Petter Bae Brandtzaeg and Asbjørn Følstad, SINTEF, Norway

Chatbots:
Changing
User Needs
and
Motivations

C
Insights Chatbots have been around for in spite of these drivers, current
→→ Chatbot initiatives often decades. However, the real buzz chatbot applications suggest that
aim for poor use cases, around this technology did not start conversational user interfaces still
ignoring user needs and until the spring of 2016. Reasons face substantial challenges, generally
user experiences. for the sudden renewed interest in speaking, as well as for the field of
→→ The potential chatbots include massive advances human-computer interaction (HCI).
unpredictable variation in artificial intelligence (AI) and a Chatbots imply not only a change
in user input and what major usage shift from online social in the interface between users and
constitutes a valid networks to mobile-messaging technology; they also imply changing
chatbot response are big applications such as Facebook user dynamics and patterns of use.
challenges for HCI. Messenger, Telegram, Slack, Kik, and We have previously outlined
→→ The expansion in chatbot Viber. The first of these reasons holds the potential implications and
I M A G E B Y A L I C I A K U B I S TA

use is revealing new user promise that intelligent chatbots opportunities that chatbots hold
needs and motivations. may well be within reach. The for the field of HCI [1]. In this
second concerns service providers’ article, we dig into what we see
need to reach users in the context as a key challenge with chatbots
of mobile messaging. However, from a user-centered perspective.

38 INTER ACTIONS S E P T E M B E R – O C T O B E R 2 0 18 INTER ACTIONS. ACM.ORG

credit tk

INTER ACTIONS. ACM.ORG S E P T E M B E R – O C T O B E R 2 0 18 INTER ACTIONS 39

 Developers and designers have an “more than 50 percent of enterprises
urgent need to know more about will spend more per annum on bots
how people experience chatbots and and chatbot creation than traditional
to understand the user needs that mobile app development” by 2021 [2].
motivate the future use of chatbots.

T
We therefore need to ask how CURRENT CHATBOTS
chatbots resonate with user needs and OFTEN FAIL
desires and, in turn, how these same The vision of a compelling
needs and desires evolve as users get conversational interface is not
more experience with chatbots. easily attainable. We know that a
The word chatbots is a derivative key success factor for chatbots and
of chat robots, understood as machine natural language user interfaces
agents that serve as natural language is how well they can support user
user interfaces to data and services needs in the conversational process
through text or voice. Chatbots seamlessly and efficiently. However,
allow users to ask questions or while chatbots may potentially

Advertise with ACM! make commands in their everyday

language and to get the needed
increase individual flexibility,
expand opportunities for information
content or service in a conversational retrieval and learning, and
Reach the innovators style. If chatbots gain the expected compensate for limitations in digital
popular uptake, the technology will competence, current chatbots often
and thought leaders dramatically change the way in which fail [3]. One possible explanation is
working at the people interact with data and services the difficulty in designing for open-
online. It has been predicted that ended conversations—chatbot
cutting edge such a conversational disruption may capabilities seldom capture all
of computing reduce service providers’ motivation the various ways in which the user
to invest in apps and instead might want to engage. There has
and information prioritize chatbots as a channel also been a substantial technology
technology through for reaching out to users. As one push in chatbot development,
messaging platform provider, Kik, potentially leaving users frustrated
ACM’s magazines, claims on its developer site: “First by a perceived lack of concern for
there were websites, then there were how people use chatbots and for what
websites
apps. Now there are bots.” purpose. Organizations are rushing
and newsletters. On average, people are spending into the space, vying to be the first
more time on messaging platforms. to deploy chatbots in their particular
Facebook Messenger, for example, is service domain. In this early phase
◊◆◊◆◊ reported to have had more than 1.2 of chatbot deployment, chatbot
billion active users per month in 2017. initiatives too often aim for poor use
Given this increase, chatbots may be a cases, ignoring user needs and user
Request a media kit critical way to reach customers. This experiences.
fundamental change in online user Anna was too human. Ikea was
with specifications behavior has propelled every major one of the first companies to provide
and pricing: company to take on chatbots. Not chatbots in customer service. Its
only the typical technology giants experience is a thought-provoking
such as Google, Amazon, Facebook, example of some of the potential
Ilia Rodriguez and LinkedIn, but also consumer challenges with which chatbot
+1 212-626-0686 service companies such as Starbucks, designers and developers currently
British Airways, and eBay now aim struggle. For more than a decade,
[email protected] to reach their customers through the chatbot Anna aimed to answer
chatbots. Gartner presented as a top customer questions on Ikea.com.
strategic prediction for 2018 that Anna was ready to listen 24/7. Her

Developers and designers have an

urgent need to know more about how
people experience chatbots and to
understand the user needs that
motivate the future use of chatbots.
40 INTER ACTIONS S E P T E M B E R – O C T O B E R 2 0 18 INTER ACTIONS. ACM.ORG
 purpose was to guide customers
around the Ikea website in an
interactive and conversational way,
adding a personal touch.
However, Ikea retired Anna in
2016, with no plans for a replacement.
While the full reason is unclear,
frustrating customer interactions
with Anna may have played a major
role in ending Ikea’s innovative
chatbot adventure. As Magnus Jern,
president of the mobile solutions
company DMI, told the BBC, “If you
try too hard to be natural, it diverts
from the real purpose of [the chatbot],
which is about giving the right
answer as fast as possible.” Ikea’s
chatbot initiative clearly struggled
in terms of balancing human versus
robot aspects, causing people to ask
“stupid questions,” often sex related,
because Anna was too human [4],
according to Ikea. This is somewhat
different from the experiences with
other chatbot initiatives where users
have complained about a chatbot
experience as being overly robotic
and lacking a personal touch.
The ultimate chatbot fail. natural language dialogue in chatbots A poorly executed conversational
However, the ultimate chatbot fail suggests a low threshold for users to user interface frustrates users, which
was Microsoft’s Tay, deployed on access data and services. However, in turn will be harmful for business. A
Twitter in 2016. Tay was perceived as whereas conversational interfaces conversational user interface also has
a cutting-edge AI-based chatbot, the are truly intuitive when applied fewer opportunities for representing
goal being that she would learn and to interactions between people, the potential offerings of a service
mimic the personality of a 19-year-old conversations between humans and provider than does a graphical user
girl through interactions with Twitter automated conversational agents are interface. The chatbot dialogue
users. The problem was that Tay more challenging. instead needs to motivate focused
not only learned from well-meaning engagement, delivering simple and

C
Twitter users but also from Twitter CHATBOT INTERACTIONS compelling user experiences. In so
trolls, giving Tay fluency in all sorts WITH HUMANS doing, designers and developers
of hateful conversation. According Chatbots in general, and in need to tackle questions such as:
to Microsoft, the teaching of Tay particular, AI-powered chatbots, How friendly should they make
was revealed to be a coordinated need substantial adaptation and the chatbot? How fast should the
effort by trolls in specific online maintenance to perform their chatbot respond? How humanlike
communities (e.g., 4chan, Reddit, task properly. The potential or personal? What about gender;
and eBaumsworld), abusing Tay’s unpredictable variation in user input, should the chatbot be female, male,
commenting skills and having Tay as demonstrated in the chatbot or gender neutral? Should the chatbot
respond in offensive ways. Many of failures above, and what constitutes include a talk-to-human option?
Tay’s comments mimicked popular a valid chatbot response, represent Perhaps the chat element itself should
repugnant memes. In less than 24 substantial challenges for the HCI be deemphasized, with a greater
hours, Microsoft removed Tay from field. In order to develop chatbots that reliance on preset answer options?
Twitter, but only after she had praised adapt to the needs of specific users The latter approach may remove
Adolf Hitler and used harsh language and conversational contexts, there is misunderstandings in free-text
to express anti-feminist sentiment. likely a need for improved user and conversations, making the process
The stories of Anna and Tay leave context models. Chatbots and their more efficient. However, it also
chatbot developers and designers interactions with humans must be represents a return to the old-style
with tricky questions. Challenges analyzed and redesigned, not only graphical interface, with a limited
I M A G E B Y A L I C I A K U B I S TA

concerning the automation of with concern for specific interaction set of predefined choices at any given
dialogue may be even more sequences, but also with the aim of time.
substantial than those we faced when improving generative responses to Despite their long history, chatbots
designing graphical user interfaces. inputs from a range of users within a are still in rapid development, with
At first this may sound strange, as the variety of conversational contexts. advances every day. We can also

INTER ACTIONS. ACM.ORG S E P T E M B E R – O C T O B E R 2 0 18 INTER ACTIONS 41

 expect that the ways in which people organizing meetings, ordering food,
will interact with conversational user or booking a flight. Chatbots can also
interfaces in the future will change, help people explore online content
ACM Transactions resulting in new user behaviors as or services. For example, Microsoft

on Accessible well as new social norms and user

expectations. Consequently, more
launched Heston Bot, which focuses
on food and cooking opportunities as

Computing knowledge about chatbot experiences

from an end-user point of view is
well as fashion. The global clothing
company H&M launched a chatbot
needed. New user insights are critical to provide shopping suggestions
for chatbot designers and developers, based on photos from users’ personal
ACM TACCESS is a who must know the desires, needs, wardrobes.
quarterly journal that and practices of chatbot users.
Designing an interactive technology
Another stream of chatbots
supports more long-term
publishes refereed such as a chatbot requires in-depth relationships and activities such as
knowledge about why people charitable giving, civic engagement,
articles addressing are choosing to make use of this work, fitness, and personal health.
technology, as well as why people stop Some of these have already proven
issues of computing using it. It is necessary to understand to be important co-workers in
as it impacts the the people who use chatbots, their
goals, the tasks they have to perform,
office environments, while others
are “smalltalk” chatbots such
lives of people with as well as their context of use. Goals as Mitsuku and Replika, which
and tasks have often been seen in seem to support the need for
disabilities. The relation to motivational issues. connectedness. Some users find a
journal will be of However, motivation theories have
led researchers to focus on factors
chat with Mitsuku to be comforting,
much in the same way as a chat with
particular interest to that inspire people to use new
technologies and factors that make
a fellow human can be. Users are
revealing intimate details of their
SIGACCESS members technology use successful over a lives to their chatbot friend Replika,
longer time period. which is growing in popularity, in
and delegates to its particular among young people

I
NEW USER NEEDS
affiliated conference AND MOTIVATIONS
between ages 18 and 25.
In general, users expect
(i.e., ASSETS), as well In our high-choice media customer-service chatbots to be
environment, users of new effective and efficient in conducting
as other international technologies often exhibit a range productivity tasks such as accessing
of motivations and patterns of use. specific content or helping with
accessibility And with ever more chatbots being administrative chores; other chatbots
conferences. launched, there is an ever-increasing
range of possibilities. The great
may be used for entertainment-based
and social experiences. Moreover,
variety of chatbots is exemplified successful chatbots seem to inform
in the BotList (https://botlist. users about what to expect from
co/), a website where people can the beginning. This means they are
find chatbots for a broad range of transparent about who the users are
purposes and messaging platforms. having a conversation with—that
An enormous variation in chatbot they are interacting with a chatbot,
alternatives can also be seen by not a human. Information about what
exploring Facebook Messenger, which chatbots are able (and not able) to
now has over 200,000 chatbots [5]. deliver is another important factor to
One stream of chatbots allows communicate to the user.
users to complete quick and specific However, while there is some
tasks such as checking the weather, existing research into people’s

For further information We can expect that the ways in which

or to submit your people will interact with conversational
manuscript, user interfaces in the future will change,
visit taccess.acm.org resulting in new user behaviors as well as
new social norms and user expectations.
42 INTER ACTIONS S E P T E M B E R – O C T O B E R 2 0 18 INTER ACTIONS. ACM.ORG
uses of and motivations for using The productivity aspect changing user behavior as well as
media technology, there is a concerning efficiency and user needs. They are also altering
dearth of research on why people effectiveness is key for chatbot users, particular user interfaces, creating
use chatbots—or stop using who need to feel in control of the new demands in the field of HCI. We
them. According to the uses-and- technology. More concretely, they have here outlined some of these new
gratifications perspective [6], want to be able to have particular needs and challenges posed by the
people’s motivations to use media tasks done quickly and efficiently. emergent trend of chatbots.
are based on social and psychological This might have to do with the
factors. People are found to use additional time demanded of Endnotes
media technologies strategically by users and people’s expectations 1. Følstad, A. and Brandtzaeg, P.B.
Chatbots and the new world of HCI.
employing different technologies for in an accelerating digital society.
Interactions 24, 4 (2017), 38–42.
different purposes. Thus, media users Therefore, the main gratification for 2. Gartner top strategic predictions for
select among media technologies the typical chatbot user is to make life 2018; https://www.gartner.com/
based on how well a certain media easier and more productive. smarterwithgartner/gartner-top-
form helps them meet specific needs Chatbot users are not necessarily strategic-predictions-for-2018-and-
or goals. A fundamental notion in the looking for social experiences with beyond/
uses-and-gratifications perspective is family and friends, which they find 3. Luger, E. and Sellen, A. “Like having
a really bad PA”: The gulf between
that people are motivated by a desire through other media channels.
user expectation and experience of
to fulfill certain needs. Therefore, Rather, they may crave a means conversational agents. Proc. of the 2016
the key is to ask not how a particular of sharing personal information CHI Conference on Human Factors in
media use influences users but rather without fear of judgment. Thus, the Computing Systems. ACM, New York,
how users’ needs or requirements experience of connectedness is not 2016, 5286–5297.
influence their particular media necessarily about other people, but 4. Wakefield, J. Would you want to
talk to a machine? BBC. Aug. 4,
choices. These choices are found to be rather about being connected to
2016; http://www.bbc.com/news/
motivated by several basic needs, such oneself. A chatbot may serve a need technology-36225980
as entertainment, social connection, for connection and support 24/7, 5. Statista 2018; https://de.statista.
identity, and information. which may not be available through com/statistik/daten/studie/662144/
So, do chatbot users use media friends and family. Research long ago umfrage/anzahl-der-verfuegbaren-
in a completely different way from [8] revealed the tendency for people chatbots-fuer-den-facebook-messenger/
previous media users, and are user to open up more easily to a computer 6. Katz, E., Blumler, J.G., and Gurevitch,
M. Uses and gratifications research.
needs shifting with changing user or a technology than to humans.
The Public Opinion Quarterly 37, 4
interfaces? Our own recent research It is safer for people to participate (1973), 509–523.
on motivational factors of chatbot in self-disclosure with a chatbot. 7. Brandtzaeg, P.B. and Følstad, A. Why
users in a U.S. sample [7] found However, we know less about what people use chatbots. In Internet Science.
slightly similar basic needs among such relationships with chatbots can INSCI 2017. Lecture Notes in Computer
chatbot users and those found in more lead to in the long run—whether they Science, vol 10673. I. Kompatsiaris et al.,
traditional media use studies, albeit can generate increased loneliness eds. Springer, Cham, 2017, 377–392.
8. Weizenbaum, J. ELIZA—a computer
with some interesting differences. or depression or whether they can
program for the study of natural
The majority of chatbot users in our be balanced in a way that can be language communication between man
study reported their main motivation beneficial to users’ mental health. and machine. Communications of the
for using chatbots was the effective Another important lesson learned ACM 9, 1 (1966), 36–45.
and efficient accomplishment of is that chatbots are not a one-
productivity tasks. Participants solution-fits-all technology. People Petter Bae Brandtzaeg has a Ph.D.
typically reported using chatbots have multiple motivations, and the in media and communications from the
University of Oslo. His main research
to obtain assistance or information. purposes for using a chatbot can vary
interests are HCI and patterns of use
However, some users also reported enormously. As such, there is a need and implications of new information and
entertainment or social and relational for an appropriate range of use cases communication technologies. He coordinates
factors as their main motivations in the chatbot context.

D
the Social Health Bots project on chatbots
for using chatbots. They described for health services, supported by the
chatbots as a fun or entertaining CONCLUSION Research Council of Norway.
means to kill time. Some of the Designing and developing chatbots is →→ [email protected]
participants addressing social and about understanding new user needs Asbjørn Følstad has a Ph.D. in
relational factors even reported that and motivations, which is required psychology from the University of Oslo and
chatbots could help reduce loneliness to make successful automatic is currently senior researcher at SINTEF.
or enable socialization, though they conversational interfaces. Such His main research interests are human-
computer interaction and human-centered
of course knew that chatbots are conversational user interfaces to data
design. He coordinates the Human-
not people. A final category of users and services means a dramatic shift Chatbot Interaction Design project and
highlighted the novelty aspect, in how designers and developers are Chatbots4Loyalty, both supported by the
emphasizing an interest in trying out used to thinking about interaction Research Council of Norway.
new technology. and user needs. Chatbots are →→ [email protected]

DOI: 10.1145/3236669  © 2018 ACM 1072-5520/18/09 $15.00

INTER ACTIONS. ACM.ORG S E P T E M B E R – O C T O B E R 2 0 18 INTER ACTIONS 43