is User Experience (UX) Design?;. [Online; accessed (08)-
May-2023]. Available from: {https://www.interaction-design.org/literature/
topics/ux-design}.
[11] Lidman ADA. Explain UX Design (Or Anything) To Anyone; 2020. Available from:
uxdesign.cc/explain-ux-design-or-anything-to-anyone-62c79fa66d62.
[12] Lee QC K F. The future, What AI cannot do. Bigthink.; 2022.
[13] Aceto VP G, Antonio P. The Role of Information and Communication Technologies

in Healthcare: Taxonomies, Perspectives, and Challenges. Journal of Network and
Computer Applications 107. 2018.
[14] Lember V, Brandsen T, Tõnurist P. The potential impacts of digital technologies on

co-production and co-creation. Public Management Review. 2019.
72
[15] Russell S, Norvig P, Davis E. Artificial Intelligence: A Modern Approach. Pearson;
2019.
[16] Fernandez P. Technology behind text to image generators. Library Hi Tech News. 2022.
[17] Huang Z. Analysis of Text-to-Image AI Generators. IPHS 300: Artificial Intelligence

for the Humanities: Text, Image, and Sound. 2022.
[18] Stappers E P Giaccardi. Research through Design, The Encyclopedia of Human-

Computer Interaction. Interaction Design Foundation; 2017.
[19] Siva Durga Prasad Nayak NKA M. Strengths and weaknesses of online surveys. IOSR
Journal of Humanities and Social Sciences. 2019.
[20] Harrell MC, Bradley MA. Data Collection Methods: Semi-Structured Interviews and
Focus Groups. Santa Monica, CA: RAND Corporation; 2009.
[21] George T. What is a Focus Group — Step-by-Step Guide Examples; 2023.
[22] Daniela K Rosner WLNTYCS Saba Kawas. Out of Time, Out of Place: Reflections on
Design Workshops as a Research Method. CSCW ’16: Proceedings of the 19th ACM
Conference on Computer-Supported Cooperative Work Social Computing. 2016.
[23] Main GM A. Guru, Partner, or Pencil Sharpener? Understanding Designers’ Attitudes

Towards Intelligent Creativity Support Tools. 2020.
[24] Grunwald A. Designing technology in the age of digitalization: needs for technology
assessment and ethics. Multidisciplinary Aspects of Production Engineering. 2019.
[25] ”Baio A. ”Exploring 12 Million of the 2.3 Billion Images Used to Train Stable Diffusion’s
Image Generator”; ”2022”.
[26] EDWARDS B. Have AI image generators assimilated your art? New tool lets you check/
New search engine combs through harvested images used to train Stable Diffusion,
others. 2022.
[27] Gershgorn D. GitHub’s automatic coding tool rests on untested legal ground. 2021.
73

FULLTEXT01

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

FULLTEXT01

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

FULLTEXT01

Uploaded by

Copyright:

Available Formats

Linköping University - Department of Management and Engineering

Master’s thesis, 120 credits - Master’s programme

Enhancing User Experience

2.1 Generated image by Midjourney . . . . . . . . . . . . . . . . . . . . . . . . . 10

4.1 List of text to image generators . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.1 Mind mapping after adjusting Fabrie’s version . . . . . . . . . . . . . . . . . . 48

7.1 Expert evaluation of the Figma prototype . . . . . . . . . . . . . . . . . . . . 64

1.3 Aim and goals

1.4 Toyota material handlling

1.5 User experience design

1.6 Research question

-A Figma prototype illustrating the idea

2.1 User experience design

Figure 2.1: Generated image by Midjourney

2.3 Artificial intelligence in design practice

Gaining an understanding of machine learning can aid us in navigating the complexities,

2.5 Text to image generators

At a more specific level, the utilization of diffusion-based techniques in text-to-image gen-

3.1.1 Research through design

3.1.2 Design framework

3.1.3 Secondary research

3.1.4 Web based survey

To accomplish the goal of this project, I conducted a series of semi-structured interviews

3.1.6 Focus group

4.1.1 Exploring current AI tools in design

Figure 4.1: List of text to image generators

• Generating low-fidelity prototypes

• Creating variations of a design for A/B testing

• Automating the production of assets like icons and illustrations

• Predicting the heatmap of the user’s interaction

• Analyzing how users interact with an interface and identifying potential UX

• Producing inspiration for designers to build upon

• Translating hand-drawn wireframes into polished user interfaces

• Generating interactive prototypes that designers can build upon

• Automating the production of design documentation like style guides, pattern

• Generating personalized design feedback for individual designers based on their

• Collaborating with designers in real-time through conversational interfaces to

4.1.2 User experience design process

4.1.3 Focus group session with design students

• Academic background: Media and communication science

• Profession: Senior UX designer

• Location: Gothenburg, Sweden

• What software do you use? We use only Figma.

• Do you think AI can be used in a professional way in UX design? Yes, of course.

• Academic background: Digital creative

• Profession: Visual designer

• Location: Stockholm, Sweden

In my experience, AI can be a valuable source of inspiration and aid in designing

• Academic background: Design for play

• Profession: Creative designer

• Location: Arhus, Denmark

• Have you used AI tools professionally? Yes

• What kind of AI tools do you use? Please identify a couple of them.

Midjourney - Dall e- Stylegan

• What kinds of changes have they experienced up to this point?

In my opinion, Midjourney is most powerful in the product design field. It can

4.2.1 Web based survey

4.2.2 Exploring Designers’ Attitudes towards Artificial Intelligence