AI vs. Physicians in Surgery, Radiology, and Dermatology: Exploring Public Trust and Preferences in the Northern Region of Saudi Arabia

Introduction

Artificial intelligence (AI) has become an essential part of many fields, revolutionizing workplaces and the nature of task performance. The last decade has seen increasing adoption of AI in medical specialties such as surgery, radiology, and dermatology, which are at the forefront of AI integration due to their reliance on image analysis and procedural precision [1-4].

In surgery, AI applications include enhanced surgical robotics and preoperative risk prediction [2]. In radiology, AI facilitates automated image analysis, improving diagnostic accuracy and workflow speed [3]. In dermatology, AI enhances the diagnosis of malignancies and supports tasks like documentation and histopathology classification [4]. Despite these advancements, physicians remain central to clinical decision-making, combining scientific knowledge with empathy, communication, and ethical reasoning—skills AI cannot yet replicate [5, 6]. Consequently, AI is increasingly viewed as a complementary tool that enhances clinical performance while the physician maintains primary authority [7].

The incorporation of AI in healthcare raises important concerns regarding its impact on patient-centered care, data privacy, and legal accountability [8, 9]. Public trust is therefore a critical factor for the successful integration of AI. The Northern region of Saudi Arabia is characterized by strong cultural values that emphasize personal interaction, making it a pertinent setting to explore public perceptions. This study aims to assess the level of public trust and acceptance toward the use of AI in surgery, radiology, and dermatology in the Northern Region of Saudi Arabia.

Methodology:

Study Design:

This study employed a descriptive cross-sectional design to evaluate public trust, perceived safety, and acceptance of AI in radiology, surgery, and dermatology among the general population in the Northern region of Saudi Arabia.

Study Area and Setting:

The study was conducted in the Northern region of Saudi Arabia, including both urban and rural communities to ensure diverse representation. Data were collected from November 2023 to June 2024 via online surveys and field-based approaches in high-traffic public venues such as universities, malls, and primary healthcare centers.

Study Population and Sampling:

The study population consisted of Saudi adults aged ≥18 years residing in the Northern region. The minimum sample size was calculated as 385 using Cochran's formula (95% CI, 5% margin of error, 50% prevalence). A total of 493 eligible and complete responses were obtained using a convenience sampling technique to enhance regional representation and feasibility.

Inclusion and Exclusion Criteria:

Eligible participants were Saudi citizens aged ≥18 years who could read and understand Arabic and provided informed consent. Respondents who declined participation, did not complete the questionnaire, or had more than 50% missing data in key outcome measures were excluded.

Data Collection Tool and Procedure:

A validated questionnaire by Yakar et al. was adapted for this study [11]. The original questionnaire created by Yakar et al. was translated into Arabic using a standardized forward-backward translation process. Two independent bilingual translators carried out the forward translation, which was then reconciled into a single Arabic version. Two more translators performed a back-translation into English to ensure it was conceptually equivalent to the original instrument. An expert panel reviewed the Arabic version for clarity and cultural appropriateness. A pilot test (n=30) confirmed clarity and reliability, with a Cronbach's alpha of 0.84 indicating good internal consistency. The final questionnaire included three sections: sociodemographic characteristics, healthcare utilization, and attitudes toward AI using a 5-point Likert scale. Overall acceptance was operationally defined as the combination of "High trust" and "Neutral trust" responses, as opposed to "Non-acceptance" (Low trust) [11].

Ethical Consideration:

Ethical approval was obtained from the Research Ethics Committee at Jouf University under number (7546). Informed consent was obtained electronically from all participants, and all data were anonymized and stored securely in compliance with the Declaration of Helsinki.

Statistical Analysis:

Data were analyzed using IBM SPSS Statistics version 23. Descriptive statistics were reported as frequencies and percentages for categorical variables. Chi-square tests examined associations between demographics and AI acceptance. A repeated-measures ANOVA was used to compare trust scores across the three specialties. Statistical significance was set at p < 0.05.

RESULTS

The demographic characteristics of the participants show that the majority were aged 31–50 years 253(51.3%), followed by those aged 18–23 years 97(19.7%), 24–30 years 90(18.3%), and older than 50 years 53(10.3%). In terms of gender, females 268(54.4%) slightly outnumbered males 225(45.6%). Most participants had a college education 339(68.8%), while 95(19.3%) had postgraduate education, 57(11.6%) had secondary education, and only 2(0.4%) had primary education. Regarding employment status, the majority were employed 333(67.5%), followed by students 86(17.4%), unemployed 50(10.1%), and retired 24(4.9%).

Table 1: Demographic information of the participants (N=493)
Demographic information	Category	N (%)
Age	18-23	97 (19.7%)
24-30	90 (18.3%)
31-50	253 (51.3%)
Older than 50	53 (10.3%)
Gender	Female	268 (54.4%)
Male	225 (45.6%)
Level of education	College education	339 (68.8%)
Postgraduate	95 (19.3%)
Primary education	2 (0.4%)
Secondary education	57 (11.6%)
Employment status	Employed	333 (67.5%)
Retired	24 (4.9%)
Student	86 (17.4%)
Unemployed	50 (10.1%)

Demographic information presented in frequencies (n) and proportion (%)

The healthcare utilization (table 2) indicated that in the past 1 month majority of the participants had consulted a general practitioner less than 5 times (343, 69.6%). In the past 12 months, majority of them had consulted a specialist less than 5 times (374, 75.9%), had been admitted in hospital less than 5 times (361, 73.2%) and majority had spent less than 5 days in hospital (423, 85.8%)

Table 2: Healthcare utilization
Statement	Category	N (%)
Number of times consulted a general practitioner in the past 1 months	0-4	343 (69.6%)
5-8	103 (20.9%)
9-12	27 (5.5%)
13-16	8 (1.6%)
More than 16	12 (2.4%)
Number of times consulted a specialist in the past 12 months	0-4	374 (75.9%)
5-8	86 (17.4%)
9-12	19 (3.9%)
13-16	4 (0.8%)
More than 16	10 (2.0%)
Number of hospital admissions in the past 12 months	0-4	361 (73.2%)
5-8	86 (17.4%)
9-12	28 (5.7%)
13-16	9 (1.8%)
More than 16	9 (1.8%)
Number of days spent in a hospital in the past 12 months	0-4	423 (85.8%)
5-8	38 (7.7%)
9-12	17 (3.4%)
13-16	9 (1.8%)
More than 16	6 (1.2%)

Healthcare utilization presented in frequencies (n) and proportion (%)

Trust in AI in radiology (table 3) revealed that nearly one-third 184 (37.3%) of the participants expressed trust in the Artificial intelligence reliability and efficiency in communicating patients’ results. However, majority of the participants expressed skepticism regarding the AI’s capacity to take over diagnostic interpretation of tasks (471, 95.5%) and clarity about medical procedures (473, 95.9%)

Table 3: Trust in AI in Radiology
Statement	Category	N (%)
Even if computers are better at evaluating scans, I still prefer a doctor	Agree	183 (37.1%)
Disagree	37 (7.5%)
Neutral	146 (29.6%)
Strongly Agree	105 (21.3%)
Strongly Disagree	22 (4.5%)
I think radiology is not ready for implementing AI in evaluating scans	Agree	123 (24.9%)
Disagree	70 (14.2%)
Neutral	204 (41.4%)
Strongly Agree	62 (12.6%)
Strongly Disagree	34 (6.9%)
It worries me when computers analyze scans without human interference	Agree	145 (29.4%)
Disagree	83 (16.8%)
Neutral	159 (32.3%)
Strongly Agree	73 (14.8%)
Strongly Disagree	33 (6.7%)
The sooner I get the results from a computer, the more at ease I am	Agree	184 (37.3%)
Disagree	52 (10.5%)
Neutral	162 (32.9%)
Strongly Agree	69 (14.0%)
Strongly Disagree	26 (5.3%)
I wonder how it is possible that a computer can give me the results of a scan	Agree	175 (35.5%)
Disagree	56 (11.4%)
Neutral	179 (36.3%)
Strongly Agree	63 (12.8%)
Strongly Disagree	20 (4.1%)

Trust in AI in radiology presented in frequencies (n) and proportion (%)

Trust in AI in surgery (table 4) revealed that majority of the participants expressed skepticism regarding the AI’s capacity to take over operative procedures of the surgeon (471, 95.5%), concerned about the clarity about medical procedures (478, 97.0%) as well as the accuracy of the procedure (474, 96.1%)

Table 4: Trust in AI in Surgery
Statement	Category	N (%)
Even if computers are better at operating patients, I still prefer a doctor	Agree	167 (33.9%)
Disagree	42 (8.5%)
Neutral	120 (24.3%)
Strongly Agree	142 (28.8%)
Strongly Disagree	22 (4.5%)
I think hospitals are not ready for implementing AI in surgeries	Agree	169 (34.3%)
Disagree	52 (10.5%)
Neutral	153 (31.0%)
Strongly Agree	95 (19.3%)
Strongly Disagree	24 (4.9%)
It worries me when computers operate patients without human interference	Agree	170 (34.5%)
Disagree	49 (9.9%)
Neutral	150 (30.4%)
Strongly Agree	109 (22.1%)
Strongly Disagree	15 (3.0%)
Through human experience, a surgeon can detect more than a computer	Agree	174 (35.3%)
Disagree	31 (6.3%)
Neutral	148 (30.0%)
Strongly Agree	121 (24.5%)
Strongly Disagree	19 (3.9%)
It is unclear to me how computers will be used in conducting operations	Agree	167 (33.9%)
Disagree	50 (10.1%)
Neutral	173 (35.1%)
Strongly Agree	82 (16.6%)
Strongly Disagree	21 (4.3%)

Trust in AI in surgery presented in frequencies (n) and proportion (%)

Trust in AI in dermatology (table 5) revealed that nearly one-third 193 (39.1%) of the participants expressed trust in the communication of patients’ results. However, majority of the participants expressed skepticism regarding the AI’s capacity to take over diagnostic interpretation tasks of the dermatologist (472, 95.7%) and clarity about medical procedures (473, 95.9%)

Table 5: Trust in AI in Dermatology
Statement	Category	N (%)
Even if computers are better at evaluating skin spot, I still prefer a doctor	Agree	193 (39.1%)
Disagree	50 (10.1%)
Neutral	134 (27.2%)
Strongly Agree	93 (18.7%)
Strongly Disagree	23 (4.7%)
I think dermatology is not ready for implementing AI in evaluating skin spots	Agree	156 (31.6%)
Disagree	65 (13.2%)
Neutral	176 (35.7%)
Strongly Agree	75 (15.2%)
Strongly Disagree	21 (4.3%)
It worries me when computers analyze skin spots without human interference	Agree	159 (32.3%)
Disagree	68 (13.8%)
Neutral	172 (34.9%)
Strongly Agree	74 (15.0%)
Strongly Disagree	20 (4.1%)
The sooner I get the results from a computer, the more at ease I am	Agree	193 (39.1%)
Disagree	45 (9.1%)
Neutral	163 (33.1%)
Strongly Agree	71 (14.4%)
Strongly Disagree	21 (4.3%)
I wonder how it is possible that a computer can evaluate my skin	Agree	182 (36.9%)
Disagree	50 (10.1%)
Neutral	173 (35.1%)
Strongly Agree	68 (13.8%)
Strongly Disagree	20 (4.1%)

Trust in AI in dermatology presented in frequencies (n) and proportion (%)

Figure 1 illustrates participants’ trust and acceptance of Artificial intelligence. About 25 participants (5.1%) had high trust, more than half 279 (56.6%) had neutral trust while about one-third 189 (38.3%) had low trust in AI in radiology, surgery and dermatology.

Figure 1: Public trust and acceptance of Artificial Intelligence

The participants’ trust and preferences of AI (figure 2) revealed that participants had slightly more trust in dermatology (8.7%) than in radiation (6.9%) and surgery (5.7%)

Figure 2: Participants trust and preferences of AI

The acceptance to AI in radiology, surgery and dermatology among the study participants (Table 6) shows that, of the 493 participants, 304 (61.7%) accepted, while about 189 participants (38.3%), were skeptical about AI in radiology, surgery and dermatology.

Table 6: Overall acceptance of Artificial Intelligence among the participants
Variables	Category	Frequency (n)	Percentage (%)
Acceptance of AI in radiology, surgery and dermatology	High trust	25	5.1%
Neutral trust	279	56.6%
Overall acceptance	304	61.7%
Non-acceptance	189	38.3%

Acceptance of AI in radiology, surgery and dermatology

The results (Table 7) revealed statistically significant associations between participants' age, employment status, and their acceptance of AI in radiology, surgery and dermatology (p = 0.005, p = 0.038). A considerable proportion of the participants aged less than 30 years and unemployed demonstrated acceptance of Artificial intelligence. However, no significant associations were observed between other variables and acceptance of AI (p > 0.05).

Table 7: The association between demographic information with Acceptance of AI in radiology surgery and dermatology
Variables	Acceptance of AI
Category	No	Yes	p-value
Age	18-23	31 (32.0%)	66 (68.0%)	0.005*
24-30	28 (31.1%)	62 (68.9%)
31-50	99 (39.1%)	154 (60.9%)
Older than 50	31 (58.5%)	22 (41.5%)
Gender	Female	105 (39.2%)	163 (60.8%)	0.675
Male	84 (37.3%)	141 (62.7%)
Level of education	College education	130 (38.3%)	209 (61.7%)	0.351
Postgraduate	41 (43.2%)	54 (56.8%)
Primary education	0 (0.0%)	2 (100.0%)
Secondary education	18 (31.6%)	39 (68.4%)
Employment status	Employed	142 (42.6%)	191 (57.4%)	0.038*
Retired	7 (29.2%)	17 (70.8%)
Student	27 (31.4%)	59 (68.6%)
Unemployed	13 (26.0%)	37 (74.0%)

* Significant at p<0.05 level.

In the multivariate logistic analysis results (Table 8), the number of times consulted a general practitioner in the past 1 month (AOR=1.13; 95% CI= 0.820–1.545; p=0.464) was identified as a positive predictor for acceptance of artificial intelligence. However, these factors were not statistically significant.

Table 8: Factors associated with acceptance of Artificial intelligence among the participants
Variables	AOR	95% CI	P-value
Age	0.81	0.636 – 1.030	0.085
Education	0.87	0.622 – 1.225	0.432
Employment	0.87	0.705 – 1.071	0.187
No. of times consulted a general practitioner past 1 months	1.13	0.820 – 1.545	0.464
No. of times consulted a specialist in the past 12 months	1.02	0.730 – 1.413	0.928
No. of hospital admissions in the past 12 months	0.92	0.677 – 1.257	0.610
No. of days spent in a hospital in the past 12 months	1.00	0.725 – 1.376	0.992

AOR – Adjusted Odds Ratio; CI – Confidence Interval

** Significant at p<0.05 level

Discussion

This study assessed public trust and acceptance of AI in three key medical specialties in the Northern Region of Saudi Arabia. The findings reveal a critical distinction: while overall acceptance was moderate (61.7%), active trust was remarkably low (5.1% high trust) while 38.3% has low trust. This suggests a pragmatic or passive acceptance, where the public acknowledges AI's role without fully trusting it, highlighting a significant barrier to its deep integration.

In the healthcare provision in Saudi Arabia, several different cultural, health system, and regional factors have an important impact on public trust in artificial intelligence. The reported low trust (38.3%) in AI among participants indicates significant skepticism even with moderate overall acceptance in the study context. Culturally, healthcare encounters here emphasize close connections between physician and patient, personal exchange of information and clinicians offering comfort and reassurance in their presence, which could reduce trust in technology and decision-making, particularly where there’s reduced human contact (like in high-stakes surgical procedures and radiology) [12]. From a health system perspective, while Saudi Arabia is making significant investment in digital health and artificial intelligence through Vision 2030, public exposure to fully autonomous AI systems in routine health care is lacking [13]. This could encourage uncertainty about accountability, error liability and clinical oversight, and lead to the greater trust to see AI tools as an adjunct to practitioners' roles rather than as a substitute for doctors.

The moderate to high level of trust in dermatology aligns with international studies [14-16], likely because AI's image-based diagnostic role in skin conditions is more tangible and perceived as less invasive compared to its applications in surgery or complex radiology interpretations. The profound skepticism regarding AI's capacity to take over operative procedures or function without human oversight reflects concerns about accountability, transparency, and the loss of human judgment in high-stakes scenarios [6]. This is consistent with Saudi studies that report hesitancy driven by worries about data quality and job replacement [17, 18].

The association between younger age, unemployment, and higher AI acceptance can be attributed to greater digital literacy and familiarity with technology among the youth. This finding may be partly explained by demographic overlap, as younger participants—who tend to be more digitally literate and receptive to technological innovation—were more represented within the unemployed group. Younger individuals may perceive AI as efficient, accessible, and aligned with modern healthcare delivery models. Additionally, employed participants, particularly those working in healthcare or related sectors, may express greater caution due to concerns about professional displacement or accountability issues.

Policy and Educational Implications:

To bridge the trust gap, targeted initiatives are essential. Public awareness campaigns should transparently demonstrate AI's role as an supportive tool not a replacement for physicians. Medical curricula and continuous professional development programs should include digital health literacy, focusing on how AI supports clinical decision-making. Ethically, ensuring transparency, robust data governance, and clear accountability frameworks will be crucial for building public confidence.

Limitations Most study participants had a college or postgraduate degree, which may limit the generalizability of the findings to the general population in the Northern region of Saudi Arabia. Due to their relatively higher level of education, this sample tends to be more familiar with digital technologies and health innovations, thus potentially shaping their views on the acceptance aspect of artificial intelligence. Additionally, the convenience sampling of the sample as well as the partial dependence on online survey distribution may have introduced selection bias, with internet users and highly digitally literate participants tending to be more likely to do the survey. The findings, therefore, ought to be read within the perspective of this relatively well-educated sample.

Conclusion

This study found that while there is moderate acceptance of AI in healthcare among the well-educated public in Northern Saudi Arabia, deep-seated trust remains low. Participants were most skeptical about AI's role in surgery and most accepting in dermatology. To support the successful and ethical integration of artificial intelligence into healthcare systems, targeted educational strategies are essential. Medical education curriculum experts should embrace AI-specific competencies in medical training programs to build future physicians' clinical competence regarding the applications, ethical issues, and limits of AI.

References

1. Rahman MA, Victoros E, Ernest JT, Davis R, Shanjana Y, Islam MR. Impact of artificial intelligence (AI) technology in healthcare sector: a critical evaluation of both sides of the coin. Clin Pathol. 2024;17. https://doi.org/10.1177/2632010X241226887

1. Ng J. The rise of robotics and AI-assisted surgery in modern healthcare. J Robot Surg. 2025;19(1). https://doi.org/10.1007/s11701-025-02485-0

1. Najjar R. Redefining radiology: a review of artificial intelligence integration in medical imaging. Diagnostics (Basel). 2023;13(17):2760. https://doi.org/10.3390/diagnostics13172760

1. Biswas S, Achar U, Hakim B, Achar A. Artificial intelligence in dermatology: a systematized review. Int J Dermatol Venereol. 2024. https://doi.org/10.1097/JD9.0000000000000404

1. Alowais SA, Alghamdi SS, Alsuhebany N, Alqahtani T, Alshaya AI, Almohareb SN, et al. Revolutionizing healthcare: the role of artificial intelligence in clinical practice. BMC Med Educ. 2023;23(1):689. https://doi.org/10.1186/s12909-023-04698-z

1. Kahraman F, Aktas A, Bayrakceken S, Çakar T, Tarcan HS, Bayram B, et al. Physicians' ethical concerns about artificial intelligence in medicine: a qualitative study: "The final decision should rest with a human". Front Public Health. 2024;12:1428396. https://doi.org/10.3389/fpubh.2024.1428396

1. Martínez-García E, Affleck A, Rakvit P, Arias-Santiago S, Buendía-Eisman A. Exploring patients' insight, concerns, and expectations at dermatology clinic: an observational study in 2 centers in Scotland and Spain. J Patient Exp. 2020;7(6):1197-202. https://doi.org/10.1177/2374373520912085

1. Witkowski K, Okhai R, Neely SR. Public perceptions of artificial intelligence in healthcare: ethical concerns and opportunities for patient-centered care. BMC Med Ethics. 2024;25(1). https://doi.org/10.1186/s12910-024-01066-4

1. Farhud D, Zokaei S. Ethical issues of artificial intelligence in medicine and healthcare. Iran J Public Health. 2021;50(11):i-v. https://doi.org/10.18502/ijph.v50i11.7600

1. Akinrinmade AO, Adebile TM, Ezuma-Ebong C, Bolaji K, Ajufo A, Adigun AO, et al. Artificial intelligence in healthcare: perception and reality. Cureus. 2023;15(9). https://doi.org/10.7759/cureus.45594

1. Yakar D, Ongena YP, Kwee TC, Haan M. Do people favor artificial intelligence over physicians? A survey among the general population and their view on artificial intelligence in medicine. Value Health. 2022;25(3):374-81. https://doi.org/10.1016/j.jval.2021.09.004

1. Alhakami H, Alsubait T, Alhakami W, Alhakami H, Alhakami R, Alhakami M, Khan RA, Ansari MTJ. Advancing Sustainable Healthcare through Enhanced Therapeutic Communication with Elderly Patients in the Kingdom of Saudi Arabia. Sustainability. 2023; 15(22):15778. https://doi.org/10.3390/su152215778

1. Maswadi K, Alhazmi A. Towards Sustainable Health Management in the Kingdom of Saudi Arabia: The Role of Artificial Intelligence—A Systematic Review, Challenges, and Future Directions. Sustainability. 2026; 18(2):905. https://doi.org/10.3390/su18020905

1. Scheetz J, Rothschild P, McGuinness M, Hadoux X, Soyer HP, Janda M, et al. A survey of clinicians on the use of artificial intelligence in ophthalmology, dermatology, radiology and radiation oncology. Sci Rep. 2021;11(1):5193. https://doi.org/10.1038/s41598-021-84698-5

1. Heinlein L, Maron RC, Hekler A, Haggenmüller S, Wies C, Utikal JS, et al. Prospective multicenter study using artificial intelligence to improve dermoscopic melanoma diagnosis in patient care. Commun Med (Lond). 2024;4(1):177. https://doi.org/10.1038/s43856-024-00598-5

1. Wu A, Ngo M, Thomas C. Assessment of patient perceptions of artificial intelligence use in dermatology: a cross-sectional survey. Skin Res Technol. 2024;30(3):e13656. https://doi.org/10.1111/srt.13656

1. Alkhatieb M, Subke AA, Alkhatieb MT. Artificial intelligence in healthcare: a study of physician attitudes and perceptions in Jeddah, Saudi Arabia. Cureus. 2024;16(3). https://doi.org/10.7759/cureus.57256

1. Alsaedi AR, Alneami N, Almajnoni F, Alamri O, Aljohni K, Alrwaily MK, et al. Perceived worries in the adoption of artificial intelligence among healthcare professionals in Saudi Arabia: a cross-sectional survey study. Nurs Rep. 2024;14(4):3706-21. https://doi.org/10.3390/nursrep14040271