The Adoption and Implementation of Artificial Intelligence Chatbots in Public Organizations: Evidence from U.S. State Governments

• Published date: 01 April 2024
• DOI: http://doi.org/10.1177/02750740231200522
• Author: Tzuhao Chen,Mila Gascó-Hernandez,Marc Esteve
• Date: 01 April 2024

The Adoption and Implementation of

Artif‌icial Intelligence Chatbots in Public

Organizations: Evidence from U.S. State

Governments

Tzuhao Chen

1

, Mila Gascó-Hernandez

1

and Marc Esteve

2,3

Abstract

Although the use of artif‌icia l intelligence (AI) chatbots in public organizations ha s increased in recent years, three crucial gaps

remain unresolved. First, little empirical evidence has been produced to examine the deployment of chatbots in government

contexts. Second, existingresearch does not distinguish clearly between the drivers of adoption and the determinants of

success and, therefore, between the stages of adoption andimplementation. Third, most current research does not use a

multidimensional perspective to understand the adoption and implementation of AI in government organizations. Our

study addresses these gaps by exploring the following question: what determinants facilitate or impede the adoption and

implementation of chatbots in the public sector? We answer this question by analyzing 22 state agencies across the

U.S.A. that use chatbots. Our analysis identif‌ies ease of use and relative advantage of chatbots, leadership and innovative

culture, external shock, and individual past experiences asthe main drivers of the decisions to adopt chatbots. Further, it

shows that different types of determinants (such as knowledge-base creation and maintenance, technology skills and system

crashes, human and f‌inancialresources, cross-agency interaction and communication, co nf‌identiality and safety rules and reg-

ulations, and citizens’expectations, and the COVID-19 crisis) impact differently the adoption and implementation processes

and, therefore, determine the success of chatbots in a different manner. Future research could focus on the interaction

among different types of determinants for both adoption and implementation, as well as on the role of specif‌ic stakeholders,

such as IT vendors.

Keywords

Chatbot, artif‌icial intelligence, AI adoption, AI implementation, innovation adoption, innovation implementation

Introduction

Over the past decade, governments worldwide have sought to

employ artif‌icial intelligence (AI) to improve public services.

In general terms, AI has the potential to answer questions,

draft, f‌ill out, and search documents, route requests, and

translate text (Mehr, 2017). Among a range of different AI

technologies, chatbots serve as an innovative tool that gov-

ernments can use to address citizens’information and

service requests (Aoki, 2020; van Noordt & Misuraca,

2019). Also known as intelligent virtual assistants, digital

assistants, or conversational interfaces, chatbots are text-

based or voice-driven computer programs that use natural

language to converse with users (Dale, 2016; Shawar &

Atwell, 2007). Due to their 24/7 availability and eff‌iciency

in handling user questions, chatbots have been deployed

widely to supplement customer service in the business

sector (Dale, 2016).

In the same vein, governments around the world have

started to explore other ways to use chatbots for providing

information and public services (Aoki, 2020; NASCIO,

2020; Neumann et al., 2022). For instance, in the U.S.A.,

in 2017, only a handful of state, county, and municipal gov-

ernments had begun utilizing basic chatbots (Quaintance,

2017). By 2019, Gartner reported that chatbots had become

the top application of AI in the enterprise, and government

chief information off‌icers (CIOs) recognized them as one

of the most inf‌luential technologies for government

1

Rockefeller College of Public Affairs and Policy & Center for Technologyin

Government, University at Albany –SUNY, Albany, NY, USA

2

School of Public Policy, University College London, London, UK

3

ESADE-Universitat Ramon Llull, Barcelona, Spain

Corresponding Author:

Tzuhao Chen, Rockefeller College of Public Affairs and Policy, University at

Albany –SUNY, 135 Western Avenue, Albany, NY 12203, USA.

Email: tchen9@albany.edu

Article

American Review of Public Administration

2024, Vol. 54(3) 255–270

© The Author(s) 2023

Article reuse guidelines:

sagepub.com/journals-permissions

DOI: 10.1177/02750740231200522

journals.sagepub.com/home/arp

organizations over the next 5 to 10 years (Moore, 2019).

Furthermore, in 2020, more state, city, and county govern-

ments started to employ chatbots due to the COVID-19 pan-

demic, making chatbots a “necessity”for governments

(Miller, 2020). More broadly, a global survey conducted in

2021 indicated that chatbots were the most widely adopted

technology among government organizations. According to

the survey, 26% of the respondents had already deployed

chatbots. Moreover, an overwhelming 59% of the partici-

pants expressed their intention to deploy a chatbot within

the next three years (Gartner, 2021). In essence, chatbots

are signif‌icantly altering the way in which citizens interact

with the public sector (Androutsopoulou et al., 2019).

Despite the importance of this topic, research on the use of

chatbots in public administration is very scarce. Some studies

have focused on the technical aspect of government chatbots,

namely, how to design and construct databases, user inter-

faces, and technical architecture to ensure a better response

to citizens’inquiries (Androutsopoulou et al., 2019;

Lommatzsch, 2018). Other studies have shed light on the

organizational, institutional, and contextual factors that

affect the use of chatbots (Baldauf & Zimmermann, 2020;

Petriv et al., 2020; Wang, Lin, et al., 2022; Wang, Zhang,

et al., 2022; Wright, 2021). Still, others have focused on

the typology of chatbot-mediated services and associated

public values (Makasi et al., 2020). However, the existing

studies of chatbots in the public sector have some limitations.

First, little empirical evidence has been collected about the

use of chatbots in public organizations. In addition, while

the current research has identif‌ied some factors that inf‌luence

the adoption and implementation of chatbots, it neither distin-

guishes clearly between the drivers of adoption and the deter-

minants of success nor uses a multidimensional perspective

to understand the deployment of AI in public sector

organizations.

In this context, the present study explores the use of chat-

bots from a holistic perspective. Our research is guided by

the following question: what determinants facilitate or

impede the adoption and implementation of chatbots in

thepublicsector?We treat adoption and implementation

as two distinct concepts, def‌ining adoption as deciding

whether to use a chatbot and implementation as the

process that follows the decision to adopt a chatbot, includ-

ing activities designed to fulf‌ill the intended goals. We have

consulted the existing literature on the value of chatbots and

technology-driven innovation adoption and implementation

to guide this empirical investigation of chatbots in public

organizations.

The remainder of this article is structured as follows. The

f‌irst section reviews the literature on chatbots in the public

sector and the next introduces frameworks of government

technology-driven innovation adoption and implementation.

These are followed by an explanation of the methodological

approach. The core section outlines and discusses the results

of the analysis. Finally, the conclusion summarizes the

article, reveals the potential limitations of the study, and

offers several suggestions for future research.

AI Chatbots in the Public Sector

Although scholars have characterized chatbots in myriad

ways, several primary features can be derived across

various characterizations. Typically, information retrieval

and natural language processing (NLP) serve as the techno-

logical backbones of chatbots (Aoki, 2020; Baldauf &

Zimmermann, 2020; Lommatzsch, 2018; Petriv et al.,

2020). Thanks to these technologies, chatbots can act as

experts, understanding spoken or written inputs and offering

useful and accurate responses to narrowly def‌ined requests

(Androutsopoulou et al., 2019; Lommatzsch, 2018; van

Noordt & Misuraca, 2019). In another essential feature, the

user interface is designed to make users feel that they are

interacting with humans, for example, by presenting the

ability to communicate in plain language, send greetings,

and manage small talk (Lommatzsch, 2018; Makasi et al.,

2020). In the private sector, chatbots are growing in popular-

ity in multiple industries, due to their strengths in reducing

the costs of handling customer requests while allowing for

more customized user experiences. Having witnessed the

potential benef‌its of chatbots in the business sector, govern-

ments around the world are seeking to leverage chatbots in

internal management by using them to respond to citizen

questions and deliver public services. For instance,

EMMA, deployed by the U.S. Department of Homeland

Security, provides users with specif‌ic immigration service

information, covering issues related to green cards and pass-

ports. Similarly, Alex, the Australian government chatbot,

and Botty Bonn, a German city government chatbot, enable

citizens to pay their bills and taxes. Finally, citizens in

Singapore can use the “Gov.sg”chatbot to raise and track

the status of complaints about public service delays.

Nonetheless, despite increasing interest in the use of chat-

bots in public organizations in recent years, few studies have

attempted to investigate the phenomenon in depth; these

approach government chatbots from various perspectives.

First, most studies discuss the functionalities that chatbots

provide in public organizations. One way to differentiate

chatbot applications is by the complexity of the tasks that

they handle. On a basic level, chatbots can provide informa-

tion in response to a user’s service query without the need for

user authentication (Makasi et al., 2020; van Noordt &

Misuraca, 2019). All of the chatbots examined in a compar-

ative case study in Germany, conducted by van Noordt and

Misuraca (2019), engaged with this level of information

service, providing straightforward answers to often-asked

questions. Moving further, public organizations can employ

chatbots to offer transactional services (van Noordt &

Misuraca, 2019). To enable a transaction to happen, the

system requires the user’s personal information to personal-

ize a response (Makasi et al., 2022). The most complicated

256 American Review of Public Administration 54(3)