Development and psychometric evaluation of the artificial intelligence attitude scale for nurses

Abstract

Background

Since artificial intelligence is transforming healthcare, targeted interventions aimed at optimizing its integration and use in clinical settings requires the assessment of nurses’ attitudes towards AI.

Aim

To develop and validate an Artificial Intelligence Attitude Scale specifically for Turkish nurses.

Method

This methodological study was conducted between October 2024 and December 2024, and its sample consisted of 678 nurses working in Turkey. The item pool was developed through a comprehensive literature review. Data analysis included descriptive statistics, item analysis, and exploratory and confirmatory factor analyses, as well as assessments of convergent and divergent validity, correlation analysis, internal consistency reliability, and test-retest reliability.

Results

The content validity index for the items ranged from 0.85 to 1.00. Exploratory factor analysis revealed that the eigenvalues for four factors were greater than one, and these four factors accounted for 77.28% of the total variance. The scale demonstrated an acceptable model fit, with a goodness of fit index of 0.921 and a root mean square error of approximation (RMSEA) of 0.064. Cronbach’s alpha coefficients ranged from 0.93 to 0.95 across the subscales, indicating high internal consistency, with the scale showing convergent and divergent validity. In addition, the Artificial Intelligence Attitude Scale for Nurses was found to have high test-retest reliability. This study may offer valuable insights into nurses’ attitudes toward digital technologies, thereby informing the trajectory of digital transformation in healthcare services.

Introduction

Digital transformation is driving significant global and far-reaching changes within healthcare systems [1], with the potential to enhance the effectiveness of workflows across diagnosis, treatment, and patient care, thereby leading to improved outcomes for patients, healthcare personnel, and institutions. It also optimizes service delivery processes, fostering greater efficiency and effectiveness within the healthcare system [2]. The World Health Organization (WHO) emphasizes the critical importance of harnessing the positive potential of digital technologies to enhance and safeguard individual health [3]. In its global strategy report on digital health, WHO underscores the need for the development of infrastructure that supports the adoption and implementation of digital health technologies across all nations [4]. In this context, as a key component of the digitalization process, Artificial Intelligence (AI) technologies represent a major focus of investment for many countries, ushering in a new era of innovation and transformation within healthcare services [1, 5, 6].

WHO defines AI as “an area of computer science that emphasizes the simulation of human intelligence processes by machines that work and react like human beings,” as outlined in its Global Strategy on Digital Health 2020–2025 report [7]. The use of AI in healthcare is expanding rapidly worldwide, particularly in clinical settings [8, 9], with projections that global spending on AI in healthcare will surpass approximately 36 billion USD by 2025 [5]. Therefore, it is of great importance that nurses, who represent over 50% of the healthcare workforce, are equipped to effectively use AI technologies and understand their implications for the healthcare system, patient care, and community health [10,11,12]. Therefore, a digitally literate nursing workforce is essential for ensuring the delivery of safe, effective care, as well as for enhancing patient outcomes [13]. In this context, the guidance and collaboration of nurse managers are crucial to the successful adoption and integration of AI technologies within nursing practice [6, 10, 14]. Nurse managers play a key role in ensuring that AI technologies are effectively utilized by nurses and contribute to reducing their workload. They are responsible for planning areas in which AI can be implemented—assessing potential risks, incorporating AI into in-service training programs, and addressing patient safety and ethical concerns related to its use [5, 6, 10, 14,15,16].

Research indicates that the integration of AI algorithms into nursing applications is progressively expanding. As nurses’ awareness of AI grows, its impact on improving patient care and enhancing nursing services continues to increase [5, 15, 17, 18]. However, the literature also highlights concerns associated with the integration of AI in healthcare, including those regarding ethical dilemmas, reliability, data privacy, potential job displacement, and accountability/transparency of the data used to develop AI systems [5, 8, 19, 20]. Based on a thorough review of the literature, this tudy identified four key themes (Nursing Care, Organization, Artificial Intelligence Readiness, and Ethics), providing a framework for understanding the integration of AI technologies within the nursing profession. AI applications in nursing care are reported to enhance the quality of patient care by supporting clinical decision-making, minimizing the risk of errors, and streamlining the development of personalized care plans. AI also helps reduce nurses’ workload, facilitates patient monitoring and care coordination, and enables nurses to focus more on direct patient care by taking over indirect tasks. It offers significant benefits in nursing education by enhancing learning experiences and supporting the development of future nursing professionals [5, 6, 8, 15, 17,18,19, 21,22,23,24,25]. However, concerns have been raised that excessive reliance on AI may diminish nurses’ critical thinking abilities, lead to errors in clinical decision-making, undermine the value of nursing roles, and potentially result in the future replacement of nurses [5, 16, 18, 23, 24]. Nevertheless, in organizational contexts, AI is known to improve decision-making processes by supporting proactive data activities, improving operational efficiency, boosting productivity, and enabling the continuous adaptation of practices to evolving conditions. In addition, AI contributes to increased nurse workforce productivity, accelerated innovation, and more cost-effective healthcare services through the optimization of data usage and the streamlining of various organizational functions [15, 16, 18, 21, 23,24,25,26,27]. Regarding nurse readiness for AI technologies, focus is placed on nurses’ knowledge of foundational and health-related applications, willingness to learn, and confidence in preparation for AI practice integration. The quality of AI integration within nursing education and nurses’ competence in using AI applications are critical factors in determining their readiness to effectively engage with these technologies [6, 8, 15, 24, 28, 29]. Among the most important topics discussed in the literature regarding AI is ethics, with concerns highlighting that AI applications may lead to ethical violations, compromise data security within healthcare institutions, and violate patient privacy. In addition, the risks of malicious coding that may potentially harm patients, the challenges of ensuring accountability in AI-driven decisions, and the inadequacy of legal regulations are considered significant threats that could jeopardize the integrity of the healthcare system [16, 18,19,20, 23, 24].

The literature review conducted by the researchers found that Schepman and Rodway’s scale (2020) designed to assess general attitudes toward AI among adults aged 18 and over in the United Kingdom has also been directly adapted for use with adults aged 18 and over in the context of Turkish culture [30], with an adaptation of the scale having also been used for Korean nursing students [31]. Review of the national literature also identified a four-item short-form scale that assesses general attitudes toward AI [32]. Such scales assess adults’ general attitudes toward AI using broad statements that reflect both positive and negative perspectives (e.g., “Artificial Intelligence is exciting”), but they do not specifically measure nurses’ attitudes toward AI from the perspective of the nursing profession. A comprehensive literature review on AI-based technologies in nursing conducted by Von Gerich et al. (2022) recommended that future research should focus on evaluating nurses’ attitudes toward AI [8], and a recent study by Yılmaz et al. (2025) has led to the development of a measurement tool, including both positive and negative dimensions, that assesses nurses’ attitudes toward AI [33]. The novelty of the current study lies in its exploration of four distinct factors (Nursing Care, Organization, Ethics, and Artificial Intelligence Readiness) that specifically relate to AI in nursing, providing a focused framework for understanding the topic. This study aimed to develop and validate a scale that can effectively measure Turkish nurses’ attitudes toward artificial intelligence.

Methods

Study design

This methodological study was conducted to develop and validate an Artificial Intelligence Attitude Scale for Turkish Nurses. Kyriazos and Stalikas’ scale development steps were used to develop the measurement tool and validate its psychometric properties [34], and the study abided by the Strengthening the Reporting of Observational Studies in Epidemiology Statement (STROBE) guidelines [35].

Creation of the item pool and development validity

In the first stage of this study, a comprehensive literature review was conducted by the research team to examine the current applications of AI in nursing. The item pool for the scale was developed based on a thorough review of the relevant literature. As a result of the content analysis, an initial pool of 75 items was identified. The research team reviewed these items during an evaluation meeting, considering factors such as similarity in phrasing, conceptual clarity, relevance, length, and overall clarity. As a result of the evaluation, the item pool was refined and reduced to 40 items to more accurately reflect the current state of AI applications in the field (Fig. 1). Finally, the 40-item draft scale was submitted to experts for their review and feedback.

Fig. 1

Development, validity, and reliability of the artificial intelligence attitude scale for nurses

To assess the content validity of the draft scale used in this study, the researchers consulted the opinions of 12 experts (Supplement Table 1). This panel included one measurement and evaluation specialist and 11 nursing professionals with experience in scale development. To determine content validity, consultation of three to 20 experts in the relevant field is recommended [36]. In this study, experts were asked to evaluate each item of the scale using one of the following Davis method responses: “(1) inappropriate,” “(2) needs serious review,” “(3) needs slight revision,” or “(4) appropriate” [37]. Experts were also asked to offer additional suggestions for each item, and the Content Validity Index (CVI) was 0.56 [37, 38]. Experts’ assessments were then analyzed using Davis’ content validity index, resulting in the removal of eight items from the draft scale [37]. To ensure adequate content validity, the Content Validity Index (CVI) for each remaining item should be ≥ 0.80 [36]. In this study, the CVI for the remaining 32 items ranged from 0.85 to 1.00 for each item, with an overall scale CVI of 0.94. In accordance with expert feedback, final revisions were made to certain items, addressing aspects such as sentence structure, clarity, and relevance to the subject, resulting in the final version of the 32-item draft scale and ensuring content validity. The process related to the psychometric evaluation of the draft version of the Artificial Intelligence Attitude Scale for Nurses is outlined in Fig. 1.

Participants

The literature recommends that the sample size for scale development and validation should be five to ten times the number of scale items [39]. In addition, the sample size should be at least 20 times the number of items, with factor analysis providing reliable results when the sample size exceeds 300 [40]. In this study, a G*Power (3.1.9.7) analysis was conducted to assess the adequacy of the sample size, indicating that a sample of 620 participants would be sufficient, with an effect size of 0.1 at an 80% power level (Critical t: 1.64; df: 619). Inclusion criteria for the sample were outlined at the beginning of the survey. These criteria required that nurses: (1) were actively employed in a hospital setting, (2) had a minimum of two years of professional experience, and (3) voluntarily consented to participate in the study. In testing the validity of a structure identified in the exploratory factor analysis (EFA), confirmatory factor analysis (CFA) using a separate sample is recommended [41]. For this purpose, the data collection process was conducted in two stages. A total of 678 nurses voluntarily participated in the study. In the first stage, EFA was performed with N = 332 participants, while in the second stage, CFA was conducted with N = 346 participants. A minimum sample size of 30 participants is recommended to accurately estimate reliability in test-retest studies [42]. Intraclass Correlation Coefficient (ICC) values between 0.5 and 0.75 indicate moderate reliability, values between 0.75 and 0.90 indicate good reliability, and those greater than 0.90 indicate excellent reliability, assuming the sample size is adequate [43]. A survey was distributed to 30 nurses who were not included in the study sample via two face-to-face interviews, with a 15-day interval between applications [44].

Data collection

The online survey (Google Form) used as the data collection tool consisted of two sections. The first section included a “Socio-Demographic Information Form” comprising 11 questions related to participants’ age, gender, marital status, educational background, institution, unit, total years of professional experience, views on artificial intelligence developments, views on the impact of artificial intelligence on daily and professional life, and experiences with AI applications (e.g., telehealth, navigation, voice assistants, and language translation). The second section of the survey consisted of a draft scale containing 32 items. It is a 5-point Likert-type scale with responses ranging from “1: Strongly Disagree” to “5: Strongly Agree.” Data were collected between October 2024 and December 2024 using a non-probability random sampling method, snowball sampling technique, and an online survey.

Only those who responded “Yes” to the question “Do you voluntarily agree to participate in the study?” were allowed to proceed with the survey. Access to the survey was automatically restricted once responses were submitted. The survey took approximately 5–10 min to complete and was distributed via the WhatsApp and email groups of nurses with whom the researchers had established contact. In the test-retest study, face-to-face data were collected from 30 nurses who were not included in the sample and who had agreed to complete the scale twice, with a 15-day interval between applications. To ensure anonymity while correlating the first and second responses, participants were asked to identify themselves using a number or pseudonym when filling out the forms. The order of the scale items was altered to minimize scoring bias [44].

Data analysis

IBM SPSS Statistics Version 26 (IBM Inc., Armonk, NY, USA) and AMOS 24 (Scientific Software International, Skokie, IL, USA) were used to determine the validity of the scale used in this study. Descriptive statistics for socio-demographic variables and psychometric tests, including Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA), were conducted to assess construct validity. Additionally, convergent and divergent validity statistics were used to further evaluate validity.

Prior to conducting the EFA, the Kaiser-Meyer-Olkin (KMO) coefficient and Bartlett’s test of sphericity were performed to assess sample adequacy and suitability of the correlation between variables for factor analysis.

The maximum likelihood estimation method was employed for the CFA. The following fit indices were used to assess model fit in the Confirmatory Factor Analysis (CFA): χ²/df (chi-square/degree of freedom), GFI (Goodness of Fit Index), AGFI (Adjusted Goodness of Fit Index), NFI (Normalized Fit Index), IFI (Incremental Fit Index), TLI (Tucker-Lewis Index), CFI (Comparative Fit Index), RMR (Root Mean Square Residual), and RMSEA (Root Mean Square Error of Approximation). The model fit criteria are as follows: values of 0.95 ≤ GFI, AGFI, NFI, IFI, TLI, CFI < 1.00, and 0 ≤ RMR, RMSEA ≤ 0.05 indicate good model fit; and values between 0.90 ≤ GFI, AGFI, NFI, IFI, TLI, CFI < 0.95, 0.05 ≤ RMR, RMSEA ≤ 0.08, and 2 ≤ χ²/df ≤ 5 indicate acceptable model fit [40, 45].

Convergent and divergent validity statistics were then applied, including Composite Reliability (CR) and Average Variance Extracted (AVE).

Internal consistency reliability of the scale was assessed by calculating Cronbach’s alpha coefficient, and the test-retest method was employed to evaluate consistency of the scale over time. The first data set was collected on 01/12/2024, and the second data set was collected on 15/12/2024. The intraclass correlation coefficient (ICC) was calculated to assess the consistency and reliability of the test-retest measurements.

Ethical considerations

Prior to conducting this study, written approval was obtained from the research ethics committee of a university in Turkey (approval date: 10.10.2024, approval number: 69553). All participating nurses electronically submitted informed consent.

Results

Socio-demographic characteristics of nurses

Since there were no missing data from participating nurses, the socio-demographic data of all participants were analyzed. Of the 678 participants, 90.1% were female, 63.7% were married, and 43.1% were between the ages of 31 and 40. Over half of the participants (56%) had received an undergraduate education, 55.5% worked in healthcare hospitals, and 69.6% were employed in inpatient treatment units. In terms of professional experience, 46.2% had between one and 10 years of experience in the nursing profession. Regarding their views on AI, 80.4% held positive opinions about AI, 93.4% believed that AI is transforming daily and professional life, and 76.1% reported having experience with AI applications (such as telehealth, navigation, voice assistants, and language translation) (Supplement Table 2).

Construct validity

Exploratory Factor Analysis (EFA)

Before conducting EFA, suitability of the data for factor analysis was evaluated using the Kaiser-Meyer-Olkin (KMO) coefficient and Bartlett’s test of sphericity [46]. For factor analysis, a KMO coefficient of ≥ 0.70 is generally recommended [39]. In this study, the KMO value was found to be 0.88, and Bartlett’s test of sphericity was statistically significant (χ² = 11.190; df = 496; p < 0.001), indicating that the data were suitable for factor analysis. Varimax rotation method was used. In determining the factor structure, the criteria of eigenvalue coefficients greater than 1.0 and factor loadings greater than 0.50 were applied [40, 46, 47]. Since the item factor loadings of the scale were above 0.50, no items were removed from the scale (Fig. 1). A total of 32 items and four factors with eigenvalues greater than one were obtained (Supplement Table 3). The total variance explained by the four factors was 72.185%. The CVI for the overall scale items was 0.94, and the total correlation scores for the scale items were above 0.30 (Table 1).

Table 1 Item analysis and exploratory factor analysis (N = 332)

Confirmatory Factor Analysis (CFA)

Validity of the four-factor structure derived from EFA was tested using CFA. Model fit indices were used to interpret CFA results. Model values of 0.95 ≤ GFI, AGFI, NFI, IFI, TLI, CFI < 1.00; 0 ≤ RMR, RMSEA ≤ 0.05 and 0 ≤ χ²/df ≤ 2 indicated a good fit, while values of 0.90 ≤ GFI, AGFI, NFI, IFI, TLI, CFI < 0.95, 0.05 ≤ RMR, RMSEA ≤ 0.08, and 2 ≤ χ²/df ≤ 5 indicated acceptable fit [40, 45]. In this study, the model fit values were found to be at good and acceptable levels (χ²/df = 2.154; RMR = 0.071; GFI = 0.936; AGFI = 0.957; NFI = 0.950; IFI = 0.965; TLI = 0.962; CFI = 0.958; RMSEA = 0.064). Accordingly, AGFI, NFI, IFI, TLI, and CFI indicated good fit, while χ²/df, RMR, GFI, and RMSEA demonstrated acceptable fit. According to the results and the content of the items, the subscales were as follows: “Factor 1: Nursing Care,” “Factor 2: Organization,” “Factor 3: Ethics,” and “Factor 4: Artificial Intelligence Readiness” (Fig. 2). When calculating the mean score of the scale in the nursing care subscale, items 13 and 14 were reverse-coded. Similarly, for the ethics subscale, items 21, 22, 23, 24, and 25 were also reverse-coded. It was determined that all items of the Artificial Intelligence Attitude Scale for Nurses had a statistically significant effect (p < 0.01). When evaluating the correlations between the items and the scale, the R² values were found to range between 0.360 and 0.932. The standardized Beta coefficients were found to range between 0.600 and 0.965 (Supplement Table 4).

Fig. 2

CFA Visual model of artificial intelligence scale for nurses (N = 346). NC: Nursing Care; OZ: Organization; EH: Ethics; AIR: Artificial Intelligence Readiness; AI: Artificial Intelligence

Convergent and divergent validity

Convergent validity refers to the extent to which multiple indicators that measure the same construct are highly correlated with each other, demonstrating that they measure the same underlying concept [46]. Divergent validity, on the other hand, assesses the degree to which a measurement does not correlate with other measurements that are assumed to diverge [48]. In general, an average variance extracted (AVE) value above 0.5 for each factor and a composite reliability (CR) value above 0.7, with the square root of AVE being higher than the inter-factor correlations, indicates strong convergent and discriminant validity [49, 50] (Table 2).

Table 2 Convergent and divergent validity statistics (N = 346)

Reliability

Internal consistency reliability—cronbach’s alpha coefficient

Cronbach’s alpha coefficient was used to calculate reliability, with values between 0 and 0.49 indicating unacceptable reliability, values between 0.50 and 0.59 indicating poor reliability, values between 0.60 and 0.69 indicating questionable reliability, values between 0.70 and 0.79 indicating acceptable reliability, values between 0.80 and 0.89 indicating good reliability, and values between 0.90 and 1.00 indicating excellent reliability [48]. In this study, Cronbach’s alpha coefficient for the 32-item, four-factor structure was found to be 0.925, with values for the subscales ranging between 0.913 and 0.960. These results indicated that Cronbach’s alpha coefficient for the Artificial Intelligence Attitude Scale for Nurses was at an excellent level.

Consistency of scale over time—pearson correlation coefficients

Consistency of the scale used in this study was evaluated using the test-retest method, and the ICCS were calculated. The average ICC for the Artificial Intelligence Attitude Scale for Nurses was 0.947, with a 95% confidence interval (CI) ranging from 0.887 to 0.975 (F = 12.382, p < 0.001). According to the scale, the ICC measurements were as follows: 0.936 for Nursing Care (95% CI = 0.867–0.970; F = 15.739, p < 0.001), 0.905 for Organization (95% CI = 0.800–0.955; F = 10.493, p < 0.001), 0.923 for Ethics (95% CI = 0.838–0.963; F = 12.998, p < 0.001), and 0.938 for Artificial Intelligence Readiness (95% CI = 0.887–0.975; F = 16.147, p < 0.001). A significant and positive correlation was found between the two measurements (p < 0.01) (Table 3). For test-retest reliability, ICC values above 0.80 and 0.90 are considered good and excellent, respectively [43]. In addition, the confidence intervals for the ICC values, ranging from 0.44 to 0.84, indicate that the sample size selected for test-retest reliability is adequate [51].

Table 3 Test-retest results of the 95% CI (N = 30)

Discussion

Use of AI is increasing globally in the healthcare sector, as in many other industries. Forecasts for global AI expenditures in 2025 also support this trend [5, 8]. Understanding the current attitudes and knowledge of nurses—who constitute the majority of the healthcare workforce and work most closely with patients—is crucial in the context of AI technologies [10,11,12]. As AI algorithms are increasingly integrated into clinical settings and nursing practices, their role in supporting patient diagnosis, treatment, and nursing care organization is also expanding [5, 8, 52]. With the use of AI, nurses can enhance their understanding of patients’ conditions and needs, thereby improving quality of care through the incorporation of AI-generated data [5, 8]. AI technologies support various aspects of patient care, including diagnosis (e.g., identification of heart rhythm abnormalities, deviations from normal vital signs, evaluation of systemic diseases and symptoms, and detection of anomalies in imaging results), treatment (e.g., medication administration, treatment planning, and patient monitoring), and nursing care organization (e.g., documentation, automated reporting, triage prioritization, and monitoring patient outcomes such as the risk of deterioration, regular management of chronic conditions, and robot-assisted home care) [5, 8, 52,53,54,55,56]. As new algorithms (the foundational components of AI) are integrated into patient care processes, it may become possible to increase the time nurses spend on direct patient care, reduce the burden of tasks related to indirect patient care, support clinical decision-making, enhance both the quality and efficiency of patient care, and ensure the effective and efficient delivery of nursing services [5, 11, 15, 56]. However, ethical concerns associated with the use of AI in nursing, such as data security, algorithmic bias, and the lack of transparency in AI models, have been widely reported [5, 8]. Therefore, understanding nurses’ attitudes toward AI ethics can contribute to the planning and development of strategies in this area.

A review of the literature on AI technologies in nursing highlights the need for studies that evaluate nurses’ attitudes towards AI [8]. A recent study in Turkey developed a scale to assess both the positive and negative dimensions of nurses’ attitudes toward AI [33]. Content analysis of the information obtained from the literature review on the intersection of nursing and AI identified four main themes: Nursing Care [5, 6, 8, 15, 17,18,19, 21,22,23,24,25], Organization [15, 16, 18, 21, 23,24,25,26,27], Artificial Intelligence Readiness [6, 8, 15, 24, 28, 29], and Ethics [16, 18,19,20, 23, 24]. The factors of the scale were defined in accordance with these themes. The study aimed to develop the Artificial Intelligence Attitude Scale for Nurses and test its validity and psychometric properties, including both validity and reliability. Scale studies should be tested for both “validity” (which indicates the degree of accurate measurement) and “reliability” (which reflects the consistency of measurement values) [36]. The validity and reliability results of the scale are presented and discussed below.

According to the Davis technique, the CVI value should be ≥ 0.56 [57]. In this study, the CVI values for the scale items ranged from 0.85 to 1.00, and the overall CVI value for the scale was 0.94. It is recommended that the variance explained in multi-factor structures should be at least 50% [46]. In this study, the total explained variance for the four-factor structure was 72.185%. Therefore, the scale was found to be comprehensive in terms of explaining nurses’ attitudes toward artificial intelligence.

The KMO value for this study was 0.94. The sample size, as indicated by the KMO value, was adequate for conducting EFA [39]. Factor loadings indicate the correlation between items and factors, with a factor loading below 0.30 considered unacceptable. When interpreting factor load coefficients, a value above 0.40 is considered a good measure [39, 46, 58]. In this study, the factor loadings of the scale items were above 0.50, meeting the desired criteria. Validity of the post-EFA scale was further tested using CFA. According to the analysis results, the AGFI, NFI, IFI, TLI, and CFI models showed good fit with the scale model, with the χ2/sd, RMR, GFI, and RMSEA values indicating acceptable fit [40, 45, 58].

After CFA, convergent and divergent validity (CR > 0.70, CR > AVE and AVE > 0.50) of the developed scale model was confirmed [40, 58]. An AVE below 0.50 and a CR value between 0.60 and 0.70 indicate that convergent validity is still achieved [59].

Cronbach’s alpha value of the scale used in this study was 0.925, with values ranging between 0.913 and 0.960 for the subscales. The data showed that Cronbach’s alpha value of the scale was high [45]. MaxR(H) values were above 0.70, and MaxR(H) was greater than the CR values [59]. A positive correlation was found between the test-retest scores (the lowest value being 0.905), indicating invariance and consistency of the scale over time [43, 44]. No significant difference was found between the mean test-retest scores and the subscales of the scale, indicating a strong, positive, and significant correlation.

Limitations

Although this study makes important contributions to the development and validation of the Artificial Intelligence Attitude Scale for Nurses, some limitations should be considered regarding the generalizability of the results. First, the study was conducted within the unique cultural context of Turkey, which may have influenced participants’ attitudes toward AI. Accordingly, cultural context may have influenced the results. Further research designed to test the psychometric properties of the Artificial Intelligence Attitude Scale for Nurses in different countries is recommended. Additionally, it should be noted that the scale was originally developed and validated in Turkish and subsequently translated into English for scientific reporting. This limitation highlights the need for special adaptations and comprehensive scale testing when used in languages other than Turkish, including English. We recommend that the scale be adapted by conducting language studies with nurses from different cultures. In this way, the scale may yield effective results in other cultures. To further explore the relationship between AI and nursing, future research should involve larger and more diverse samples to assess dimensions such as nursing care, organization, ethics, and AI readiness among nurses. Finally, as no similar scale has been previously used with nurses, the results of this study could not be compared with others. Therefore, the study may have limitations in terms of sensitivity and originality.

Conclusions

The Artificial Intelligence Attitude Scale for Nurses is a valid and reliable measurement tool used to effectively assess nurses’ attitudes toward AI. It can also be used to examine nurses’ attitudes towards AI based on various variables. As the use of AI continues to expand in clinical and nursing practice, future studies on this topic may offer valuable insights into managing change processes within healthcare settings. In addition, the scale can contribute to the development of AI-related training programs, helping to identify and address the educational needs of nurses regarding AI technologies. By adapting the scale to different cultures, nurses’ attitudes towards AI related to “nursing care,” “organization,” “ethics,” and “readiness for artificial intelligence among nurses” may be identified and compared. The data obtained from use of this scale may especially contribute to the development of future strategies used by policy makers, managers of health institutions, and nurse managers.