1 Introduction

AI competence is a universal objective in educational settings. Using technology to improve learning performance in higher education is a decision that is based on the knowledge to use AI. AI tools, such as Grammarly, ChatGPT, Gemini, and Copilot, etc. have been designed to help learners and educators in higher education as innovations that are frequently built for teaching and learning [1, 2]. Students’ individual choices, previous knowledge, and contextual factors can all have an impact on how AI tools are incorporated into educational institutions and how students can be influenced by using these tools in an online learning setting [3, 4]. Educational encouragement and enthusiasm for the learner’s future requirements could be equally important. Investigations into the utilization of popular social networks, such as Twitter and Facebook, as well as the application of broad and specialized technologies, have been conducted in higher education. However, there have not been numerous studies examining the use of AI tools for improved learning performance from the perspective of online learning.

The successful use of AI tools differs depending on the technology and its application, but it often includes improved access to information, practical application of complex ideas, approaching educational materials, and exploration of new content [5]. The overall upsides to using AI in online learning for learners involve boosted learning performance, AI self-efficacy regarding available material, collaboration of new knowledge for critical analytical thinking, and social engagement [6, 7]. Furthermore, academic learners who utilize AI tools frequently show increased flexibility, retention of understanding, content engagement, and active learning. However, there are numerous advantages to adopting AI tools and few instances when these developments do not appear to aid students in higher education. Higher-education students have expressed discomfort and anxiety about transitioning to these AI tools, and a few learners choose alternative and common approaches to learning. Considering previous interactions that influence student participation and satisfaction within the setting of higher education, instructors should analyze the use of different AI tools to discover possibilities for accessing obstacles and tailored requirements [8].

AI may also provide studies and assessments; it could reduce dependence on teachers, whereas some argue that it encourages communication between people. However, not everyone agrees on these points [9,10,11,12]. The benefits comprised online learning, convenience, and easy access, although the drawbacks were ineffectiveness and issues protecting reliability in education. The suggestions put forward included educating teachers on how to use digital tools and developing course designs with less intellectual burden and more interactivity [11, 13].

The key issue of this study is: What factors influence students’ continuous use of AI in an online learning setting to improve their learning performance? To answer this question, this research will (1) identify the main factors that influence students’ desire to use AI tools to improve learning performance and (2) examine how constructs such as AI in online learning, AI self-efficacy, and collaboration interact to influence higher education students’ learning performance. By achieving these objectives, the study seeks to provide AI-based insights that can help HEIs create AI implementations that are more effective and tailored to the needs of students, resulting in better learning performance.

Difficulties comprised instructors’ uncertainty about utilizing AI tools for observing and evaluating the learners’ achievement and a belief that the application fails to comprehend the complications and significance of collaboration. As a result, there is additionally satisfactory proof of the educational disadvantages of difficult assessment [14, 15].

The study was conducted by Ref [16]. in Poland in which the findings enabled the investigators to identify factors that are critical for the effective execution of creative use of AI tools in an online learning environment, as well as emerging opinions. Most of the benefits of online learning have a favorable association with the views of learners of the suitability of obtaining information and are not associated with students’ insights. Hence, the advantages of integrating students’ learning performance in HEIs encouraged researchers to investigate the factors influencing the use of AI tools in online learning surroundings with the help of social constructivism theory in the present study.

As an effect of the social constructivist approach to teaching and learning relevant topics, a research investigation was required to observe the usefulness of the social constructivist process in successful students’ learning. This interaction allows learners to react to concerns posed by the instructor, their classmates, or observers. Additionally, reflection allows learners and educators to analytically assess and analyze their personal and group development. This inspires every learner to organize their information, relate educational items to additional curriculum components, and predict upcoming educational experiences [17,18,19].

Considering the extensive factors that can influence students’ learning performance and the ongoing use of AI tools in online higher education, and the difference in functionality, and educational settings in which they can be implemented, there will be an immediate need to develop and assess the current condition of advancement in this field, as well as recognize the significant potential for further investigation in this field of study.

2 Literature review

2.1 Implementation and use of AI tools in online higher education

Currently, educational organizations rely on conventional learning processes and face-to-face engagement in classrooms worldwide. While most educational fields have originated with unified education, many still use traditional methods [20]. The incorporation of AI into HEIs can revolutionize the educational process, but it also raises concerns regarding ethics and privacy issues [21,22,23]. The emergence of AI innovation is causing important changes in individual employment interactions. Educational organizations have been required to reassess their methods of instruction after the COVID-19 pandemic. HEIs have implemented online learning approaches as an alternative to classroom instruction. Therefore, online learning has developed as an important educational technology skill by implementing and using AI tools in facilities of higher education. Still, the efficacy of online informative tools is primarily contingent on learners’ willingness to use these platforms [24, 25].

On the other hand, integrating difficulties are linked to the requirement to teach educators about the implementation of technological methods and construct instructional materials involving reduced psychological strain and improved interactivity. Different disciplines, such as STEM (Science, Technology, Engineering, and Mathematics) education, have problems, resulting in inadequate achievement among students, particularly in rural locations. Prior investigation has demonstrated that mobile learning can help to address the problems associated with the learning performance of students within rural locations [26].

AI has been presented for improving the writing abilities of learners, self-esteem, and knowledge concerning academic ethical behavior in higher education. Fortunately, many students were worried about the potential properties of innovation, analytical thinking, and professional writing behaviors. This investigation emphasized an equitable strategy for integrating AI into higher education, in which AI works alongside humans, and discovered common AI tools used by students of higher education in online learning environments. Consequently, the study revealed AI’s essential role in facilitating an educational learning environment. While keeping human originality and intellectual curiosity, emphasizing the significance of upholding an equitable combination to maintain human creativity and thoughtfulness in the realm of academia [27, 28]. The use of AI in informative settings enables teachers to evaluate data taken from higher education systems, discover trends regarding pupil actions, and create solutions for learners in trouble. Learning online has struggled to anticipate students’ learning performance due to a lack of use of educational processes, evaluative data, and an accurate estimation of statistical connections between components and outcomes [29,30,31]. To solve these challenges, this research builds an AI-based approach to better educational results for students that utilize the method of learning and collective information.

2.2 Influence of AI on improved learning performance in higher education

In the current study, AI tools are defined as an explanation for the dependent variable learning performance. AI tools are technologies used in higher education to facilitate learning and instruction. AI usually includes ideas, innovations, and developments that are used in a variety of educational contexts and have the potential to redefine or transform education. The technologies involved are comparatively new; however, they may not be used to their maximum benefit, highlighting the demand for additional research. The capabilities of AI have still not been fully realized in education. Additionally, according to the exceptional ethical considerations raised by AI, educators must be able to evaluate AI-based conclusions [23, 32, 33].

Previous studies investigated the influence of higher education students’ educational behaviors. Investigations assessing higher education students’ willingness and utilization of integrating technological advances in educational settings, including online learning [34], motivation of students [35]; mobile technologies [36, 37], and collaboration [38]. Neither of the previously mentioned studies looked at higher education learners’ intentions to utilize AI tools for improving their learning performance, which is the attention of the existing research.

3 Study framework and hypotheses development

3.1 Theoretical framework: social constructivism theory

Prior studies on the influence of students’ learning performance in higher education on their acceptance and usage of AI tools have been conducted from a range of views, including the social constructivism theory. Constructivism and social constructionism have regularly impacted the field of academic psychology using broader constructivism. In the current work, social constructivism differs from construction in that it emphasizes how individuals interrelate conceptually with the development of insight, whereas social construction asserts that expertise and its importance are historically and socially acquired by human relationships and behaviors. Under the social constructivism model, researchers explored the factors AI in online learning, AI self-efficacy, and collaboration, which helped to improve the learning performance of higher education students (see Fig. 1).

Fig. 1
Fig. 1
Full size image

Conceptual Framework for learning performance

3.2 Research hypotheses

The current research investigation examines higher education students’ influence to employ AI tools for improving learning performance in an online educational environment. Based on the social constructivism theory, the following hypotheses were developed for analysis:

3.2.1 AI in online learning

In the last several decades, innovation has drastically revolutionized the teaching and learning environment worldwide. Researchers predict that AI will drive the next major technological shift in how people communicate, work, deal in businesses, and acquire new knowledge. Prior studies about the intent to use AI technology have an optimistic influence on education quality, learning evaluation, and career opportunities. According to Ref [39]., the engagement of AI technology to improve educational performance has gained popularity around the world, eliminating the time and geographical problems associated with conventional education. However, with the numerous advantages, maintaining learners on digital platforms is difficult. The execution of the learning experience consists of a course taken online using a combination of online communication tools such as social networking sites, messengers, and video conferencing, which is intended to satisfy pupil requirements and competencies [40,41,42]. In the present investigation, factors inducing the use of AI tools for improving learning performance would be high if the learners supposed that the utilization of these tools would be helpful in their educational approaches. The Ref [43]. confirmed the benefits of environmentally friendly practices and the use of online educational platforms. In this study, participants’ intention for using AI tools to positively influence learning performance suggested that AI in online learning surroundings would be beneficial to self-efficacy and collaboration in their educational practices. According to the information given, the following hypothesis was developed:

Hypothesis 1

AI in online learning positively influences higher education students’ learning performance.

3.2.2 AI self-efficacy

The present investigation examined the associations between AI in online learning, AI self-efficacy, collaboration, and learning performance, in which past knowledge and satisfaction differ with gender and age. The present investigation emphasizes learners’ knowledge management of self-efficacy and examines how much learners believe they can manage and analyze the substantial quantity of material accessible through the Internet [44, 45]. With the advancement of AI, researchers, educators, and administrators must recognize how someone feels confident in using AI tools. Assessing individuals’ future actions concerning the growth of AI tools is equally important. Considering the increased emphasis on AI, an appropriate rating system for evaluating AI self-efficacy has still to be established [46]. Knowledge-based self-confidence and motivation for learning are both thought to serve crucial parts in the learning performance of students. The earlier indicator is thought to predict student learning performance, whereas the subsequent is thought to perform a crucial role in fostering AI self-efficacy.

Many learners try while studying but recognize the impact on their knowledge, self-efficacy, and inspiration for learning experiences. AI innovation has grown rapidly, improving its function and depth in higher education. This research explores the factors influencing learning performance and the constant use of AI tools among higher education students [47, 48]. Students’ information literacy and AI self-efficacy were predicted by personality traits such as readiness for diversity, risk and difficulties, and determination. The investigation indicates that users’ AI self-efficacy and collaboration can improve learning performance in AI in online learning arrangement and effectiveness. Therefore, HEIs prioritize online learning as an ongoing learning plan. According to the information given, the following hypothesis was developed:

Hypothesis 2

AI self-efficacy positively influences higher education students’ learning performance.

3.2.3 Collaboration

Accessibility and accuracy are critical in current interactive AI tools and educational environments. These tools have disrupted the conventional educational context, notably by using the collaborating strategy for solving problems and accelerating the learning process. The framework is essential to the incorporation of technological advances into educational activities to understand the prospective consequences of new digital technology [49, 50]. Social learning analytics are predominant in institutional and entirely online environments, with the analysis of social networks serving as the primary statistical tool. Many social and educational analysis researchers tried to understand the ways students learn and used the social constructivist approach to understand their educational habits [51, 52].

Learning together with collaboration is one approach to increasing placement for students’ capabilities and possibly enhancing all aspects of educational engagement. The implementation of AI for collaboration has been recognized in learning performance, cooperative accomplishments, learning content, social relationships and techniques, attitudes and feelings, conversations and discourse move around, and student traits and attitudes [53]. The rising popularity of technological resources for learners and educators for learning highlights the prominence of assumed consequences of technological-mediated approaches. Following the concepts of social cognition and flexible organization meta-theory, the successful outcomes revealed a strongly favorable effect that collaborative learning facilitated [54]. In the present investigation, participants’ goal to employ AI tools to influence learning performance positively suggested that collaboration would benefit their practices in education. According to the information given, the following hypothesis was developed:

Hypothesis 3

Collaboration positively influences higher education students’ learning performance.

3.2.4 Learning performance

The element that may limit learner-teacher connections and, thus, learning performance is the intention of using virtual learning in computerized exercises by using AI. The advantages seen in past research may take longer to reproduce in the initial users’ curricula. This investigation underlines the implication of persistence in investigating the implementation and consequences of educational changes as they are adopted [55, 56]. The participants in the study had better experience with the digital platform, utilized it regularly, and found it easier. The results imply that digital AI technology can improve commitment and learning performance in higher education settings.

Additional studies could investigate techniques to improve the use of AI tools and flexibility as well as the feasibility of utilizing educational innovations in HEIs [35, 57]. The outcome of this study underlines the need to constantly examine the execution and impacts of learning improvements as they are adopted [56]. The utilization of data collected in this study has been generally acknowledged as a means of coordinating operations throughout managerial levels and improving the quality of education. The present research suggests a broad framework to recognize the factors that stimulate the learning performance of higher education learners’ aspirations to use AI tools for the improvement of their learning results in an online learning environment.

4 Methods

4.1 Participants

The participants were HEI students enrolled in various undergraduate and postgraduate degrees in universities in Jharkhand, India. These participants were chosen for this study because they are capable of being agents in the implementation and usage of AI technology at universities in future years, and they have prior experience using these AI tools in their learning environments. The initial representation included 918 higher education students from several disciplines such as MBBS, Allied Health Sciences, Nursing, Education, Library & Information Science, management, and commerce fields chosen to represent their wide range of skills. Participants in the research were chosen over a simple random sampling procedure. This sample approach was chosen since the participants were convenient and comfortable with the use of AI tools. Nevertheless, simple random sampling is a popular method of sampling in quantitative investigations using instrumentation for surveys. Simple random sampling is seen to be more advantageous in homogeneous and uniformly selected groups. In this selection approach, everyone will have a chance to get involved in the research. There are both advantages and disadvantages to using simple random sampling. It ensures impartial, accurate, and equal probabilities of the population; yet, it can be tedious, rarely supported by a publicly available list of people, and difficult when the population is varied and far spread [58, 59]. Before collecting the data, prior approvals were obtained from the concerned authorities of each institute. For the present study, the data were collected manually by visiting the respective institutes. Printed survey forms were distributed to all accessible students. They were briefed on the purpose of the study which was also mentioned at the top of the questionnaire. We ensured that all students get equal opportunities to be part of the survey. We followed the same procedure in all five selected Indian higher education institutes.

Table 1 presents the demographic data of the participants (N = 918). In terms of gender distribution, 69.28% (636 participants) were female, while 30.72% (282 participants) were male. The higher number of female candidates is due to the data collected by two women’s universities out of five selected universities. Regarding age groups, most participants, 52.40% (481 individuals), were aged between 21 and 25 years. This was followed by 41.94% (385 individuals) in the 16–20 years age group, 5.12% (47 participants) aged 26–30 years, and a small proportion, 0.54% (5 participants), aged over 30 years. When considering education levels, most participants, 86.06% (790 individuals), were pursuing undergraduate studies, while 13.94% (128 individuals) were at the postgraduate level. This data provides a comprehensive overview of the participant demographics.

4.2 Survey instrument and data collection

This investigation employed a quantitative research strategy. An offline survey was established to identify feasible pupil attitudes regarding the influence of using AI tools in an online learning environment. The investigation followed the social constructivism framework. The items that influenced the structures were selected from earlier investigations. The survey consists of 14 items composing four constructs (see Table 2). The questionnaire was rigorously constructed using a methodical procedure that included scale transformation, interaction with subject matter experts, and revision to meet specific requirements. Measurement scales were modified from multiple proven sources in the fields of technology acceptability and organizational behavior research. These scales were carefully tailored to reflect the unique dynamics of AI tools in the framework of online learning. To verify the items that comprise the study, we reached an understanding of > 70% agreement among respondents following the Q-sort. This process increased reliability and provided comprehensive knowledge into the elements impacting learning confidence by using AI technology in the online learning settings.

Participants comprised individuals enrolled in higher education across various disciplines and degree levels, all of whom were familiar with the use of AI tools in an online learning environments. Responses were measured using a five-point Likert scale ranging from “strongly disagree” (1) to “strongly agree” (5). The measurement instrument was adapted and refined from prior validated studies (e.g., 73, 74). Comprehensive assessments of validity and reliability were conducted to ensure the instrument’s robustness. All 14 measurement items exhibited outer loading values above 0.70, indicating strong reliability by established thresholds.

The research instrument was divided into two halves and included fourteen items. The first section comprised three pieces of personal information, including gender, age, and education level. The second portion contained fourteen items with four sections of construct AI in online learning (AIOL), AI self-efficacy (AISE), Collaboration (CL), and Learning performance (LP). The AIOL constructs contained four items, each taken from [33], the AISE and CL constructs contained three items [60, 61], and the LP construct contained four items [62], respectively, and revised the items by modification. To fulfill the information needs, this study used simple random sampling. The above-mentioned sampling procedures were judged to be the most appropriate [63]. The estimate of sample size for this investigation is extremely challenging due to the large quantity of higher education students in the intended population. However, with the help of specialists from other fields, researchers have attempted to determine a suitable number that is deemed true to represent those specific locations in the state of Jharkhand. Given the difficulty level, the overall sample size for this investigation was 918 respondents from various universities in Jharkhand, India.

4.3 Data analysis

This paper examined a total of 918 participants from higher education. Smart PLS version 4 was used to examine all four variables for statistical significance and to perform Partial Least Squares Structural Equation Modeling (PLS-SEM). Using the survey, researchers investigated the intention of higher education students to utilize AI tools in an online learning environment. First, confirmatory factor analysis (CFA) was used to evaluate the measurement model’s associations with latent and observed variables. Second, a structural component of the PLS-SEM was calculated to determine the associations between the latent variables. The measurement model and its results are shown before being followed by the results for the structural model.

PLS-SEM is a popular multivariate analysis technique for constructing variance-based structural equation models, especially in the social sciences. PLS-SEM is widely used in project management research due to its ease of use with complicated models and lack of need for data normality distributions or large sizes of samples [64, 65]. The present study investigated the impact associated with organizational online readiness on competence in the digital age, involving a strong analytical framework that can deal with complicated interactions and numerous variables. We used PLS-SEM to evaluate study constructs and test hypotheses in a single phase.

PLS-SEM consists of two primary stages: the first stage is the measurement model, and the other stage is the structural model. The reliability and validity of the constructs are evaluated as part of the measurement model assessment process. Cronbach’s alpha and composite reliability were used to determine internal reliability, with values of 0.70 deemed suitable in exploratory research examinations [66]. The average variance extracted (AVE) was used to assess convergent validity, with an appropriate value of greater than 0.50. The discriminant validity was determined by comparing the latent variables’ correlations to the square root of AVE and the Heterotrait-Monotrait (HTMT) ratio. An HTMT ratio under the minimum value of 0.85 indicates that the two structures under consideration are distinctive and not replaceable [67, 68]. The structural model analysis requires finding relevant path coefficients employing a technique called bootstrapping of 5000 subsamples. Bootstrapping generates VIF, R2, t-values, and p-values for model parameters, facilitating hypothesis testing.

Table 1 Participant demographic data
Full size table

5 Results

Survey research on higher education student learning performance and continuous use of AI tools was conducted in an offline higher education survey. The findings of the measurement model are offered first, followed by the structural model.

5.1 Measurement model

5.1.1 Reliability and validity of data surveyed for the entire questionnaire

Table 3 highlights the construct reliability and validity of various constructs in this present study, focusing on both independent and dependent variables. The constructs include AI in online learning (AIOL), AI self-efficacy (AISE), Collaboration (CL), and Learning performance (LP). Each construct is assessed through a specific number of items. Reliability is measured using Cronbach’s Alpha, Composite Reliability (CR), and Average Variance Extracted (AVE). Notably, all constructs demonstrate acceptable reliability with Cronbach’s Alpha values ranging from 0.681 to 0.764 and composite reliability scores above 0.8. Furthermore, the AVE values exceed 0.5, affirming sufficient convergent validity.

Table 2 List of constructs and items
Full size table

5.2 Structural model and hypothesis testing

These outcomes demonstrate sufficient reliability and validity for each construct, with AIOL, AISE, CL, and LP showing satisfactory Cronbach’s Alpha values above 0.7. Table 4 outlines the discriminant validity of the constructs based on the Fornell-Larcker criterion. Each construct’s diagonal element represents the square root of its AVE. The discriminant validity is verified as these values (e.g., AIOL: 0.754, AISE: 0.781, CL: 0.813, LP: 0.766) are higher than the corresponding inter-construct correlations, ensuring distinctiveness among constructs. This reinforces the robustness and validity of the measurement model employed in the analysis.

Table 3 Construct reliability and validity
Full size table
Table 4 Discriminant validity—Fornell-Larcker criterion
Full size table

Multivariate techniques are classified into dependence and interdependence techniques. SEM is based on two well-known multivariate methods: multiple regression analysis and factor analysis, making it a unique combination of both approaches. Partial least squares structural equation modeling (PLS-SEM) is a multivariate statistical technique for examining complicated relations between variables. Despite its expanding use over decades, PLS-SEM continues to be overlooked in idle research. Figure 2 explains the hypothesized paths of the structural model. All the path coefficients are positive and statistically significant as specified by the high critical ratios and significant p-values, demonstrating that each of these factors has a meaningful impact on learning performance in the online learning setting.

Table 5 displays the path coefficients related to the interactions between variables AIOL, AISE, CL, and LP. Each relationship is evaluated using the original sample value (O), sample mean (M), standard deviation (STDEV), T statistics, and P values. The findings indicate that all the relationships (AIOL→LP, AISE→LP, CL→LP) are statistically significant, as their P values are less than 0.05. Moreover, the respective T-statistics for these relationships confirm the strength of the results. For the association among the variables AI in online learning (AIOL) and Learning performance (LP), the original sample path coefficient is 0.390, with a sample mean of 0.391 and a standard deviation of 0.063. The T statistics are 6.201, and the p-value is 0.000, representing concrete statistical implication, and Hypothesis 1 is supported. The relationship between AI self-efficacy (AISE) and LP shows a path coefficient of 0.189, with a sample mean of 0.192, a standard deviation of 0.070, a T statistic of 2.691, and a p-value of 0.007, representing concrete statistical implication and Hypothesis 2 is supported. This also indicates a significant relationship, and Hypothesis 2 is supported. Lastly, the relationship between Collaboration (CL) and LP presents a path coefficient of 0.352, with a sample mean of 0.351 and a standard deviation of 0.055. The T statistic is 6.447, and the p-value is 0.000, confirming the Hypothesis 3 is supported. These results highlight significant positive relationships between the independent variables (AIOL, AISE, CL) and dependent variables (LP), underscoring the implication of these factors in inducing learning performance.

Table 5 Path coefficients-mean, STDEV, T values, p values, and hypotheses supported results
Full size table
Fig. 2
Fig. 2
Full size image

Path co-efficient with p-values of the structural model

6 Discussions

The present study aims to determine the relationship between continuous uses of AI technologies, AI self-efficacy and collaborative learning and their impact on learning performance in an online learning environment. The findings of this study show significant empirical evidence to the hypothesized connections between AI in online learning (AIOL), AI self-efficacy (AISE), and collaborative learning (CL) and learning performance (LP) in higher education online environments. AIOL (β = 0.390, t = 6.201, p = 0.000), CL (β = 0.352, t = 6.447, p = 0.000), and AISE (β = 0.189, t = 2.691, p = 0.007) all have statistically significant effects on learning performance (p < 0.05). These findings demonstrate that every factor individually improves students’ learning performance, and all hypotheses are validated.

The present study investigated higher education students’ (N = 918) influence on using AI tools to improve learning performance in an online learning setting. Although several of these innovations originated and evolved in the field of education, AI tools could increase students’ learning performance, self-efficacy and collaboration. This study generated three hypotheses based on social constructivism theory (H1, H2, and H3). The results showed that all three hypotheses were accepted.

Like other studies related to students’ influence on using AI tools in higher education comparable to the current finding, beneficial advantages when accepting an AI tool, it is critical to use this knowledge for subsequent module delivery services to assist the various student populations in the future [69]. Assessments performed online are presently employed for assessing the academic achievement of pupils by using AI tools in learning, such as online assessments, presentations, report submissions, etc., all of which have a substantial effect on the learning performance of higher education students. Despite the advantages, including convenience, and ease of access, the disadvantages were inefficiency and concerns with educational reliability. The ideas involved educating teachers a way to use AI tools and creating instructional programs that are less intellectually demanding and more interactive [11, 13].

This finding has important implications, including motivating higher education students who have effectively explored the use of an AI tool to assist other learners who might wish to use an AI tool. AI will not merely enhance present teaching and learning methods and procedures but will radically alter educational experiences. Focusing on contemporary literature and research into AI tools in an online learning environment among higher education students, this paper investigates potential future consequences for learning. We recognize the revolutionary and transformative possibilities for influence to employ AI tools in higher learning and analyze how this links with the ethical and interpersonal elements of education that are discussed.

New literacy skills and technological breakthroughs are currently discussed among students who have a strong desire to transform and increase educational performance. Simultaneously, global education strategies demand instructors to transition from knowledge receivers to facilitators. This implies that educator training also needs to evolve, and additional training centered on effective instruction is crucial to make a difference [70]. Investigating the analysis aspects that affect the usage of AI tools in online learning environments in higher education, there were also conflicting results [28, 71, 72].

Believing in the use of AI technology is influenced by the combination of AI self-efficacy, and collaboration with educators and peers, and the availability of individual and institutional assistance. These elements work together to improve the learning performance of higher education students in online learning contexts. Before implementing or incorporating AI technology into education, information based on evidence might help students believe it is easy to use. A positive interaction with educational development may influence the overall use of AI technologies [73]. The indicators of skills for the 21 st century have identified the most important competencies required for continuous development by all citizens, such as instructors and learners. In this regard, learning plays a significant part in making sure that citizens obtain the skills they need [74, 75].

The contemporary setting requires improved educational methods that allow students to take part in their learning process. Technology influences how teaching is provided and how knowledge is discovered and shared. Prior, most recently, instructional techniques emphasized memorizing as a key learning capacity. Nowadays, technology has altered the approach to learning as well as accessibility to content. Learning is widely available on the internet, largely for free and with ease of access. Reading, collaborating, paying attention, and engaging in have become essential abilities in learning. Portable electronic devices have developed into a comprehensive collection of uses, support, and assistance for educational institutions [7].

Contrary to a few of the answers of the previous investigation, researchers found that the educational atmosphere typically causes many kinds of anxiety created by the capacity of students to employ technological advances and modify their negative emotions of being observed and dismissed constantly. The impact of AI ethical behavior awareness perceptions of the use of AI in online learning on the intentions of improved learning performance is required to introduce a favorable mindset toward offering these innovative educational materials [48, 76].

Additionally, higher education students participating in the present investigation sought to enhance their learning performance through AI basically in the online learning atmosphere. The confidence in preparing for an online learning environment by accessing the internet and computers had a transmitted influence not only on the acceptance of AI tools but also on learning views and fulfillment [39]. The outcomes of this investigation are useful for higher education students with the effect of utilizing AI in an online learning environment, which particularly emphasizes learning performance with AI self-efficacy and collaboration.

7 Implications

The theoretical implication of the study demonstrates that the continuing use of AI technology, AI self-efficacy, and collaboration all improves learning performance in online higher education by employing social constructivism theory. These aspects have a major influence, thus HEIs should integrate AI tools into online platforms, give instruction to enhance students’ learning performance in using these AI tools, and foster collaboration through collaborative tasks and AI-supported tools. These tactics not only enhance learning performance but also improve student involvement and motivation [35]. The findings also advocate for comprehensive educational strategies and additional research into the long-term effects of these variables in many situations and disciplines. The study offers practical implications for HEIs, emphasizing the need to implement structured training programs for lecturers to facilitate effective technology integration [42]. Furthermore, it underscores the importance of developing institutional policies that promote and support the widespread use of AI tools among higher education students.

8 Conclusion, limitations, and future research

The higher education sector strives to involve students in interactive education to strengthen expertise and enhance learning performance. AI technologies can be an essential component of active learning activities since they increase learners’ motivation to learn. There has been increasing evidence that the positive effects of AI tools include greater learner participation and achievement [35, 77]. Although, it is uncertain how students intend to incorporate the use of these AI tools. The existing study revealed that in higher education, learners desire to use AI technologies in their education within the context of social constructivism theory. The results on what aspects influence students’ learning performance when using these innovations might help instructional designers, teachers, and educational authorities plan the combination, deployment, and execution of AI tools for growth in higher education.

A main limitation of this investigation transmits to the demographic characteristics of the study; a greater number of respondents were engaged in MBBS, Allied Health Sciences, Management, and Education, which may indicate that they would be more prepared for utilizing AI technologies compared to individuals from various fields. Future studies could be conducted to include more disciplines. Second, the scale itself could be adjusted in subsequent research to incorporate more components in aiding flexibility-related conditions that are more common while using AI tools. Future studies should look at how higher education students responded to this assessment to identify their actions and intentions to use AI tools in online educational settings. Additionally, as these modern AI tools increase in popularity and novel tools such as ChatGPT, copilot, and Gemini are introduced, as well as AI search options in various online databases consisting of text, graphic synthesis devices, and large language-generated chatbot examples, existing elements that constitute AI tool use should be redefined.

Third, the present study was conducted in India and was conducted among higher education students only. Future studies should comprise participation from other countries to ensure the wider generalizability. Fourth, a predominantly female sample may introduce gender-based perspectives that are not fully representative of the broader student population. Responses may reflect attitudes, experiences, or learning outcomes that are more salient among female students, particularly in domains like AI self-efficacy, collaboration, or learning performance of educational technologies. To enhance the generalizability of findings, future studies should adopt stratified and regionally diverse sampling techniques that ensure a balanced representation across gender, academic level, and geographic location. Fifth, concentrating on urban regions potentially overlooks socio-economic, technological, and educational disparities experienced by students in rural or semi-urban areas. This may affect constructs related to accessibility, digital readiness, and environmental support in online learning. Including participants from rural and semi-urban areas would allow for a more comprehensive understanding of contextual factors influencing AI-based learning in higher education. Grounded on the results of this investigation, further studies should focus on building an AI-centered acceptance model for learning that is ideal for assessing views and discovering factors that encourage students’ approaches to the use of AI tools in an online learning setting.