Using Virtual
Assistants and
Chatbots for Crisis
Communication
Uso de asistentes
virtuales y chatbots
para la comunicación
de crisis
aDResearch ESIC
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
Bonales Daimiel, G. y Martínez Estrella, E. (2021)
Using virtual assistants and chatbots for crisis
communication
Revista Internacional de Investigación en Comunicación
aDResearch ESIC. Nº 25 Vol 25
Monográco especial, marzo 2021 · Págs. 70 a 91
https://doi.org/10.7263/adresic-025-04
Gema Bonales Daimiel, Ph.D.,
Associate professor at the department
of Applied Communication Sciences,
Complutense University of Madrid
gbonales@ucm.es
https://orcid.org/0000-0003-2085-2203
Eva Citlali Martínez Estrella, Ph.D. student,
Complutense University of Madrid
evacmart@ucm.es
https://orcid.org/0000-0001-6599-1050
Purpose: To study the role of bots and the main virtual assistants as communication tools and
support for citizenship. The goal is to observe if this type of technology achieves the designed
objectives, if it is useful for the users and to know how they have contributed to the communi-
cation crisis strategies of the government and other institutions in the context of the Covid-19
pandemic.
Design / Methodology / Approach: The current study therefore pursued a mixed method:
a qualitative eldwork whit a quantitative assessment by adding an additional measurement
from user’s interest. In order to achieve the objectives, the research has a comparative analysis
that contrasts the data collected in surveys, interviews with experts and in the eldwork done
on the selected platforms, which have been analyzed from seven dierent dimensions: pre-
determined questions, initial greeting, privacy policies, languages, data on the health center,
health recommendations and a diagnosis option.
Results: The result of this work shows that 30% of the analyzed sample uses bots, and that
those who have used this type of assistants during the health crisis have done so mainly for
informational purposes, being the months of April and October the ones that have reected
the greatest activity in this regard; while 38.3% claim to use virtual assistants, Google is the
virtual assistant most used in Spain.
Limitations / Implications:The study is focused on the Spanish environment in specic time
during health crisis.
Originality / Contribution: Chatbots and virtual assistants are considered as a communication
channel that can help to strengthen sanitary measures.
Objetivo: Estudiar el papel de los bots y los principales asistentes virtuales como herramientas
de comunicación y apoyo a la ciudadanía. El objetivo es observar si este tipo de tecnología
consigue los propósitos pensados, si es útil para los usuarios y conocer cómo han contribuido
a las estrategias de comunicación de crisis por parte del gobierno y de otras instituciones en el
contexto de la pandemia de la Covid-19.
Diseño/Metodología/Enfoque: El presente estudio ha seguido un método mixto: un trabajo
de campo cualitativo con una evaluación cuantitativa, añadiendo una métrica adicional a par-
tir del interés de los usuarios. Para lograr los objetivos, la investigación cuenta con un análisis
comparativo que contrasta los datos recogidos en las encuestas, las entrevistas a expertos y
en el trabajo de campo realizado en las plataformas seleccionadas, que han sido analizadas
desde siete dimensiones diferentes: preguntas predeterminadas, el saludo inicial, políticas de
privacidad, idiomas, datos sobre centros de salud, recomendaciones sanitarias y opción de un
diagnóstico.
Resultados: El resultado de este trabajo señala que el 30 % de la muestra analizada utiliza bots,
y que quienes han utilizado este tipo de asistentes durante la crisis sanitaria lo han hecho con
nes informativos, en este sentido, fueron los meses de abril y octubre los que han reejado
mayor actividad; mientras que el 38,3 % arma utilizar asistentes virtuales, siendo Google el
asistente virtual más utilizado en España.
Limitaciones/Implicaciones: El estudio se centra en el entorno español en un momento con-
creto de crisis sanitaria.
Originalidad/Contribución: Los chatbots y asistentes virtuales se consideran un canal de co-
municación que puede ayudar a reforzar las medidas sanitarias.
ABSTRACT
RESUMEN
JEL Classication:
M31
Key words:
communication
channel,
chatbot,
virtual personal
assistant,
technology,
usability
Clasicación JEL:
M31
Palabras clave:
Canal de
comunicación,
chatbot,
asistente virtual,
tecnología,
usabilidad
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
aDResearch ESIC
72
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
1. Introduction
During the last years and specially with the
lockdown established in Spain, as a consequence
of the state of alarm decreed in March 2020,
the new information technologies have played
a relevant role thanks to their immediacy and
coverage possibilities. These have emerged in
the eld of health communication, as in other
areas, providing a change in the way of conceiving
medicine and, consequently, new terms have
emerged such as eHealth or telemedicine.
In this context, conversational platforms
have become one of the most used tools by
institutions in order to alleviate the lack of
resources in telephone and face-to-face attention
when dealing with the rst consultations made
by citizens who suffer from symptoms, or by
those who request some other type of urgent
information related to the disease. This demand
for agility was attended to by different companies
that offered the development of bot services in
record time. “Chatbots are a solution that allows
for the rapid implementation of an immediate
response to users” (Aguilar et al., 2020). This
article analyzes the level to which conversational
platforms have had the decongestion of health
services in Spain and whether their availability
has been useful for citizens. For this purpose, a
sample of eight conversational platforms has been
taken: four virtual assistants and four chatbots
in order to review their particular characteristics
and contributions. It also integrates the data
obtained through a survey to users and from
several interviews with experts.
The result of this research shows that the
level of use of specic health chatbots for the
Covid-19 currently reaches about 30% of the
sample analyzed and that people who have
used this type of assistance have done so for
information purposes, reecting greater activity in
this regard during April and October. Previously
chatbots were present for ecommerce activities,
but now conversational attendees have gained
strength in the Covid-19 pandemic. In this
way, they are no longer just a technological
development, because they are transforming into
another communication channel that works for
the dissemination of important information for
citizens, helping to decongest the health system
and with a possibility of taking a step towards
“telemedicine”.
The role of technology in crisis situations has
been shown to be fundamental and even more
so when it affects health on a global level, as is
currently the case. Technological developments
in the area of communications technology and,
in particular, the implementation of bots can
contribute signicantly to improving health care
and optimizing its resources, solving problems of
communication and information and mitigating
other problems of various kinds, including
psychological problems or those arising from
isolation. However, conducting the literature
review for this study, it has been observed that
there is no research focused on the effectiveness
and real scope of these applications (Zarouali et
al., 2018, Brandtzaeg and Folstad, 2017). Nor
any studies have been found that relate the role
of conversational platforms in health emergencies
or during the collapse of health systems.
Consequently, it seems entirely relevant that
the experience acquired in this extreme situation
should serve to stimulate the development
of studies related to the evolution on digital
technology, and, specifically, on the use of
different conversational platforms, which will be
very useful for the development and improvement
of tools related to the communication processes
between men and machines, promoting a
prevention culture.
73
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
2. AI application in the communica-
tion between human-machine
When talking about the human-machine
topic, especially about Articial Intelligence, it
is important to take into account the Restless
Valley Theory, which assures that the degree of
preference of the human being for an articial
intelligence will depend on whether the latter
becomes more human. If this is not the case,
the inclination for it diminishes. There is a great
afnity of systems that use avatars with human
forms, because it is easier for the user to identify
with the technological development in question
(Mori, 1970; Wilks, 2010).
Bots, according to Ferrara (2016), are computer
robots that operate online and remotely. However,
depending on the specic function for which they
have been designed, they adopt one name or
another and the term chatbot refers to a “machine
conversation system that interacts with human
users through a natural conversation language”
(Shawar and Atwell, 2005, p. 489).
Since the creation of human-computer
interface applications, the use of social keys
in the interaction with the computer has had
a continuous development (Prendinger and
Shizuka, 2004). In this sense, it is important to
go deeper into the functions of bots and virtual
assistants in communication acts or processes in
a period as serious as the current one. Intelligent
machines have appropriate computer programs
to solve some problems and not others, and
although they use the syntactic elements of the
language, they do not use semantics (Malpica,
2016, p.9), and this is where their main difculty
lies in communicating with man and imitating
him one hundred percent. Their algorithms are
logical, they are not human; so they do not form
part of the conscious reasoning process of the
machine. Consciousness is the capacity of the
human to debate, reect and realize his own
existence.
In recent years, studies about communication
and technology have focused on nding more
advanced ways to measure the usefulness and
effectiveness resulting from human-computer
interaction, and at the same time, to know how
the user responds. Through different studies
such as those of Lind and Salomonson (2006),
Brandtzaeg and Folstad, (2017), Van den Broeck
et al., (2019), among other researchers, it has
been proven that conversational robots can
improve the experience of online customers,
especially because they enhance the feeling
of being attended to at the precise moment,
they need it. Currently, however, it is difcult
to analyze the degree of interaction that exists
between these assistants and consumers as well
as the level of real understanding that users have
of the information provided by this media.
One of the main advantages of conversational
platforms is that they can be used by different
organizations for individual communication with
the consumer, achieving —despite an automated
process —optimal levels of personalization and
interaction (Van den Broeck et al., 2019 and
Kunse, 2016). This is possible because chatbots
are specically designed to be able to have highly
personalized conversations with users, using
pre-programmed dialogue schemes where the
client can make use of natural language or use
certain buttons with proposed options and thus
obtain particular recommendations and relevant
information.
The interaction between the user and the bot
is framed within conversational genres, based on
direct communication with the audience through
dialogue (Trillo-Domínguez et al., 2017). Due to
this connection, a relationship of trust is sought
to be established in order to respond to their
aDResearch ESIC
74
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
preferences and concerns. These conversational
media dominate digital rhetoric through natural
language, which should be as close as possible
to that of a human being: “the potential of bots
lies in their personalized distribution and in their
conversational power with audiences, through
instant messaging applications” (Sánchez-
González et al., 2017: 67).
Chatbots embody four fundamental charac-
teristics that detail their form and behavior: the
ability to act autonomously, full or partial auto-
mation, which require initial orientation, and are
used through online media (Grimme et al., 2017,
Hofeditz et al., 2020). In addition, its opera-
tion system is very simple, and anyone is trained
to operate it as the program is ready to answer
your questions (Dahiya, 2017). Overall, part of
its appeal lies in the fact that these bots are in-
stantly available 24 hours a day, 7 days a week.
The chatbot can make information available on
a larger scale and also improves customer service
by providing a quick response to the interested
party (Van den Broeck et al., 2019 and Herriman
et al., 2020)
When a user inputs any question the program
employs articial intelligence (AI) to give the an-
swer immediately, presenting the text in a conver-
sational way. This facilitates the collaboration with
researchers and help centers, because thanks to
the set of information collected in human-com-
puter conversations, it is possible to have a high
level of automation and enrich the databases.
The conversational ow of chatbots based
on a linguistic model, as shown in Figure 1, is
Figure 1. Chatbot working: conversational ow
Source: Villar, 2017.
NO
Welcome!
Question
Answer
2
Answer
1
End
Human
takeover
None
Try again
NO
Yes
Try again
or human
More
questions?
75
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
designed to be based on a linear block-structured
dialogue. These diagrams always begin with
a welcome message and a default answer.
However, the answers they provide are limited
to answering frequently asked questions (FAQs).
The possibility of generating more intelligent and
versatile answers that go beyond that predened
script would require constant updating. This
automatic learning is achieved by robots that use
AI in their design, requiring, therefore, a greater
investment. There are also hybrid models, which
have the advantage of allowing solutions to be
developed even in the absence of data and which
can go beyond the scope of linguistic rules to make
intelligent interferences in contexts where it is
complicated, or impossible, to use language tools
exclusively (Articial Solutions, 2020). An effective
chatbot provides results by helping patients to do
something without human intervention.
A virtual personal assistant in technical terms,
it is a software go-between that interacts with
individuals through voice, and can perform tasks
or services for us. The process is the result of a
merger of articial intelligence (AI), linguistics,
and computer science (Heneveld, 2018). So, these
tasks or services are based on user input data,
location recognition, process human speech, as
well as the ability to access information from a
variety of online resources.
“Combining NLP with machine learning,
personal assistants’ home on who we are, what
we like, and deliver personalized responses”
(Heneveld, 2018). The interaction that occurs
between a virtual assistant and a person should
be natural.
A person communicates using voice and the
virtual assistant processes, interprets and responds
in the same way. One of the key aspects of an
intelligent personal assistant is its ability to organize
and maintain information. organize and maintain
information. This includes the management of
emails, calendar events, les, to-do lists calendar,
les, to-do lists, etc. Having virtual assistants on
smart devices is now commonplace. The assistant
interacts with the user using the device through a
speech-based interface to answer questions, make
recommendations or perform actions in other
digital services to which it is able to connect on
demand, providing results in an individualized
and targeted way to the user with whom it
interacts (Guzman, 2017).
Some examples of personal assistants are Apple’s
Siri; Google Assistant and Google Now; Amazon
Echo; Microsoft Cortana; Voice Mate from LG;
Silvia and Hidi from HTC; Bixby from Samsung
and Braina. Siri is perhaps the best-known and
most popular of the virtual assistants, along with
Cortana, Cortana virtual assistants, along with
Cortana, Alexa and Google Assistant. These are
used in tablets, smartphones, cars, speakers’
smartphones, cars, speakers and electronic devices
of all kinds. Their function is to help users with
certain tasks to users. By means of a simple voice
command, they can carry out certain open a
certain application, making life easier. People just
have to give them the order of what is needed.
“Many believe consumers will increasingly
become more dependent on smart speakers and
their ilk, because they will offer news in a much
more interactive manner than traditional news
sources and even query the assistant for specic
topics and further customize and personalize the
experience over time”, (Martin, 2017).
An example of the workow of these virtual
assistants is shown in Figure 2. Its computational
characteristic is that they respond to a specic
user’s intention, which is the main purpose that
the each virtual assistant has.
Thus, virtual personal assistants are innovations
that allow the user to interact with a computer
aDResearch ESIC
76
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
textually and/or verbally and whose interactions are
digitized to be subsequently stored, analyzed and
processed through various AI techniques (Tabarés,
2020). This opens up a natural path to add more
smart home products, enhance the customer
experience, and build customer loyalty through the
empowerment of voice control (Heneveld, 2018).
Currently this technology uses female voices -so
present in IOT technology nowadays- as authority,
referring us to gender stereotypes (Piñeiro-Otero,
2011). Therefore, it is of interest to study why only
female voices are used in these technologies and
what is the scope they have.
Nevertheless, virtual assistants do not yet
have the ability to recognize a certain illness
through voice analysis; however, technological
developments are working on it to achieve the
application of “telemedicine”. “Chatbots will play a
leading role by embodying the function of a virtual
assistant and bridging the gap between patients
and clinicians. Powered by AI and machine
learning algorithms, chatbots are forecasted to
save healthcare costs when used in place of a
human or assist them as a preliminary step of
helping to assess a condition and providing self-
care recommendations” (Fadhil, 2018).
Additionally, it offers a simple, low-friction
experience (Jenkins et al., 2007). Thus, this
digital development transforms the alternatives
in the communication processes and speeds
up the delivery of information. In this context,
it can be said that bots avoid the congestion
of communication networks and health
system facilities, especially in care centers, and
consequently contribute to minimizing the
risk of contagion by contact. Furthermore, the
development of these platforms based on articial
intelligence can also collaborate in the health
education of the population, for example, reducing
the possibility of spreading false information about
Covid-19. Similarly, the presence of this type of
technology helps institutions to have a friendlier
and closer approach to concerned citizens (Ahuja
et al., 2020).
Figure 2. Virtual assitants’s workow
Source: Paul, 2018.
77
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
So, we are facing a techno-social ecosystem,
extended by various platforms, devices, protocols,
algorithms and interfaces that seek at all times
to digitize and enhance the interactions of its
users with the environment. Therefore, both
conversational assistants have unique possibilities
to strengthen general well-being, because
they can mitigate the emotional burden that
infectious diseases cause in the short and long
term in pandemics and thus, promote an active
communication between health authorities or
managers with each individual.
3. Use of virtual assistants and bots
According to a study conducted by HubSpot
(2019), the 48% of consumers would instead
connect with a company via live chat than any
other means of contact; besides, 63% of people
would consider messaging an online chatbot to
communicate with a business or brand (Joseph,
2019). So, the chatbot can give the opportunity
of boosting a brand perception, because the
consumer has a friendly experience that offers a
visual interaction.
With Statista’s 2019 results, 78% of service
organizations are taking advantage of chatbots in
simple self-service scenarios. Just in second place
with 77% is the use of bots to evaluate the type
and difculty of a query before passing it on to the
commercial agents. It should be noted that this
communication channel has surpassed the use
of social networks between brand and consumer.
In Spain there are 43 million internet users,
of which 52% use social networks and 62%
make online purchases, with product and travel
inquiries being the main information sought to
be answered by a chatbot (We are social, 2019).
As for virtual assistants, around 4.3 million
families use them on a daily basis, which is
equivalent to 10.7% of the Spanish population,
according to the results of the 1st Wave of the
General Media Study 2019 (EGM) and the
Association for Media Research (AIMC). The
leading virtual assistant is Siri, owned by Apple,
while the assistant of Google, Alexa (Amazon),
Cortana (Microsoft) and Bixby (Samsung) have
an average share between 7% and 12%. Aura
(Movistar) is beginning to be used.
Figure 3. Virtual assistant insertion in Spain
Source: Estudio General de Medios (EGM).
Asistentes utilizados
Siri
(Apple)
Asistente
de Google
Alexa
(Amazon)
Cortana
(Microsoft)
Bixby
(Samsung)
Aura
(Movistar)
0
aDResearch ESIC
78
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
Furthermore, during 2020, only 8% of the
Spanish population said they use virtual assistants
several times a day; while almost 20% said they
use conversational platforms a couple of times a
week (Statista, 2021). Virtual assistants become
an important channel for generating two-way
communication in emergency management
(Bird et al., 2012), in which there is a need to
disseminate accurate information to combat
rumors, give warnings and recommendations for
action; as well as communicate data on specic
situations (Hofeditz et al., 2020).
Another benet of a virtual assistant is that it
can be used in text or voice format, thus helping
people with disabilities, or elderly people without
sufcient knowledge to search manually for the
required information on the web. According to a
study carried out by the Digital Marketing Agency
Trend (2020) in Spain, assistants are used mainly
to ask questions, check the weather, nd places
or listen to music and their presence represents
4.3 million homes in this country.
For these reasons, the conversational platforms
have been a key piece in this crisis because they
have provided information, follow-up, monitoring
and health control to the citizen in the rst phase
of the detection of the contagion, when the health
system was especially saturated with sick people
and people who presented some symptoms
compatible with the Covid-19 infection.
4. Inuence of conversational
platforms on a crisis communication
Natural disasters and pandemics are unforeseeable
and pose unique challenges to health care delivery;
as a result, the Covid-19 pandemic has been a huge
undertaking for both researchers and technology
developers, as technical teams around the world
have sought digital solutions to help relieve the
pressure on medical issues in healthcare facilities.
Researchers Miner, Laranjo and Kocaballi
(2020) believe that pandemics have three unique
characteristics that make them susceptible to
personalized interventions through these robots.
First, individual actions can signicantly worsen
outcomes in these crises, as a single individual
can infect more people, depending on their
behavior. Second, there is a fear of infection
and a “stigma associated with the disease, as a
result, people may feel personally responsible for
poor outcomes during a pandemic and also hide
symptoms” (Miner, Laranjo, & Kocaballi, 2020).
The third feature is that during such a situation,
many people should avoid physical gatherings,
which may worsen the risk of future mental health
problems.
For this reason, several government
organizations and health institutions have
used social media platforms such as Facebook,
WhatsApp, Instagram and Twitter to provide
specific information about Covid-19 to the
population. WhatsApp is one of the main methods
of providing information about Covid-19 in many
countries (Hutchinson, 2020). An example is the
governments of Australia, India, Singapore and
the United Kingdom.
The creation of all Covid-19 applications
has served to collect and register data related to
the user’s health. The rst to be developed and
widely distributed were the contact tracking
applications, which were created to notify users
if they had crossed paths with another person
infected with the coronavirus. The rst of these
was developed in Singapore (Baharudin, 2020).
Later, applications were developed to monitor
the compliance of citizens with the lockdown, as
well as conversational assistants that answered the
frequent questions of the population.
The applications of symptom surveillance
have been useful in identifying disease trends
79
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
and possible areas of high transmission. Most
countries made the applications and chatbots
based on the disease symptom manual and on the
questionnaires that users were answering (Hanson
et al., 2020).
Table 1 lists some of the bots that emerged
in Spain (2020) to join the health struggle in
the pandemic corresponding to the institutions
contacted for this study.
Carina has been developed by 1Millionbot in
Spain. It is a free conversational platform about
coronavirus based on AI that uses ofcial sources
such as the World Health Organization (WHO)
and ofcial government sources. This technology,
launched in the rst half of March 2020, has been
inserted by different entities in their web pages.
Like Carina in Spain, this company developed
a month later for the Republic of Ecuador,
the Catalina chatbot. The United Nations
Development Programme (UNDP) awarded
this Alicante-based company the development
of Catalina to provide an urgent response to the
demands of the citizens of Ecuador. This bot is
hosted in 79 ofcial pages, and in three months
of operation had an interaction of 135,000 users
(Torre Juana, 2020).
Hubtype has created, along with the Ministry of
Health, a conversational robot on the coronavirus
for the Generalitat through instant messaging
applications such as WhatsApp. This company
denes itself as a company that “creates quality
conversational experiences’’ (Hubtype, 2020).
The bot, available since June 2020, shares content
on health and civil protection, as well as data on
those affected, advice, lockdown measures and
information on the evolution of the pandemic for
free.
Chatbot Chocolate is “the leading agency in
the design, development and training of chatbots
and voice apps” (Chatbot Chocolate, 2020). Maia,
conversational assistant launched by this company
in early April 2020, is capable of answering more
than 125 different questions asked in more than
50,000 different ways in Basque and Spanish
(Planeta Chatbot, 2020).
The Community of Madrid decided to create
a platform inspired in South Korea, which was
launched at the end of March 2020. The function
of CoronaMadrid is to evaluate through a series of
questions the clinical situation of the citizen. So,
the user can know the magnitude of his symptoms
and is guided on the procedure to follow in each
Table 1. Conversational robots in Spain.
Company Media Chatbot Institution
1. 1millionbot Web and WhatsApp Carina Bot
Ayuntamiento de Alicante
y Elche
2. Hubtype WhatsApp and Telegram gencatBot Generalitat de Catalunya
3. Chatbot Chocolate Web Maia Servicio Vasco de Salud
4. CARTO, ForceManager y
Mendesaltaren, con el apoyo
de Telefónica, Ferrovial,
Google y Googo.
Web and WhatsApp CoronaMadrid Comunidad de Madrid
Source: Own elaboration.
aDResearch ESIC
80
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
case. The evaluation can be done again every 12
hours to incorporate the new developments that
could have happened in his evolution. In addition,
this system allows to obtain epidemiological data
and to offer personalized information, and to offer
the best preventive and evaluation measures, it
includes a geolocalization system” (Pérez, 2020).
5. Methodology
The technologies from the virtual assistants
need massive amounts of data, which include
algorithms to learn from data input and become
each time even better at predicting the user’s
needs. The current study therefore pursued a
mixed method: a qualitative eldwork whit a
quantitative assessment by adding an additional
measurement from user’s interest. The main goal
is to nd patterns of usability and to know which
virtual assistant has the higher human speech
understanding. Also, the collected data was used
to determine the content quality of the answers
and to know if the virtual assistants and bots
could provide a diagnosis of a possible Covid-19
infection.
A eld work was made to obtain rsthand
information from four virtual assistants with
the largest market share in Spain: Siri, Alexa,
Cortana and Google; as well as four conversational
platforms that emerged during the Covid-19 and
were used by Spanish people. This work consisted
of making a list of generic questions about the
virus information, then the answers of this AI
technologies will show the understanding level
and its ability to help in a health crisis. For these
reasons, the following parameters were used to
explore both kind of platforms:
• Privacy Policy.
• The initial greeting.
• Languages that have.
• Information: battery of questions and
answers you have.
• Recommendations about the disease.
• If there is a diagnosis of Covid-19.
• Information about health centers.
These criteria were chosen because
through them, it was possible to determine its
functionality, whether it was possible to have
a uid conversation between the program and
the user and also, to know if the information
provided by the virtual assistants was useful
and helped the user’s doubts. With the language
criteria, the inclusive possibilities provided were
established, while the initial greeting and privacy
policies are important elements for building trust
with the user. These parameters have provided
the necessary data (qualitative and quantitative)
to make a comparative analysis between virtual
assistants and chatbots.
Then, an in-depth interview was conducted
with four experts representing the above-
mentioned companies, which took place during
July and August (2019). This type of interview
is dened as a “social technique that puts into a
relationship of direct face-to-face communication
a researcher/interviewer and an individual
interviewed with whom a peculiar relationship
of knowledge is established that is dialogical,
spontaneous, concentrated and of variable
intensity” (Canales, 2006, p. 219). This research
technique produces information of a qualitative
nature, seeking to generate greater richness in the
linguistic material collected.
And nally, in order to complete the analysis,
a survey was carried out, because it was
important to explore if both programs achieve
their initial objective. So, the survey participants
added validation to the data collected in the
fieldwork. The purpose of any survey is to
81
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
measure the degree or the way in which the
subjects possess certain variables and concepts
of interest for the research. In order to facilitate
the process of carrying out and later measuring,
the questionnaire had closed answers. This was
conducted online in September and October
(2020). The sample universe (n=100) was
18 to 65 years of age and was intended to be
simple random to minimize the probability of
bias. Participating people were from ve Spain’s
Autonomous Communities: Madrid, Valencia,
País Vasco, Castilla la Macha and Andalucía.
All people participating in the study have a
smartphone. This ensured that they are familiar
with some type of AI technology.
Then, six categories were reviewed by the
survey: use of specic chatbots during covid,
evaluation of their performance, usability,
handling, usefulness to replace face-to-face
assistance and proposed improvements to the
platforms; as well as a general assessment of
the tool used during the pandemic and possible
suggestions for improvement. In this way, the aim
was to contrast the experts’ point of view about the
bots with the users’ own assessments.
Throughout research described, the study
aimed to examine the complex interaction
between the strengths and weaknesses of current
AI conversational technology. Through the use
of both quantitative and qualitative data, this
text provides a comprehensive examination of
user’s interest experiences of usability and health
information quality.
6. Results
The following tables show the most relevant
results of the experiments carried out on the four
virtual assistants used in the study.
Table 2. Comparison of the conversational robots
Carina Bot gencatBot Maia CoronaMadrid
Privacy Policy no no yes yes
Initial greeting yes yes yes yes
Languages Spanish
and English
Catalan Spanish and
Euskera
Spanish, English, French
and German
Questions preset preset preset preset
Answers preset preset preset only yes or no
Recommendations yes yes yes yes
Background/Prole yes no no yes
Diagnosis no no no yes
Information about health centers no no no yes
Source: Own elaboration.
aDResearch ESIC
82
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
Table 3. Comparison of virtual assistants
Siri Google Alexa Cortana
% Preset Answers 69 % 56 % 49 % 38 %
% Complete answers 75 % 32 % 25 % 20 %
% Humor 0,8 % 0,3 % 0,1 % 0 %
Recommendations Yes NO Yes NO
Source: Own elaboration.
Table 4. Comparison between bots and virtual assistants
Bots Virtual Assistants
Privacy Policy Yes no
Initial greeting Yes
It only has it if the user requests it.
Languages Several
Depending on the programming.
Questions Preset
No questions asked
Answers Preset Preset
Background/Prole
Yes no
Diagnosis Yes no
Information about health centers. Yes
It only has it if the user requests it.
Source: Own elaboration.
Table 5. Functionality of chatbots
Question Answer
Did the chatbots help in the health crisis? 100 % - Yes
Which is the advantage of a chatbot? 100 % - Accessibility of information
When was it used the most?
50 % - at the peaks of the pandemic.
50 % - at the beginning of the lockdown
Do you think that in time this technology will be implemented? 100 % - Yes
Possible improvements
Adapted for people with disabilities.
Transactional capacity.
Source: Own elaboration.
83
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
According with the Table 2, CoronaMadrid
is the more complete chatbot, because it has
several languages and also provides health centers
information.
As shown in Table 3, Siri is the virtual assistant
that has the highest understanding of language,
because it can provide answers with a more
complete grammatical structure and even adds
some humorous phrases.
The following table shows the main differences
between bots and virtual assistants in terms of the
criteria that had been evaluated.
Later, From the interviews conducted
with experts, the following answers can be
highlighted:
The survey measures the user’s interest
about Covid- 19 chatbots. More than half of the
participants answered that these technologies have
helped them with “useful” information about the
Covid- 19 sickness. While, a 15% answered that
with the chatbot they knew which health protocol
have to follow. Only a 4% mention that the chatbot
was useful to identify the infected people who
were nearby.
Figure 4. Percentage of responses to the question Did it help?
Source: Own elaboration.
YES
NO
45,8 %
54,2 %
Figure 5. Usability of Covid-19’s Chatbots
Source: Own elaboration.
YES
NO
92,3 %
7,7 %
aDResearch ESIC
84
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
In terms of usability, almost 100% of the sam-
ple found easy to use the Covid-19 chatbot.
Following, 30,8% indicates that this kind of
chatbots were very useful to provide information
on the disease. Only the 15,4% replied that it was
“not useful at all”.
The next gure shows how the bots were used
in relation to the request for information on covid
during 2020. We can appreciate the peaks during
2 months: April and October.
Then, 38.3% of the participants said they
use some virtual assistant, usually that of their
own cell phone.
In general, there is a closed relationship
between the age and the interaction with any
AI programs, because people in their 20s and
35s are the ones who use them the most.
Afterwards, once the results are analyzed,
the paper will be closed showing the usability
According whit the participants, the
chatbots are just ne, because the 70% of
the sample answered that the do not make
changes in the platforms. A 2% said that may
will be a good option to review the health
terms and use some less complex for having
an easy understanding.
patterns of the platforms and the main
conclusions.
7. Discussion and Conclusions
This study has yielded several conclusions.
Firstly, it has been observed that all the
bots analyzed coincide in several of their
Figure 6. Bots usage curve (2020)
Source: Own elaboration.
35.00%
30.00%
25.00%
20.00%
15.00%
10.00%
5.00%
0.00%
March April May June July August September October November
85
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
Figure 8. Changes in Covid-19 chatbots
Source: Own elaboration.
Figure 9. Use of virtual personal assistants
Source: Own elaboration.
Nothing Funcionality Speed Health terms
60
40
0
80
20
characteristics. Two of the platforms studied
-Carina and Maia, which do not belong to
the same company, have a similar look &
feel (green and white color) and bot use a
female voice and name, coinciding with the
research made by Piñeiro-Otero (2011),
which stablished that female voices are
used as an authority reference. The rest of
the platforms have an assistant that is not
based on the virtualization of a human being.
However, this fact doesn’t make a difference
for the users.
Siri
Google
Alexa
Cortana
Other
None
61,7 %
14,9 %
19,1 %
aDResearch ESIC
86
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
On the other hand, the interview carried out
has been fundamental to know the opinions of
the professionals who are behind the development
of this technology. This study’s results provide
a picture of the importance of the chatbots in
this health crisis and the rst thing that can be
deduced, once the answers have been analyzed,
is the informative nature that most of these
conversational robots have, in addition to the
fact that all the interviewees emphasize the
importance of their work during the rst weeks
of the state of alarm. It has been detected through
this analysis that the systems incorporate a low
degree of Articial Intelligence, especially when
it comes to the use of natural language in the
interaction between chatbot-user; in several cases
it happened that the bot did not understand the
questions asked, which means that the options in
the conversational ow are limited.
Other conclusion derived from the in-depth
interviews with the experts and protagonists of
the design of these platforms is that the 96% were
satised with the fulllment of the objectives that
had been proposed in their design, specically
those of being a reliable informative tool that
would collaborate to create new communicative
channels that would manage to decongest the
traditional ones. In contrast, as reected in the
user surveys, only 29% of the participants stated
that they did not contact their health center after
making the consultations through these means.
Despite the fact that these tools were designed
to decongest a health system that was beginning
to collapse, the reality is that they have not
significantly prevented people from going to
their health centers, since these tools -still in the
rst phase of development in terms of Articial
Intelligence (AI)- are not prepared to respond to
certain complex questions or to do so with the
security that person-to-person contact provides
when explaining their problems in detail.
Likewise, the surveys have helped to complete
the study and to get the perspective of the
citizens’ use experience. In this sense, it is clear
the main reason why the respondents have
come to these platforms and that is the search
for information. Specically, there are two peaks
of use corresponding to the months of April and
October, which also coincide with the two waves
of infections in Spain so it can be deduced that the
worsening of the gures of infections and deaths
is related to the demand for access to such sources
of information. Additionally, experts agree that
these periods were the ones when the platforms
were most used.
It is also found that the lockdown —released
in March 2020, as a result of the state of alarm—
accelerated the adoption of this technology in
health care area. During these months, the bots
have improved; for example, they have included:
recommendations, prole information and data to
contact health centers. However, they still follow a
tree of answers and one of their clearest limitations is
that they only answer certain predened questions.
Figure 9. Use of virtual personal assistants
Figure 10. Conversational robots that
have the look of a woman
Source: Carina y Maia, 2020.
87
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
The information provided by the sample about
the most used platform places CoronaMadrid in
rst place, a fact that is in agreement with the
origin of most of the respondents, although
others said they do not remember the name of
the chatbot used. It is noteworthy that some
people pointed to RadarCovid as the platform.
Nevertheless, this is an app that requires
downloading and is primarily intended to
register infected people to warn users of having
been in contact with a positive case, but it is not
a chatbot. This indicates that users are unaware
about the difference between AI technologies
and the services offered by each one. Currently,
most people install different applications on their
cell phones for entertainment, professional and
even educational use, but one of the things that
distinguishes conversational robots from other
apps is that they do not need to be downloaded,
as they work online through websites or social
networks.
Then, as a summary of the results obtained in
the series of questions asked to the four assistants,
it can be concluded that the answers of Alexa and
Cortana are the most elaborated grammatically.
Therefore, they are more advanced in terms of
intellect. Alexa, especially, is one step ahead in
terms of natural language. Siri is the only one
who calls you by your name when answering.
Alexa has order information, has the bank details,
address and name, but does not address the user
by name, as Siri does. And none of the four
attendees includes invasive advertising of the
type of conventional advertising such as spots or
banners on the Internet.
Subjectively, assistants fall short in intellect
because they give very primary answers to the
most open or complex questions. Their articial
intelligence, despite appearing complex, is
still very simple and leaves a lot of margin for
improvement. It can be said that the level of
language comprehension is basic. Occasionally,
they do not respond to questions formulated
in natural language that have an average level
of complexity because they do not understand
them or know how to process them, in addition
to not distinguishing properly between statements
and questions. Their intonation is simple, and
they should work on using it to communicate
the results more effectively. Likewise, they do not
process the information in a logical way, their
answers are determined; therefore, they limit
themselves to giving standard answers, since they
respond in the same way to different questions
and say the same thing to some and to others,
there is no difference between interlocutors.
Virtual assistants can perform voice recognition
awkwardly. In addition, they do not remember
whether they have been asked the same question
ve or ten times. So, this lack of memory may
make the virtual assistant less useful for people
with disabilities.
So, it can be concluded that all these conversa-
tional platforms have collaborated to disseminate
the preventive measures established by the health
authorities such as the promotion of the famous
“Stay at home”, a recommendation that all include.
Furthermore, due to the type of language used
and the structuring of the programmed responses,
they are a reinforcement tool that emphasizes the
messages regarding how to act in case of contagion
and, by maintaining an informative tone, they
also manage to transmit a reassuring sensation,
preventing users from panicking and acting in
a chaotic way that is not very effective for their
interests and those of the rest of the citizens. For
these reasons, the analyzed conversational assis-
tants are positioned as another useful means of
communication to be used in communication
strategies in case of crisis.
aDResearch ESIC
88
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
In general, and for all that has been mentioned,
it seems pretty evident that a transformation is
taking place in the semiotic patterns of culture and
of advertising in particular around this technology.
A transformation that has only just begun, since
such motivations have not yet completely taken. At
his moment, there is a lot of space for innovation
and development of these assistants. We are
very close to the real transformation. However,
as devices become more interconnected to the
social network proles of each user and can collect
more data, privacy and security concerns will
increase. The way companies choose to balance
their customers’ privacy with this abundance
of data will be a crucial factor in maintaining
user condence and may even be marked as a
competitive advantage among conversational
platforms.
While it is true that the processing of natural
language in virtual assistants and chatbots still has
much to improve, what nobody doubts is that the
great opportunity they will provide to advertising,
since through a conversational business you can
establish a relationship with a more personal level
with the brand. The future, without a doubt, is
going that way, where the stars of the ads are not
the celebrities who star, but the product being
promoted. The brand itself is mutating, because
in this case, the conversational assistants will talk
about their values and their advantages to the
consumers, providing more credibility.
A possible scenario for these AI tools is that
they will be used to enhance human capabilities
even further and that they can spend more time
on strategic rather than operational activities. With
these technological developments, a path is being
promoted in ow automation projects, especially
within the health sector, since thanks to the user
experience and data collection, there is a constant
improvement in software performance. In the
long term, when this crisis is over, it is likely that
chatbots will become digital portals for interactive
health care, helping patients and doctors and thus
facilitating the rst care in a health emergency.
Consequently, monitoring the development and
implementation of conversational robots in the
health area and their application in crisis situations
is still a very new eld that undoubtedly invites
researchers to remain alert and open new avenues
of research.
89
Using virtual assistants and chatbots for crisis communication · págs. 70 a 91
Bibliography
Agencia Digital Marketing Trend (2020). Mobile en España
y en el Mundo. https://www.amic.media/media/les/
le_352_2531.pdf
Aguilar, J & Villegas, O. (2020). COVID-19 y sus implicacio-
nes sociales: una mirada desde las ciencias computacionales.
https://tinyurl.com/y58k32x2
Ahuja, A., Reddy, V. & Marques, O. (2020). Articial Intelli-
gence and COVID-19: A
Multidisciplinary Approach, Integrative Medicine Research,
9(3) doi.org/10.1016/j.imr.2020.100434.
Articial Solutions. Chatbots: la guía denitiva (2020). https://
www.articial-solutions.com/es/chatbots-pdf/
Baharudin, H. (2020). Coronavirus: Singapore Government to
make its contact-tracing app freely available to developers
worldwide. https://tinyurl.com/yyz7ldwk
Bird, D., Ling, M., & Haynes, K. (2012). Flooding Facebook
-the use of social media during the Queensland and Victo-
rian oods, Australian Journal of Emergency Management,
27(1), 27–33.
Brandtzaeg P., Folstad A. (2017). Why People Use Chatbots.
Internet Science. https://doi.org/10.1007/978-3-319-70284-
1_30.
Canales, M. (2006). Metodologías de investigación social. Chile:
Lom Ediciones.
Dahiya, M. (2017). A Tool of Conversation: Chatbot, Inter-
national Journal of Computer Sciences and Engineering, 5(5).
Fadhil, A. (2018). Beyond Patient Monitoring: Conversatio-
nal Agents Role in Telemedicine & Healthcare Support For
Home-Living Elderly Individuals. Cornell University. https://
arxiv.org/pdf/1803.06000.pdf
Ferrara, B; Varol, O; Davis, C; Menczer, F & Flammini, A.
(2016). The Rise of Social Bots, Communications of the Acm,
59(7).
Grimme, C., Preuss, M., Adam, L., & Trautmann, H. (2017).
Social Bots: Human-Like by Means of Human Control?, Big
Data, 5(4), 279–293, doi: 10.1089/big.2017.0044.
Guzman, A. L. (2017). Making AI safe for humans: A
conversation with Siri. In R. W. Gehl & M. Bakardjieva
(Eds.), Socialbots and Their Friends: Digital Media and the Au-
tomation of Sociality (pp. 69–85). New York: Routledge.
https://doi.org/10.4324/9781315637228
Hanson, J., Couch, D., & Yap, K. (2020). Mobile Health Apps
That Help With COVID-19 Management: Scoping Review.
JMIR Nursing. 3(1). https://www.nursing.jmir.org/2020/1/
e20596/#ref75.
Heneveld, H. M. (2018). Let’s talk about embracing voice
technology. Sdm, 48(2), 40–40.
Herriman, M., Meer E., Rosin R., Lee V., Washington V., &
Volpp, K. (2020). Asked and Answered: Building a Chatbot
to Address Covid-19-Related Concerns, NEJM Catalyst. doi:
10.1056/CAT.20.0230.
Hofeditz, L., Ehnis, C., Bunker, D., Brachten, F., & Stieglitz,
S. (2019). Meaningful use of social bots? Possible applica-
tions in crisis communication during disasters. En Procee-
dings of the 27th European Conference on Information
Systems (ECIS), Stockholm y Uppsala, Suecia, Junio 8-14,
2019. ISBN978-1-7336325-0-8 Research Papers. https://
aisel.aisnet.org/ecis2019_rp/138.
Hubtype. (2020). Mensajería escalable. https://www.hubty-
pe.com/es/.
Hutchinson, A. (2020). WhatsApp launches new COVID-19
info hub amid concerns around misinformation being shared in-
app. https://tinyurl.com/y6hj3pll.
Jenkins, M., Churchill R., Cox S., & Smith D. (2007) Analy-
sis of User Interaction with Service Oriented Chatbot Sys-
tems. Computer Science, 4552. Doi: 10.1007/978-3-540-
73110-8_9.
Joseph, R. (2019). Chatbot Statistics All the Data You Need.
https://www.intellectyx.com/blog/chatbot-statstics-2019/
Lind, M., & Salomonson. N. (2006). The Role of Virtual
Servants in e-Interaction. https://tinyurl.com/y2ybrnhv.
Martin, E. (2017). How Echo, Google Home, and Other Voice
Assistants Can Change the Game for Content Creators.
E.Content, 40(2),4-8.
Miner, A., Laranjo, L., & Kocaballi, A. (2020). Chatbots in
the ght against the covid-19 pandemic, Npj Digital Medici-
ne, 3(1). doi.org/10.1038/s41746-020-0280-0.
Mori, M. (1970). The Uncanny Valley. Energy, 7(4), 33-35.
Doi:10.1109/MRA.2012.2192811
Paul, S. (2018). How to build a Google Home App with
Dialogow |Overview. Medium. shorturl.at/demRU
Pérez, J. (2020, 18 de marzo). Madrid lanza una web para
atender a posibles afectados de coronavirus. https://tinyurl.
com/y3f4tvxm.
aDResearch ESIC
90
Nº 25 Vol 25 · Monográco especial, marzo 2021  págs. 70 a 91
Piñeiro-Otero, M. (2011). La utilización de la voz femenina
como autoridad en la publicidad radiofónica española.
https://www.researchgate.net/publication/279500825
Planeta Chatbot. (2020, 28 de abril). Tecnología de chatbot
Chocolate en Osakidetza. https://tinyurl.com/y3mxxbdn.
Prendinger, H. & Ishizuka, M. (2004). Life-like characters:
Tools, affective functions, and applications. Springer Science &
Business Media.
Sánchez-Gonzales, Hada M.; Sánchez-González, María
(2017). Los bots como servicio de noticias y de conectivi-
dad emocional con las audiencias. El caso de Politibot. Doxa
comunicación, 25, 63-84.
Sayobo. (2019). Chatbot COVID- 19. www.sayobo.io.
Shawar, A. & Atwell, E. (2005). Using corpora in machine-
learning chatbot systems, International Journal of Corpus Lin-
guistics, 10(4), 489–516.
Statista. (2020). Frecuencia de uso de asistentes virtuales
de voz en España en 2019. https://es.statista.com/estadisti-
cas/1018541/uso-de-asistentes-virtuales-de-voz-en-espana/
Stieglitz, S., Bunker, D., Mirbabaie, M. & Ehnis, C. (2017).
Sense-Making in Social Media During Extreme Events, Jour-
nal of Contingencies and Crisis Management, 26(1), 4-15.
Tabarés, R.(2020). Conversando con cajas negras; sobre la
aparición de los interfaces conversacio-nales. Teknokultura.
Revista de Cultura Digital y Movimientos Sociales, 17(2), 179-
186.
Torre Juana. (2020, 25 de julio). Naciones Unidas y ‘Catali-
na’: ayudando más allá de los síntomas de COVID-19. https://
tinyurl.com/y4voya46.
Trillo-Domínguez, Magdalena; Alberich-Pascual, Jordi
(2017). Deconstrucción de los géneros periodísticos y nue-
vos medios: de la pirámide invertida al cubo de Rubik. El
profesional de la información, 26(6),1091-1099. doi.
org/10.3145/epi.2017.nov.08
Van den Broek, E., Zarouali, B., & Poels, K. (2019). Chatbot
advertising effectiveness: When does the message get
through?, Computers in human behaviour, 98. doi.
org/10.1016/j.chb.2019.04.009.
Villar, C. (2017). Creating Conversational Experiences.
Landbot.io. https://medium.com/landbot-io/creating-con-
versational-experiences-ii-build-and-design-20ac88d7ee72
We are social (2019). Digital en 2019 en España. https://
wearesocial.com/es/digital-2019-espana
Wilks, Y. (2010). Close Engagements with Articial Compa-
nions. Key social, psychological, ethical and design issues. Ho-
landa: John Benjamins.
World Health Organization. (2020, 12 de marzo). WHO
announces COVID-19 outbreak a pandemic. https://tinyurl.
com/y48wdsed.