📄 social science

Influence of users’ perceived self-efficacy and perception of generative artificial intelligence chatbots’ problem-solving ability on chatbots’ value and intention to continue using them

This study empirically demonstrates that users' self-efficacy and their perception of generative AI chatbots' problem-solving ability enhance both utilitarian and hedonic perceived values, which subsequently drive the intention to continue using these chatbots.

Original authors: Shih-Tse Wang, Yun-Xiu Cai

Published 2026-09-09

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Shih-Tse Wang, Yun-Xiu Cai

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the bustling world of modern technology, a new kind of digital assistant has emerged, one that does not just retrieve information but creates it. These tools, known as generative artificial intelligence chatbots, can write stories, generate images, and compose code in response to simple requests. They have become so common that keeping people interested in using them has become a major challenge for the companies that build them. To understand why people stick with these tools, researchers often look at a concept called perceived value. This is simply the idea that a person decides whether something is worth their time based on two main feelings: how useful it is for getting a job done, and how much enjoyment it brings. While many studies have looked at what makes people like these tools, fewer have explored how a user's own confidence and the tool's ability to solve problems shape those feelings.

A team of researchers at National Chung Hsing University set out to investigate this specific relationship. They wanted to know if a person's belief in their own ability to handle tasks, known as self-efficacy, and their belief in the chatbot's ability to fix issues, influenced how much value they found in the technology. The team gathered data from 612 people in Taiwan who had used these chatbots. These participants filled out a detailed survey about their experiences, rating their confidence in solving problems, how well they thought the chatbot could solve problems, and how much they enjoyed or found value in using the tool. The researchers then analyzed these responses to see how these different factors connected to one another.

The study revealed a clear chain of cause and effect. First, the researchers found that when users felt confident in their own problem-solving skills, they were more likely to see the chatbot as both useful and enjoyable. Similarly, when users believed the chatbot itself was capable of solving complex issues, their perception of its value increased. This held true for both the practical side of using the tool, such as saving time or money, and the emotional side, such as feeling entertained or satisfied. The data showed that these feelings of value were the direct drivers behind a person's decision to keep using the chatbot in the future. In fact, the feeling that the tool was useful had a much stronger influence on the decision to continue using it than the feeling of enjoyment did.

The researchers concluded that for these digital assistants to remain popular, developers and marketers need to focus on two things. They must help users feel more confident in their own abilities to work with the technology, perhaps by breaking down complex tasks into smaller, manageable steps. At the same time, they need to clearly demonstrate the chatbot's ability to solve real problems, showing users that even when they are stuck, the tool can provide a solution. The study suggests that when users feel capable and see the tool as a reliable problem-solver, they are far more likely to keep coming back. This understanding offers a practical path forward for improving these technologies, moving beyond simple features to focus on the genuine value they provide to the people who use them.

Technical Summary: Influence of Users' Perceived Self-Efficacy and Perception of Generative AI Chatbots' Problem-Solving Ability on Chatbots' Value and Intention to Continue Using Them

Problem Statement
Despite the rapid adoption of Generative Artificial Intelligence (GAI) chatbots, retaining users remains a critical challenge for marketers. While existing literature identifies various factors influencing continued usage—such as task-technology fit, perceived usefulness, and trust—there is a significant gap in understanding the specific antecedents that shape users' perceived hedonic (emotional) and utilitarian (functional) value of GAI chatbots. Previous research has largely focused on general benefits or specific AI features like anthropomorphism, but few studies have simultaneously examined how individual user characteristics (specifically self-efficacy) and technology characteristics (specifically problem-solving ability) interact to drive these value perceptions and, subsequently, the intention to continue using the technology.

Methodology
This study employs a quantitative research design grounded in Perceived Value Theory (PVT).

Data Collection: An anonymous online survey was conducted in Taiwan between March and April 2024. The target population consisted of individuals who actively used GAI chatbots. After filtering for validity, 612 usable responses were collected. The sample was predominantly young (83.9% aged 20–39) and highly educated (90% held at least a college degree), with the vast majority (88.4%) being ChatGPT users.
Measures: The study utilized a 7-point Likert scale to measure five latent constructs:
1. Perceived Self-Efficacy: Adapted from Gao et al. (2015), measuring the user's confidence in their own problem-solving capabilities.
2. Perceived Problem-Solving Ability of GAI: Adapted from Heppner and Baker (1997), measuring the user's belief in the chatbot's competence to solve problems.
3. Perceived Hedonic Value: Adapted from Wang (2022), assessing enjoyment, pleasure, and entertainment.
4. Perceived Utilitarian Value: Adapted from Wang (2022), assessing functional benefits, quality, and value for money.
5. Continued Usage Intention: Adapted from Wang (2012), measuring the willingness to keep using the chatbot.
Analysis: Data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) via SmartPLS 3.2.9. The study assessed common method bias (Harman's one-factor test and VIFs), internal reliability (Cronbach's alpha and Composite Reliability), convergent validity (AVE), and discriminant validity (Fornell-Larcker criterion and HTMT).

Key Contributions and Theoretical Framework
The paper extends Perceived Value Theory by integrating two distinct trait perspectives:

Individual Characteristics: The user's perceived self-efficacy.
Technology Characteristics: The GAI chatbot's perceived problem-solving ability.

The research proposes a model where these two antecedents influence the dual dimensions of perceived value (hedonic and utilitarian), which in turn drive the intention to continue usage. This approach moves beyond generic "perceived usefulness" to dissect the specific emotional and functional drivers of GAI adoption.

Results
The structural model analysis (N = 612) supported all six hypotheses with statistical significance (p < .001):

Antecedents to Value:
- Users' perceived self-efficacy positively influences both perceived hedonic value ( $\beta = 0.27$ ) and perceived utilitarian value ( $\beta = 0.26$ ).
- The perceived problem-solving ability of GAI chatbots positively influences both perceived hedonic value ( $\beta = 0.32$ ) and perceived utilitarian value ( $\beta = 0.28$ ).
Outcomes:
- Perceived hedonic value positively influences continued usage intention ( $\beta = 0.12$ ).
- Perceived utilitarian value has a significantly stronger positive influence on continued usage intention ( $\beta = 0.63$ ) compared to hedonic value.
Model Fit and Explanatory Power: The model demonstrated acceptable fit (SRMR = 0.059, NFI = 0.853). It explained 20.1% of the variance in hedonic value, 17.6% in utilitarian value, and 50.7% in continued usage intention.

Significance and Claims
The authors claim that this study provides empirical evidence clarifying the psychological and technological drivers of continued GAI chatbot usage.

Theoretical Significance: The study extends PVT by identifying specific antecedents (self-efficacy and problem-solving ability) for both hedonic and utilitarian values, addressing a gap in the literature regarding how user traits and technology traits jointly shape value perception.
Practical Significance: The findings suggest that GAI marketers should shift communication strategies from focusing merely on features ("what it is") to emphasizing benefits and value delivery ("what it delivers"). Specifically:
- Value Proposition: Marketers should highlight tangible savings in time, money, or resources to bolster utilitarian value, which the study identifies as the primary driver of retention.
- User Empowerment: To enhance perceived value, marketers should design strategies that boost user confidence (self-efficacy), such as breaking down complex tasks into manageable steps and providing structured assistance.
- Trust in Capability: Communication should emphasize the chatbot's ability to solve problems even when the user lacks an initial solution, thereby reinforcing the perceived problem-solving ability of the technology.

Limitations Noted by the Authors
The paper modestly acknowledges limitations, including the demographic skew toward younger, highly educated users in Taiwan, the dominance of ChatGPT users in the sample (limiting generalizability to other GAI platforms), and the use of a single-dimensional measure for problem-solving ability. The authors suggest future research should explore diverse user groups, different GAI platforms, and multi-dimensional aspects of problem-solving capabilities.

More like this