Conference PaperPDF Available

Design After the Rise of AI-Driven Services: Learning from Literature Review


Abstract and Figures

Technology is playing a significant role in shaping the future of design. We are moving fast into a digital era where Artificial Intelligence, Machine Learning, Deep Learning, Big Data, the Internet of Things, Blockchain, Spatial Computing, and several other technologies are becoming part of the designers' lexicon. The designers' roles are evolving, and the touchpoints they need to consider are growing in complexity. Integrating AI developments with User-centered Design and User Experience Design is becoming a challenging task. Motivated by this, we reviewed the literature to understand how AI is shaping the way designers think about their process and how they design for AI artifacts. This research was important for evidencing how designers are adapting their mindsets, skills and knowledge to address these new technological possibilities. The overall search was conducted on ACM Digital Library, Google Scholar, and Springer, for publications about design and AI. The work identifies promising research clusters in the crossroads of Intelligent Systems, Human-Computer Interaction and Design, but few studies were found with concrete guidance on how to design for AI-driven services. By mapping the current literature on AI and Design, we contribute to a broad understanding of how current design methods need to adapt when interactions become living ecosystems and represent complex trade-offs to the designers. Therefore, new principles for human-AI interaction are becoming urgent to study.
Content may be subject to copyright.
367DIGICOM 3rd International Conference on Digital Design & Communication
Artificial Intelligence;
Design; Interaction
Design; Service Design;
Joana Cerejo1; Miguel Carvalhais2
1 Faculty of Engineering
of the University of Porto,;
2 INESC TEC, Faculty of Fine
Arts of the University of Porto,
Technology is playing a signicant role in shaping the future of de-
sign. We are moving fast into a digital era where Articial Intelligence,
Machine Learning, Deep Learning, Big Data, the Internet of Things,
Blockchain, Spatial Computing, and several other technologies are be-
coming part of the designers’ lexicon. The designers’ roles are evolving,
and the touchpoints they need to consider are growing in complexi-
ty. Integrating AI developments with User-centered Design and User
Experience Design is becoming a challenging task. Motivated by this,
we reviewed the literature to understand how AI is shaping the way
designers think about their process and how they design for AI arti-
facts. This research was important for evidencing how designers are
adapting their mindsets, skills and knowledge to address these new
technological possibilities. The overall search was conducted on ACM
Digital Library, Google Scholar, and Springer, for publications about
design and AI. The work identies promising research clusters in the
crossroads of Intelligent Systems, Human-Computer Interaction and
Design, but few studies were found with concrete guidance on how to
design for AI-driven services. By mapping the current literature on AI
and Design, we contribute to a broad understanding of how current
design methods need to adapt when interactions become living eco-
systems and represent complex trade-offs to the designers. Therefore,
new principles for human-AI interaction are becoming urgent to study.
368 Portugal 2019
1. Introduction and Background: The age of AI
New conventions in experience economy1 alongside new technologi-
cal developments as Articial Intelligence (AI) are shaping how busi-
nesses invest in integrating user-centered design (UCD) and user ex-
perience (UX) as a crucial part of their whole service design strategy
plan. AI, machine learning (ML) and deep learning (DL)2 will be the
most important means to improve UX (Yang, 2017). Businesses are
chasing AI transformation either to enhance customer experience or
to automate businesses and dening a system that extends human
capabilities (Daugherty & Wilson, 2018).
Among the paradigms for AI services are UCD and UX. In the mid-
1980s, two publications introduced the user-centered design (UCD)
methodology (Gould & Lewis, 1985; Norman, 1986). In essence, UCD is
a design approach that consists in taking the user perspective in all
stages of product development. The user is part of the testing and eval-
uation process, providing relevant considerations for the nal prod-
uct development. With today’s new technological possibilities, UCD is
moving from a traditional static convention process to an emergent
living ecosystem. By changing the design paradigm, is becoming hard-
er to foresee the effects of our designer’s solutions because the center
of the problem is now an organic and unpredictable evolving system.
The center of design is now an organic and unpredictable intelligent
evolving system (Dove et al. 2017; Fischer, 2002; Van Allen, 2017). A sys-
tem that is designed to constantly adapting to context possibilities,
which produces different effects on user behavior according to the
situation. For example, an AI service that meets this circumstance is
Waze. Waze is a mobile application that combines AI algorithms and
real-time data to create living, dynamic, optimized maps that allows
peoples’ to get to their destinations as quickly as possible (Daugherty &
Wilson, 2018, pp. 6–7). Design is evolving into a new context,
1 “As services, like goods before them, increasingly become commoditized
experiences have emerged as the next step in what we call the progression of
economic value.” (Pine II, Joseph B.; Gilmore 1998)
2 Technically, DL is a subset of ML as ML if from AI. But their capabilities are
different, meaning a different impact on how designers may design services
for each type of technology. Basic ML models do become progressively better
at whatever their function is, but they still some guidance. If a ML algorithm
returns an inaccurate prediction, then an engineer needs to step in and make
adjustments. But with a DL model, the algorithms can determine on their
own if a prediction is accurate or not. (Grossfeld 2017)
369DIGICOM 3rd International Conference on Digital Design & Communication
An evolving, negotiated, inconsistent, improvised, serendipitous inter-
action that does not easily resolve to task accomplishments, efciency,
certainty, ROI, customer expectations, or for that matter, one user’s ex-
perience. (Van Allen 2017, 431).
The International Organization for Standardization (ISO) in the ISO
9241 norm, titled Ergonomics of Human-System Interaction, and in par-
ticular in its part 210, Human-Centered Design for Interactive Systems;
dened UX as a “person’s perceptions and responses resulting from
the use and/or anticipated use of a product, system or service.” These
perceptions and responses are very broad and include “emotions, be-
liefs, preferences, physical and psychological perceptions, color re-
sponses, behaviors and achievements that occur before, during, and
after use of the system, product, or service.” This denition is in line
with our analysis. With the rise of AI-driven services, how are design-
ers taking into their process all these dimensions when designing for
autonomous services?
Design with AI pursues has the goal of enhancing the human ex-
perience by extending their capability. Each eld gathers data to in-
terpret and predict human behavior and to anticipate what people
might do next. But they understand the human signicantly differ-
ent. However, both elds have the power to shape each other. To de-
sign for AI-driven services, designers will need to domain the context
awareness and personalized customization enabled by AI and its sub-
domains; and AI needs UXD to be perceivably valuable to the users. For
achieving social responsibility, designers will have to learn how to de-
sign AI-driven services that address issues of fairness, accountability,
interpretability, and transparency (Riedl, 2019).
1.1. New Problems and Perspectives
Design is in a rapidly changing landscape of opportunities. One of
AI’s affordances is the reduction in the number of users’ interactions
and decisions within a system. Does this automation, and decrease of
qualitative interactions, affects the users’ perception of quality? These
services are being designed to decide and performing tasks on behalf
of users. And the success or failure of a digital service is intrinsically
correlated to how users perceive the systems’ qualities (Hassenzahl,
2008). AI services are able to reduce users’ cognitive overload by
370 Portugal 2019
facilitating their decision-making processes. Google service Nest3 ex-
emplies this approach.
2. Methodology
This paper presents a literature review. Wwe aim to identify dimen-
sions that have an impact on how designers design for and within AI,
to address its new technological possibilities.
2.1. Research Question and Search Terms
The guiding questions of this research aimed to select studies that
explore the conuence of AI or ML with UCD, and that consider UX:
1. How is AI shaping the way designers think about their design pro-
cess? This question is concerned with identifying all possible di-
mension where AI is shaping the way they think about design,
and design for AI. We need to understand how designing for AI is
changing the designer’s mindsets, skills and knowledge to address
these new technological possibilities.
2. How does the integration among AI technologies with UCD or UX oc-
cur? This question is concerned with identifying the aspects that
affect both elds of knowledge and development. Identifying these
aspects will help us identify how designers can engage with AI also
as a design material.
Derived from the research questions we identied the following
search terms:
Research Question Extracted Search Terms
RQ 1 Design, Decision-making, User-Centered Design, Design
Process, Service Design, Human-Centered Design
RQ 2 Artificial Intelligence, Machine Learning, User Experience,
Human-centered Computing, Interaction Design, AI-driven
3 Nest is an AI service. Is a self-learning thermostat that is programmable to optimize
homes and business. Nest can learn people’s schedule, and at which temperature they
are used to and when, shifting into energy-saving mode when it realizes nobody is at
home to conserve energy.
Table 1 Research Questions
and the extracted keywords
371DIGICOM 3rd International Conference on Digital Design & Communication
2.2. Eligibility Criteria and Search
The eligibility criteria were: 1) Papers that specify at least Articial
Intelligence or machine Learning regarding design methodologies
and concerns; 2) The study needed to consider design for an AI service
or AI as a design material to designers.
˅Phase 1: We follow the PICO strategy (Santos, Pimenta, & Nobre,
2007). The search terms were dened as follows: ((“Articial
Intelligence” OR “Machine Learning”) AND (Design OR UX OR
“user-centered design” OR “Interaction Design”)).
˅We selected three main digital databases: Springer, Association
for Computing Machinery (ACM), and Google Scholar. The search-
es were carried out in the period from 2017 to August-2019. Books,
journals and conference proceedings were considered with this
time frame of three years and a publication language restriction
of English.
˅Phase 2: For each database the terms were used to lter publi-
cations titles, abstract and keywords, to verifying compliance
with the established eligibility criteria;
˅Phase 3: The search generated multiple sets of publications clus-
ters. These results were not particularly informative or insight-
ful in terms of revealing solutions for the research questions.
However, it was useful to understand the tendencies of keyword
clustering among the elds of expertise. The inclusion criteria
were if the publication answers one of the RQs, either partially or
entirety. Some publications were excluded from Google Scholar
results because they were duplicated from ACM or Springer
search results.
2.3. Results & Discussion
With the support of the publication obtained in the research, it was
possible to analyze the research questions.
Table 2 Numbers
of publications found
in each database
ACM Google Scholar Springer Total
2017 2018 2019 Total 2017 2018 2019 Total 2017 2018 2019 Total
RQ 1 4 10 13 27 16 16 8 39 5 7 13 25 92
RQ 2 10 18 18 46 3 11 4 18 2 6 9 17 81
Total 14 28 31 73 19 27 12 58 7 13 22 42 173
Excluded 0 Excluded 28 Excluded 0
372 Portugal 2019
Question: How is AI shaping the way designers think about
their design process?
Answer: Seeking for ubiquity on design. Design is a human fab-
rication. Almost everything in the environment around us is designed
which make us in part a product of design (Fry, 2015). Designers have
to learn how to design AI-driven services that address issues of fair-
ness, accountability, interpretability, and transparency (Riedl, 2019).
Design has spread across several domains and AI is following a sim-
ilar path, which makes them both ubiquitous in many senses. In the
beginnings of UCD, users started to be invited into the world of the
designer through participatory design4 methods (van der Bijl-Brouwer
& Dorst, 2017). With the conuence of design with AI maybe we should
produce processes were data scientists learn how to summon design-
ers into their processes. However, often designers join AI development
after the functional decisions have been made which rise another is-
sue, “UX cannot be an afterthought of AI.” (Yang et al. 2018, p. 469).
In this perspective, integrating UCD or UX into AI-driven services be-
comes a great challenge for designers when designing for anticipa-
tion actions in living ecosystems.
Answer: Investing on a new educational system. An AIGA’s (2018)
report, a professional organization for design, mentions that a big
portion of today’s designers still focus on an object-driven process that
addressed one independent physical constraint at a time. “Current ed-
ucational systems around the world continue to focus on teaching our
students to undertake tasks for which machines are now better suited”
(Susskind & Susskind, 2015, p. xi). Institutions are struggling for edu-
cating for the future (Fry, 2015). We question if designers are seeking
to be prepared to explore and exploit the integration of experience and
interaction design with AI? Which will be the designer’s professional
demands of the future? Are we preparing young designers for long
and productive careers? We need a new educational agenda for UXD
eld that is able to exploit AI domains. Given the potential of AI to im-
pact our lives, building intelligent algorithms without a purpose will
not serve us properly. “There is a growing awareness that algorith-
mic advances to articial intelligence and machine learning alone
are insufcient when considering systems designed to interact with
and around humans” (Riedl, 2019, p. 2). To understand and master
4 Scandinavian participatory design movement of the 1990s.
373DIGICOM 3rd International Conference on Digital Design & Communication
this complexity, designers will have to understand the technologies
behind it (Hebron, 2016). With the proviso that understanding AI is
not the same as learning it from a technological point of view. As a
result, several authors propose that current design students should be
prepared with additional skill sets like basic interpretation – Human-
centered Articial Intelligence (Riedl, 2019) – understanding and ma-
nipulation, or either one, of quantitative data (Girardin & Lathia, 2017;
Yang et al. 2018) – Co-designing with data (Dove et al. 2017).
Answer: Demanding for new set of skills. As more services are
built with AI, it becomes clear that designers still have a lot to learn
about how to make users feel in control of the technology (Dove et al.
2017). “Designers face challenges in understanding ML capabilities, in
envisioning new products and services, and in collaborating effective-
ly with data scientists” (Yang et al. 2018, p. 591). The author express
that designers don’t think that learning more about AI will make
them better designers. This could be a fallacy. Because they perceived
they don’t need to learn about AI to be a better designer when actually
they need to. AI might not achieve the productivity gains expected, be-
cause designers do not understand how to exploit it into their projects.
Almost all designers do not know how to bring their UXD expertise to
bear on AI (Dove et al. 2017). Now, designers do not possess enough
inferred knowledge of AI to operationalize interaction ows that pro-
actively adapt or evolve over time to this new organic leaving state
(Yang, 2017). Furthermore,
It is no longer enough for UX designers to only improve experience by
paying attention to usability, utility, and interaction aesthetics. Instead,
the best UX may come from services that automatically personalize
their offers to the user (...) that leverage more detailed understanding
of people (Dove et al. 2017, p. 278).
The consequence is a disconnection between design practices and
innovation. Nevertheless, there is a lack of new resources and tools
that may allow designers to move into a more effective engagement
approach towards AI as a design material or as a service design. But
can designers work with a material they do not fully understand?
Answer: Requesting a shifting paradigm with design process.
Powered by AI developments, the current paradigm of design offers an
exciting and challenging stimulus for innovation.
374 Portugal 2019
We are shifting from …
Formulating research questions
based on user’s insights:
˅ Surveys
˅ Interviews
˅ Focus Groups
˅ Direct Observation
To a hybrid approach: Users’ insights
as well as formulating research
questions that arise from the
current availability of data:
˅ Data Analysis
˅ Business Intelligence
˅ Pattern Recognition
We may consider an evolution on the HCD eld by moving into
coexistence with the formal method of putting human beings in the
center of the design process as well as putting data. Also, shifting from
formulating research questions based on users’ needs and insights,
towards formulating research questions that arise from data and pat-
terns recognition – “smart hypothesis” ( Daugherty & Wilson, 2018, p.
72). Which raises an important question. What will happen to the sci-
entic process and to design problem framing process when hypoth-
eses can be generated automatically? Now, designers have the advan-
tage of not only being able to formulate hypotheses from the users’
perspective, but also to formulate hypotheses from big data. Designers
can start co-designing with data (Dove, 2015). The author proposes a
new collaborative approach to seek insights from data through design
workshops in which working with domain-relevant data is the key
distinguishing feature. He helps designers to gain an understanding
of the context these data might come from, and to inspire creative
design ideas. Creating a synergy between the two methods to achieve
an increment on the efciency of AI-driven services and establishing
a new discipline that moves between humans and data.
Question: How does the integration among AI technologies with
UCD or UX occur?
Answer: Requiring new principles for human-AI interaction.
Design problems require several knowledge domains and a broad
range of skills and they are one of the most complex problems to
tackle within AI (Grecu & Brown, 1998). It is time to rethink the
standards of interaction design inside AI and IoT (Marenko & van
Allen, 2016). First, we need to raise awareness among designers to
the importance and relevance of educating themselves to new tech-
nological practices and domains such as AI, ML and DL. And second,
prepare an ecosystem where designers can ground fundamental
knowledge about AI and their subdomains within mind best prac-
tices for designing with and for it. Notwithstanding, a few steps al-
ready were made in this sense.
Table 3 Shifting paradigm
with design process
(Girardin & Lathia, 2017)
375DIGICOM 3rd International Conference on Digital Design & Communication
The University of Helsinki5, developed a pilot project to educate the
general population to AI awareness. The program is a free course with
the goal to help people to be empowered, not threatened, by articial
intelligence (AI). They built Elements of AI6, a non-technical course to
teach the basics of AI to people from a wide range of backgrounds. In the
spring of 2018, they launch the rst part of the course with the intuition
to introduce people to AI concepts. For the end of 2019, they are prepar-
ing to launch the second phase of the program focused on Building AI.
Another course was developed by Deeplearning.ai7, also a free
non-technical course, AI for Everyone8, that unleash an understanding
of AI technologies by teaching how to spot opportunities to apply AI to
problems in people’s own organizations. Both courses were aimed at
the general population. A similar approach, oriented to design led and
designers, could be a strategy to literate designers to AI and orient them
to a new set of skills for designing AI-driven services and products.
3. Conclusion
Without proper methods and prototyping tools, it becomes hard to
successfully prototype for interactions that may follow an unpredict-
able course (Dove et al. 2017). AI are “not yet a standard part of UX
design practice, in either design patterns, prototyping tools, or educa-
tion” (Yang, 2017). There is an opening door for designers to collaborate
with data scientists (and other stakeholders) to innovate and co-cre-
ate exciting meaningful experiences that will benet users and the
future of the interaction design.
A possible method to address UXD within AI is by approaching
Animistic Design, a method for “fostering affects, sensibilities and
thoughts that capitalize on the uncertain, the unpredictable and the
nonlinear, and their capacity to trigger creative pathways” (Marenko
5 The course was designed by University of Helsinki and Reaktor. The lead instructor
of the course is Associate Professor Teemu Roos with industry insights from Hanna
Hagström. The course is a part of the AI Education program of the Finnish Center for AI,
and offered in cooperation with The Open University, and Mooc..
6 ]
7 Is an organization to promote the world-class AI education accessible to people
around the globe so that we can all benet from an AI-powered future https://www.
376 Portugal 2019
& van Allen, 2016, p. 432). This may become a way to reimagine digi-
tal interaction between the human and nonhuman. The method can
help the design shifting from crafting task-oriented experiences for
users, to building evolving, diverse, autonomous ecologies that sup-
port collaborative exploration and creativity for machine and human
participants alike.
Another possible method to address the synergy between design
and AI is by approaching Anticipatory Design a design method that
personalizes the user ow by making and eliminating user choices
(Shapiro, 2015; Van Bodegraven, 2017). This design pattern has the deter-
mination of predicting UX. The premise behind it is to reduce users’ cog-
nitive overload by facilitating their decision-making process on behalf
of them. This new design method moves around three major concepts:
If one of these three actors fail, we cannot design for anticipatory
design. “The goal is not to help the user make a decision, but to create
an ecosystem where a decision is never made” (Shapiro, 2015). In sum,
designers lack prototyping tools for working with AI (Dove et al. 2017).
Without good tools for tools to prototyping unpredictable intelligent evolv-
ing courses that support Anticipatory Design and Animistic Design princi-
ples and methods, designers struggle to explore the space of possibilities.
It is vital to comprehend how will designers design for unpredicta-
ble courses, and how they are preparing themselves to deal with data
to design responsible AI solutions. Aiming to contribute to the eld in
that sense, we already started to map the biggest challenges inherent
to AI and design. This will allow us to initially understand how design
methods need to adapt when interactions become living ecosystems
and represent complex trade-offs to the designer. This will lead us to
a major area of inquiry: How to do anticipatory experience design for
AI-driven services?
Fig.1 Anticipatory Design
structure retrieved from
Van Bodegraven (2017).
377DIGICOM 3rd International Conference on Digital Design & Communication
Daugherty, P., & Wilson, J. (2018). Human
+ Machine: Reimagining Work in
the Age of AI. United States: Harvard
Business Review Press.
Dorst, K. (2015). Frame Creation and
Design in the Expanded Field. She Ji:
The Journal of Design, Economics, and
Innovation, 1(1), 22–33. https://doi.
Dove, G. (2015). CoDesign with data.
Retrieved from
Dove, Graham, Halskov, K., Forlizzi, J.,
& Zimmerman, J. (2017). UX Design
Innovation. In Proceedings of the
2017 CHI Conference on Human
Factors in Computing Systems —
CHI ’17 (pp. 278–288). https://doi.
Dubberly, Hugh; Pangaro, P. (2018).
Making sense in the data economy
Making sense in the data economy.
Retrieved from https://www.
Fischer, G. (2002). Beyond “Couch
Potatoes”: From consumers to
designers and active contributors. First
Monday, 7(12).
Fry, T. (2015). Whither Design / Whether
History. In Design and the Question of
History (1st ed., p. 320). Bloomsbury
Girardin, F., & Lathia, N. (2017). When
User Experience Designers Partner
with Data Scientists. In Aaai 2017
Spring Symposia (pp. 376–381).
Gould, J. D., & Lewis, C. (1985, March).
Designing for usability: key
principles and what designers
think. Communications of the
ACM, 28(3), 300–311. https://doi.
Grecu, D., & Brown, D. (1998). Dimensions
of machine learning in design.
Articial Intelligence for Engineering
Design, Analysis and Manufacturing,
(April), 117–122.
Grossfeld, B. (2017). A simple way
to understand machine learning
vs deep learning. Retrieved from
Hassenzahl, M. (2008). User experience
(UX): Towards an experiential
perspective on product quality. ACM
International Conference Proceeding
Series, (September 2008), 11–15. https://
Hebron, P. (2016). Machine Learning for
Designers. (A. Runo, Ed.) (1st ed.).
O’Reilly Media, Inc.
Marenko, B., & van Allen, P. (2016).
Animistic design: how to reimagine
digital interaction between the human
and the nonhuman. Digital Creativity,
27(1), 52–70.
Norman, D. (1986). The design of everyday
things. New York, USA: Basic books.
Riedl, M. O. (2019). Human-Centered
Articial Intelligence and Machine
Learning. Human Behavior and
Emerging Technologies, Volume 1(Issue
1), 1–8. Retrieved from
Santos, C. M. D. C., Pimenta, C. A. D.
M., & Nobre, M. R. C. (2007). The
PICO strategy for the research
question construction and evidence
search. Revista Latino-Americana de
Enfermagem, 15(3), 508–511. https://doi.
Schwartz, B. (2004). The Paradox
of Choice, Why More Is Less.
Retrieved from
Shapiro, A. (2015). The Next Big Thing In
Design? Less Choice. Retrieved April 22,
2018, from https://www.fastcompany.
Susskind, R., & Susskind, D. (2015).
The Future of the Professions: How
Technology Will Transform the Work
of Human Experts. new York: Oxford
University Press.
Van Allen, P. (2017). Reimagining the
Goals and Methods of UX for ML/AI A
new context requires new approaches.
In The AAAI 2017 Spring Symposium
on Designing the User Experience
of Machine Learning Systems (pp.
Van Bodegraven, J. (2017). How
Anticipatory Design Will Challenge
Our Relationship with Technology A
Future Without Choice. In The AAAI 2017
Spring Symposium on Designing the
User Experience of Machine Learning
Systems (pp. 435–438).
van der Bijl-Brouwer, M., & Dorst,
K. (2017). Advancing the strategic
impact of human-centred design.
Design Studies, 53, 1–23. https://doi.
Yang, Q. (2017). The Role of Design in
Creating Machine-Learning-Enhanced
User Experience. The AAAI 2017 Spring
Symposium on Designing the User
Experience of Machine Learning
Systems Technical Report SS-17-04,
406–411. Retrieved from http://aaai.
Yang, Q., Banovic, N., & Zimmerman,
J. (2018). Mapping Machine Learning
Advances from HCI Research to
Reveal Starting Places for Design
Innovation. In Proceedings of the 2018
CHI Conference on Human Factors in
Computing Systems — CHI ’18. https://
Yang, Q., Scuito, A., Zimmerman, J.,
Forlizzi, J., & Steinfeld, A. (2018).
Investigating How Experienced UX
Designers Effectively Work with
Machine Learning. In Proceedings
of the 2018 on Designing Interactive
Systems Conference 2018 — DIS
’18 (pp. 585–596). https://doi.
... The prototyping process of ML is distinct from that of traditional technologies because the two types of technology are essentially different. For traditional technologies like electronics and the Internet of Things, designers usually build the prototype after the initial design proposal has been determined (Cerejo and Carvalhais, 2019). This is because these technologies are based on predetermined rules, and their capabilities are predictable. ...
... Some designers might only have a rough idea about how ML operates, and they tend to treat ML as a completely black box system (Patel et al., 2008). They barely know the capabilities, working mechanism, limitations, and other characteristics of ML (Cerejo and Carvalhais, 2019), which hinders them from using ML as a design material (Yang, 2018) for making functional ML prototypes. ...
Machine learning (ML) is increasingly used to enhance intelligent products in the field of product design. However, ML has a never-ending lifecycle in which its capabilities and technical properties iteratively change as new annotated data are utilized. The never-ending lifecycle of ML (which includes data annotation, model training, and other steps) has led to challenges to the prototyping of ML-enhanced products and requires a high level of ML literacy in designers. To facilitate the prototyping of ML-enhanced products and improve the ML literacy of designers, we draw inspiration from a design method called Material Lifecycle Thinking (MLT), which regards ML as a continuously evolving design material. Based on the MLT, we proposed a cyclical prototype workflow and developed inML Kit, a toolkit enabling designers to make functional ML prototypes and improve ML literacy by involving them in the never-ending ML lifecycle. The toolkit was designed, iterated, and implemented through the participatory design process with experienced designers in this field. We evaluated inML Kit by conducting a controlled user study where our toolkit was compared with Google AIY. The evaluation results imply that our inML Kit helps designers to make functional ML prototypes while improving their ML literacy.
Conference Paper
Full-text available
Machine learning (ML) plays an increasingly important role in improving a user’s experience. However, most UX practitioners face challenges in understanding ML’s capabilities or envisioning what it might be. We interviewed 13 designers who had many years of experience designing the UX of ML-enhanced products and services. We probed them to characterize their practices. They shared they do not view themselves as ML experts, nor do they think learning more about ML would make them better designers. Instead, our participants appeared to be the most successful when they engaged in ongoing collaboration with data scientists to help envision what to make and when they embraced a data-centric culture. We discuss the implications of these findings in terms of UX education and as opportunities for additional design research in support of UX designers working with ML.
Conference Paper
Full-text available
HCI has become particularly interested in using machine learning (ML) to improve user experience (UX). However, some design researchers claim that there is a lack of design innovation in envisioning how ML might improve UX. We investigate this claim by analyzing 2,494 related HCI research publications. Our review confirmed a lack of research integrating UX and ML. To help span this gap, we mined our corpus to generate a topic landscape, mapping out 7 clusters of ML technical capabilities within HCI. Among them we identified 3 under-explored clusters that design researchers can dig in and create sensitizing concepts for. To help operationalize these technical design materials, our analysis then identified value channels through which the technical capabilities can provide value for users: self, context, optimal, and utility-capability. The clusters and the value channels collectively mark starting places for envisioning new ways for ML technology to improve people's lives.
Conference Paper
Full-text available
Machine learning (ML) applications that directly interface with everyday users are now increasingly pervasive and powerful. However, user experience (UX) practitioners are lagging behind in leveraging this increasing common technology. ML is not yet a standard part of UX design practice, in either design patterns, prototyping tools, or education. This paper is a reflection on my experience designing ML-mediated UX. I illustrate the role UX practice can play in making machine intelligence usable and valuable for everyday users: it can help identify 1) how to choose the right ML applications. 2) how to design the ML right. The separation of the two concerns is a first step to untangling the tight interplay between ML and UX. I highlight the unique challenges and the implications for future research directions.
Full-text available
This article puts forward the notion of animistic design as an uncertainty-driven strategy to reimagine human–machine interaction as a milieu of human and nonhuman. Animistic design is suggested as capable of fostering affects, sensibilities and thoughts that capitalize on the uncertain, the unpredictable and the nonlinear, and their capacity to trigger creative pathways. Informed by post-human philosophies, theories of mediation and materiality, as well as by affect, agency and aesthesia, animistic design eschews the anthropomorphic and the cute playfulness often associated with animism. Instead, it proposes a practical–theoretical framework to articulate the nexus of digital innovation, interaction design practices, technical materialities and affective responses already emerging in the digital cohabitation of the human and the nonhuman. Using a ‘research through making’ approach, the article describes in detail a series of animistic design experiments and prototyping methods that explore ways of rethinking interaction as an open-ended and creative enterprise. Animistic design offers an investigative strategy that exploits degrees of collaboratively curated uncertainty and unpredictability to imagine forms of digital interaction, and to engender creative human–nonhuman relationships within a given digital milieu.
Full-text available
Design-trained people have access to a very broad range of professions. Yet there is something paradoxical about this development: ostensibly, many of these highly successful people have moved out of the field of “design.” This phenomenon deserves deeper consideration: how do design practices spread across society? What key design practices are particularly relevant to the problems of today’s society? Should what these people do still be considered design? To answer these questions, first we need to understand various ways that practices can be adopted and adapted from one discipline to the other. Problem framing emerges as a key design practice that can be adopted and adapted to other fields, and one which provides a valuable alternative to conventional types of problem solving. An example will illustrate how this frame creation allows practitioners to approach today’s open, complex, dynamic, networked problems in new and fruitful ways. The paper goes on to argue that the practice of frame creation is still part and parcel of the domain of design, and explores how design can develop into an expanded field of practice.
Humans are increasingly coming into contact with artificial intelligence (AI) and machine learning (ML) systems. Human‐centered AI is a perspective on AI and ML that algorithms must be designed with awareness that they are part of a larger system consisting of humans. We lay forth an argument that human‐centered AI can be broken down into two aspects: (a) AI systems that understand humans from a sociocultural perspective, and (b) AI systems that help humans understand them. We further argue that issues of social responsibility such as fairness, accountability, interpretability, and transparency.
Many organisations realise that becoming more human-centred is key to dealing with today's innovation challenges. Human-centred design (HCD) has potential to contribute to this goal. However, its current impact on strategic innovation is limited. In this paper we describe the evolution of HCD methods to date, and the challenges and opportunities of applying HCD in strategic innovation. We show that these challenges could be addressed by augmenting HCD with methods from the field of design innovation. To do this, we propose the NADI-model that links these two worlds by considering the different layers of practices and knowledge they contain, and show how the deepest level of this model can bridge human-centred design and strategic innovation.
Conference Paper
Machine learning (ML) is now a fairly established technology, and user experience (UX) designers appear regularly to integrate ML services in new apps, devices, and systems. Interestingly, this technology has not experienced a wealth of design innovation that other technologies have, and this might be because it is a new and difficult design material. To better understand why we have witnessed little design innovation, we conducted a survey of current UX practitioners with regards to how new ML services are envisioned and developed in UX practice. Our survey probed on how ML may or may not have been a part of their UX design education, on how they work to create new things with developers, and on the challenges they have faced working with this material. We use the findings from this survey and our review of related literature to present a series of challenges for UX and interaction design research and education. Finally, we discuss areas where new research and new curriculum might help our community unlock the power of design thinking to re-imagine what ML might be and might do.