Chatbot design, trust and social influence key to digital government adoption in Sri Lanka

Trust emerged as a pivotal variable with a significant impact on perceived ease of use. When users felt secure and confident in the chatbot’s reliability, they found it easier to interact with. However, trust did not significantly influence attitude directly, suggesting that citizens may value performance and efficiency more than abstract trustworthiness when evaluating new digital tools.

As developing nations race to modernize public administration, Sri Lanka has taken a bold step into the digital frontier. A groundbreaking study recently published in Administrative Sciences titled “Factors Influencing AI Chatbot Adoption in Government Administration: A Case Study of Sri Lanka’s Digital Government” offers unprecedented insights into how citizens perceive, accept, and are influenced to adopt AI chatbot applications in e-governance.

Amid efforts to streamline government services, improve efficiency, and ensure 24/7 access for citizens, Sri Lanka is leveraging artificial intelligence as a cornerstone of its digital transformation strategy. The study, conducted using an extended Technology Acceptance Model (TAM), dives deep into what motivates citizens to accept AI-driven interfaces, especially chatbots, as an interface for interacting with public institutions. It combines behavioral science with data-driven analysis to uncover critical psychological, design-based, and social factors that influence adoption behavior.

What Makes People Say Yes to AI Chatbots in Government?

At the core of the study is the recognition that AI chatbot adoption in public administration hinges on more than just functional efficiency. The research introduces three external factors, trust, application design/appearance, and social influence, to the classical TAM framework. These external variables are evaluated alongside traditional TAM indicators like perceived ease of use (PE), perceived usefulness (PU), attitude (AT), and behavioral intention (BI).

Using a sample of 207 citizens and public sector employees across urban and rural regions of Sri Lanka, the researchers deployed a structured questionnaire and covariance-based structural equation modeling (CB-SEM) to test nine key hypotheses. Results showed that perceived usefulness had the strongest impact on user attitudes, and those attitudes in turn were the most decisive factor in shaping the intention to adopt AI chatbots.

Interestingly, perceived ease of use did not significantly influence attitude directly, contradicting traditional assumptions in TAM. Instead, it strongly influenced perceived usefulness, highlighting that usability supports perceived utility more than emotional affinity in Sri Lanka’s context.

Trust emerged as a pivotal variable with a significant impact on perceived ease of use. When users felt secure and confident in the chatbot’s reliability, they found it easier to interact with. However, trust did not significantly influence attitude directly, suggesting that citizens may value performance and efficiency more than abstract trustworthiness when evaluating new digital tools.

How Do Social and Design Factors Shape Trust and Usefulness?

The study found that social influence significantly enhances trust in AI systems but has limited direct impact on perceived usefulness. This suggests that endorsements by peers or community leaders can build foundational trust in emerging technologies, even if they don’t immediately translate to perceptions of utility. This is particularly relevant in Sri Lanka’s highly relational, community-oriented society.

Application design and appearance also significantly influence perceived ease of use. Respondents favored intuitive navigation, clear interfaces, and aesthetic simplicity. The design factor was especially critical in Sri Lanka’s multilingual, digitally diverse environment, where varying degrees of tech literacy exist across age groups and urban-rural divides.

From a structural standpoint, the model’s goodness-of-fit indices were robust across multiple validation techniques. Cronbach’s alpha scores exceeded 0.7 for all constructs, while AVE and composite reliability validated convergent and discriminant validity. Multicollinearity was ruled out, ensuring that the constructs functioned independently and precisely within the model.

What Should Policymakers and Designers Take Away?

The implications of the study are significant for digital governance, not just in Sri Lanka but across developing countries with similar sociotechnical landscapes. First, the research underscores that a technology's success in public administration depends on a combination of psychological confidence (trust), peer validation (social influence), and interactional comfort (design).

Policymakers are encouraged to prioritize transparent data practices and user education campaigns to bolster trust. Legal and regulatory assurances around data privacy and chatbot accountability can further reduce the risk barrier. Simultaneously, designers should focus on creating multilingual, mobile-first chatbot interfaces that feel familiar, responsive, and culturally aligned with user expectations.

Public sector strategists might also consider leveraging social capital, through endorsements by respected figures, testimonials, and community-driven pilot programs, to accelerate adoption. The study’s extended TAM framework offers a replicable model for other nations aiming to integrate AI into their e-governance systems.

The researchers emphasize the need to go beyond intention-based assessments in future studies. While this paper rigorously analyzes what makes people likely to adopt chatbot applications, actual usage metrics over time would provide even more granular insights into long-term behavioral patterns. They also highlight the need for periodic evaluation of different chatbot designs to assess impact variation based on interface evolution.