Artificial intelligence in dentistry: between clinical efficiency, human judgment, professional responsibility, and humanized care

Full text

Leonardo Lins2

Antonio Canabarro3

ABSTRACT The incorporation of artificial intelligence represents one of the most significant transformations in contemporary dentistry, expanding the capacity for pattern recognition, image analysis, clinical data organization, and support for treatment planning. However, its impact should not be assessed solely through algorithmic performance indicators, but also through its implications for clinical judgment, professional responsibility, and humanized care. This study aimed to critically discuss the incorporation of artificial intelligence in dentistry based on four interdependent axes: clinical efficiency, human judgment, professional responsibility, and humanized care. This is a critical and reflective literature review, guided by transparency principles inspired by the SANRA tool. Scientific publications, institutional documents, regulatory frameworks, and theoretical references related to digital dentistry, clinical decision-making, automation bias, explainable artificial intelligence, bioethics, and patient-centered care were analyzed. The analysis indicates that artificial intelligence can enhance clinical efficiency and improve stages of dental diagnosis and treatment planning, particularly in tasks based on pattern recognition. However, its uncritical use may favor automation bias, a false sense of security, displacement of decision-making responsibility, and weakening of the professional-patient relationship. It is concluded that artificial intelligence should be understood as an instrument of augmented intelligence, designed to support, rather than replace, clinical judgment. Keywords: Artificial Intelligence. Dentistry. Clinical Judgment. Professional Responsibility. Humanized Care.

RESUMO

1 The authors declare that they used Generative Artificial Intelligence (GenAI) tools to support the conception, writing, and translation of this manuscript. The content was critically reviewed by the authors, who assume full responsibility for the final version of the manuscript. Translations of excerpts from works written in another language, when rendered into English, are likewise the responsibility of the authors, unless a different authorship is declared.

2 Doutor e Mestre em Odontologia – Reabilitação Oral. Universidade Veiga de Almeida. Email: [email protected]. ORCID: https://orcid.org/0000-0002-6768-9071.

3 Professor Titular. Coordenador e Professor do Programa de Pós-Graduação Stricto Sensu em Odontologia. Universidade Veiga de Almeida. Email: [email protected]. ORCID: https://orcid.org/0000-0002-3970-9047.

A incorporação da inteligência artificial representa uma das transformações mais relevantes da odontologia contemporânea, ampliando a capacidade de reconhecimento de padrões, análise de imagens, organização de dados clínicos e apoio ao planejamento terapêutico. Entretanto, seu impacto não deve ser avaliado apenas por indicadores de desempenho algorítmico, mas também por suas implicações sobre o julgamento clínico, a responsabilidade profissional e a humanização do cuidado. Este estudo teve como objetivo discutir criticamente a incorporação da inteligência artificial na odontologia a partir de quatro eixos interdependentes: eficiência clínica, julgamento humano, responsabilidade profissional e humanização do cuidado. Trata-se de uma revisão narrativa, crítica e reflexiva, orientada por princípios de transparência inspirados na ferramenta SANRA. Foram analisadas publicações científicas, documentos institucionais, marcos regulatórios e referenciais teóricos relacionados à odontologia digital, tomada de decisão clínica, viés de automação, inteligência artificial explicável, bioética e cuidado centrado no paciente. A análise indica que a inteligência artificial pode ampliar a eficiência clínica e qualificar etapas do diagnóstico e do planejamento odontológico, especialmente em tarefas baseadas em reconhecimento de padrões. Contudo, seu uso acrítico pode favorecer viés de automação, falsa sensação de segurança, deslocamento da responsabilidade decisória e enfraquecimento da relação profissional-paciente. Conclui-se que a inteligência artificial deve ser compreendida como instrumento de inteligência aumentada, destinado a apoiar, e não substituir, o julgamento clínico. Palavras-chave: Inteligência Artificial. Odontologia. Julgamento Clínico. Responsabilidade Profissional. Humanização do Cuidado.

RESUMEN

La incorporación de la inteligencia artificial representa una de las transformaciones más relevantes de la odontología contemporánea, al ampliar la capacidad de reconocimiento de patrones, análisis de imágenes, organización de datos clínicos y apoyo a la planificación terapéutica. Sin embargo, su impacto no debe evaluarse únicamente mediante indicadores de desempeño algorítmico, sino también a partir de sus implicaciones sobre el juicio clínico, la responsabilidad profesional y la humanización de la atención. Este estudio tuvo como objetivo discutir críticamente la incorporación de la inteligencia artificial en la odontología a partir de cuatro ejes interdependientes: eficiencia clínica, juicio humano, responsabilidad profesional y humanización de la atención. Se trata de una revisión narrativa, crítica y reflexiva, orientada por principios de transparencia inspirados en la herramienta SANRA. Se analizaron publicaciones científicas, documentos institucionales, marcos regulatorios y referentes teóricos relacionados con la odontología digital, la toma de decisiones clínicas, el sesgo de automatización, la inteligencia artificial explicable, la bioética y la atención centrada en el paciente. El análisis indica que la inteligencia artificial puede ampliar la eficiencia clínica y mejorar etapas del diagnóstico y de la planificación odontológica, especialmente en tareas basadas en el reconocimiento de patrones. Sin embargo, su uso acrítico puede favorecer el sesgo de automatización, una falsa sensación de seguridad, el desplazamiento de la responsabilidad en la toma de decisiones y el debilitamiento de la relación profesional-paciente. Se concluye que la inteligencia artificial debe comprenderse como un instrumento de inteligencia aumentada, destinado a apoyar, y no a sustituir, el juicio clínico. Palabras clave: Inteligencia Artificial. Odontología. Juicio Clínico. Responsabilidad Profesional. Humanización de la Atención. INTRODUCTION

Contemporary dentistry is undergoing a profound technological transformation,

driven by the progressive incorporation of digital tools capable of enhancing diagnostic

precision, optimizing clinical workflows, and supporting therapeutic planning. Intraoral

scanners, CAD/CAM systems, cone-beam computed tomography, virtual planning,

three-dimensional printing, and digital data integration have substantially changed the way

clinical information is acquired, processed, and incorporated into dental care (Mangano et

al., 2017; Joda; Brägger; Gallucci, 2015; Van Noort, 2012).

Within this context, artificial intelligence (AI) has assumed an increasingly

prominent role, particularly because of its ability to recognize patterns, interpret imaging

exams, assist in risk classification, and support clinical decision-making in areas such as

cariology, periodontology, radiology, orthodontics, implant dentistry, endodontics, and

three-dimensional image analysis (Schwendicke; Samek; Krois, 2020; Chen; Stanley; Att,

2020; Thurzo et al., 2022; Lee et al., 2025). These advances have raised expectations

regarding increased clinical efficiency, reduced diagnostic variability, and expanded

analytical capacity for dental professionals. However, the incorporation of intelligent

systems into dental practice should not be assessed solely through indicators such as

accuracy, sensitivity, or specificity, since their presence also modifies the dynamics of

clinical decision-making, professional responsibility, and the professional-patient

relationship.

Although algorithms may achieve high performance in circumscribed tasks, clinical

practice cannot be reduced to pattern recognition. Clinical judgment involves the

integration of scientific evidence, professional experience, contextual interpretation,

patient values, risk assessment, communication, ethical responsibility, and the management

of uncertainty. In dentistry, many relevant decisions are not made in the presence of

absolute certainty, but rather within probabilistic, ambiguous, and individualized scenarios.

The indication of restorative treatment, the decision to preserve or extract a tooth, implant

planning, or the interpretation of a radiographic lesion requires more than the visual

identification of signs; it requires an understanding of the patient in his or her biological,

functional, psychological, and social singularity (Croskerry, 2009; Norman; Eva, 2010;

Simpkin; Schwartzstein, 2016).

In this context, artificial intelligence introduces a central paradox. On the one hand,

it can enhance clinical efficiency, reduce perceptual errors, organize large volumes of data,

and provide relevant support for professional reasoning. On the other hand, it may favor

uncritical dependence, inappropriate displacement of decision-making responsibility, and

weakening of human judgment when used as a substitute rather than as a supportive tool.

Phenomena such as automation bias and automation complacency show that professionals

may place excessive trust in automated systems, accepting incorrect recommendations or

reducing their critical vigilance when faced with algorithmic outputs perceived as reliable

(Parasuraman; Riley, 1997; Goddard; Roudsari; Wyatt, 2012; Lyell; Coiera, 2017). Thus,

the issue lies not only in the technical capacity of artificial intelligence, but also in the way

it is integrated into the clinical process.

The ethical discussion becomes even more relevant when considering that

responsibility for care remains linked to the professional who establishes the clinical

relationship with the patient. Even when systems are regulated, validated, or commercially

disseminated, the final decision still requires human interpretation, prudence, transparent

communication, and professional responsibility. The literature on bioethics and digital

health has emphasized that the implementation of artificial intelligence in healthcare must

preserve autonomy, beneficence, non-maleficence, justice, transparency, and accountability

(Beauchamp; Childress, 2019; Char; Shah; Magnus, 2018; World Health Organization,

2021; UNESCO, 2021). In dentistry, these principles are particularly important because

clinical decision-making takes place within a fiduciary relationship, in which the patient

places trust not only in the technical competence of the dental professional, but also in his

or her capacity for judgment, listening, and care.

Given this scenario, it becomes necessary to shift the debate from professional

replacement to the paradigm of augmented intelligence. From this perspective, artificial

intelligence should not be understood as a new form of autonomous clinical intelligence,

but as an instrument capable of expanding the analytical capacity of the dental professional

while preserving the centrality of human judgment. Technology may assist in organizing

information, recognizing patterns, and reducing specific technical uncertainties, but it does

not replace clinical deliberation, ethical responsibility, or the humanization of care (Topol,

2019b; Miller, 2019; Amann et al., 2020).

Thus, this article aims to critically discuss the incorporation of artificial intelligence

in contemporary dentistry based on four central axes: clinical efficiency, human judgment,

professional responsibility, and humanized care. It is based on the premise that the future

of digital dentistry will not depend solely on the sophistication of algorithms, but on the

profession’s ability to integrate them in a critical, ethical, and patient-centered manner.

Artificial intelligence may expand information processing and support specific diagnostic

and therapeutic stages; however, the task of interpreting, deciding, communicating, and

caring in the face of the uncertainty that characterizes clinical practice remains the

responsibility of the dental professional.

Methodology

This study is characterized as a narrative literature review with a critical and

reflective approach. The bibliographic search was conducted in the first semester of 2026

and included publications related to artificial intelligence, digital dentistry, diagnostic

imaging, clinical decision-making, automation bias, explainable artificial intelligence,

bioethics, professional responsibility, governance of health technologies, and

patient-centered care. The databases PubMed/MEDLINE, Scopus, Web of Science,

Cochrane Library, SciELO, and Google Scholar were consulted, in addition to institutional

and normative documents published by recognized international organizations and

professional entities.

Controlled descriptors and free-text terms were used in Portuguese and English,

including combinations of “artificial intelligence”, “dentistry”, “clinical judgment”,

“clinical decision-making”, “automation bias”, “explainable artificial intelligence”,

“professional responsibility”, “bioethics”, and “patient-centered care”. Original articles,

reviews, conceptual studies, institutional documents, international guidelines, regulatory

frameworks, and classic works relevant to the proposed discussion were included. Studies

exclusively focused on algorithmic architecture without direct relation to clinical practice,

opinion texts without scientific support, and commercial materials used solely for product

promotion were excluded.

The final selection of references considered conceptual relevance, timeliness,

methodological quality, pertinence to the central theme, and capacity to critically support

the axes of the article. The methodological design was guided by transparency principles

applicable to narrative reviews, inspired by the SANRA tool (Scale for the Assessment of

Narrative Review Articles), proposed to assist authors, reviewers, and editors in assessing

the quality of narrative reviews (Baethge; Goldbeck-Wood; Mertens, 2019). SANRA was

not applied as a formal scoring instrument, but was used as a reference to clarify the

relevance of the topic, the objectives of the review, the search strategy, the criteria for

source selection, and the coherence of the scientific reasoning.

The selected texts were analyzed through critical reading and narrative synthesis,

with no intention of performing a meta-analysis or quantitative synthesis. The information

was organized into four analytical categories: a) clinical efficiency and applications of

artificial intelligence in dentistry; b) human judgment, clinical reasoning, and management

of uncertainty; c) professional responsibility, transparency, and risks associated with the

uncritical use of automated systems; and d) humanized care, fiduciary relationship, and

patient centrality.

DEVELOPMENT Dental artificial intelligence and clinical efficiency

The incorporation of artificial intelligence into dentistry should be understood

within the broader context of the digital transformation of clinical practice. Even before the

dissemination of intelligent systems, technologies such as intraoral scanners, CAD/CAM

systems, cone-beam computed tomography, virtual planning, three-dimensional printing,

and digital integration of clinical data had already been changing the way information is

acquired, processed, and incorporated into dental care (Van Noort, 2012; Joda; Brägger;

Gallucci, 2015; Mangano et al., 2017).

Within this scenario, artificial intelligence (AI) represents an additional stage in the

digitalization of dentistry, as it is not limited to recording or organizing information. Its

main contribution lies in its ability to recognize patterns, correlate large volumes of data,

and support clinical interpretation in specific tasks. Systems based on machine learning

and deep learning have been used for caries detection, periodontal assessment,

identification of periapical lesions, radiographic interpretation, cone-beam computed

tomography analysis, orthodontic planning, implant dentistry, and clinical decision support

(Schwendicke; Samek; Krois, 2020; Chen; Stanley; Att, 2020; Thurzo et al., 2022).

The clinical efficiency provided by AI largely derives from its ability to perform

repetitive and standardized tasks with speed, consistency, and high processing capacity. In

imaging exams, for example, algorithms can assist in the triage of radiographic alterations,

highlight regions of interest, suggest diagnostic classifications, and reduce perceptual

failures associated with fatigue, cognitive overload, or interobserver variability

(Schwendicke; Samek; Krois, 2020; Lee et al., 2025).

Beyond the academic field, artificial intelligence has already reached regulated and

commercial application in dentistry, especially in solutions aimed at dental image analysis.

The existence of systems authorized by regulatory agencies for clinical use indicates that

these technologies are already part of the reality of dental offices, clinics, and diagnostic

centers. Thus, the central question is no longer whether AI will be incorporated into

dentistry, but how it should be incorporated in a critical, safe, and ethically responsible

manner (Shujaat et al., 2025).

However, clinical efficiency should not be confused with clinical sufficiency. The

ability of an algorithm to identify patterns in images or suggest diagnostic classifications

does not correspond to the entirety of the clinical process. Dental practice involves

anamnesis, physical examination, listening to the patient, analysis of systemic factors, risk

assessment, individual preferences, and responsibility for therapeutic decision-making.

Moreover, algorithmic performance depends on the quality of the data used for system

development, training, and validation, and may present limitations when transferred to

heterogeneous clinical environments (Thurzo et al., 2022; Lee et al., 2025).

Therefore, artificial intelligence should be understood as a tool capable of

increasing clinical efficiency in dentistry, especially in tasks related to pattern recognition,

information triage, and diagnostic support. Its value, however, depends on integration with

human judgment. When used as an auxiliary instrument, it may contribute to more

informed decisions and more organized workflows; when interpreted as a substitute for

clinical deliberation, it may produce a false sense of security and inappropriately displace

professional responsibility.

Clinical judgment, automation bias, and calibrated trust

The incorporation of artificial intelligence makes it necessary to revisit a

fundamental question: what characterizes clinical judgment? At first glance, part of dental

practice may appear to be based on pattern recognition. The dental professional observes

images, identifies morphological alterations, correlates clinical signs, and formulates

diagnostic hypotheses. In this respect, AI shows relevant performance, since algorithms

can recognize statistical regularities in large volumes of data. However, reducing clinical

judgment to pattern recognition excessively simplifies the nature of decision-making in

healthcare.

Clinical judgment involves the integration of information from multiple sources:

anamnesis, physical examination, complementary exams, medical history, systemic

conditions, functional factors, psychosocial context, patient expectations, therapeutic

possibilities, and biological limits. This integration requires clinical reasoning, professional

experience, prudence, communication, and the ability to weigh risks and benefits in

situations that are often incomplete or ambiguous (Croskerry, 2009; Norman; Eva, 2010).

Uncertainty is a constitutive dimension of clinical practice. In dentistry, many

decisions do not present themselves as choices between true and false, but rather as

deliberations within a field of probabilities. The interpretation of a radiographic image, the

maintenance of a tooth with a doubtful prognosis, the indication of implants, or the choice

between monitoring and intervening requires weighing possible alternatives, rather than

the simple automatic application of protocols. In this sense, clinical judgment does not

consist merely of eliminating uncertainty, but of managing it responsibly (Simpkin;

Schwartzstein, 2016).

This ability differentiates clinical reasoning from algorithmic inference. AI systems

operate based on patterns present in the data used for their training and validation. When

applied to situations like those represented in training datasets, they may offer high

performance. However, in the face of atypical scenarios, incomplete information,

overlapping clinical conditions, or patients not adequately represented in the data, their

results may become less reliable. It is up to the clinician to recognize these zones of

indeterminacy and decide when to accept, question, or relativize an automated

recommendation.

The literature on clinical reasoning shows that decision-making involves both

intuitive and analytical processes. Fast thinking, based on pattern recognition and

accumulated experience, may be efficient in familiar situations, but it is also subject to

cognitive biases. Reflective thinking allows hypotheses to be reviewed, alternatives to be

considered, and errors in judgment to be reduced (Kahneman, 2011; Croskerry, 2009). AI

may function as a second layer of analysis, drawing attention to findings that might

otherwise go unnoticed; however, when received as a decision-making authority, it may

weaken precisely the reflective component of clinical judgment.

At this point, relevant cognitive risks emerge, such as automation bias and

automation complacency. Automation bias occurs when professionals place excessive trust

in automated recommendations, accepting results without sufficient critical verification.

Automation complacency manifests when professionals reduce their active vigilance

because they consider the system reliable enough to monitor or interpret certain

information. Both phenomena are particularly concerning in clinical contexts, in which

decisions involve uncertainty, professional responsibility, and direct consequences for the

patient (Parasuraman; Riley, 1997; Goddard; Roudsari; Wyatt, 2012; Lyell; Coiera, 2017).

In dental practice, the risk lies not only in the possibility that the algorithm may err,

but also in the professional failing to perceive that the algorithm can err. When an

automated recommendation is received as an objective conclusion rather than as auxiliary

information, a silent transfer of the decision-making burden to the machine may occur.

This transfer may weaken professional autonomy, reduce clinical investigation, and

compromise the responsibility of the dental professional toward the patient (Gaube et al.,

2021; Asan; Bayrak; Choudhury, 2020).

Calibrated trust therefore emerges as an essential competence. Trusting an

intelligent system appropriately means recognizing its usefulness without ignoring its

limitations. A system should be considered reliable when it operates within its validated

domain, with adequate data and in tasks for which it has demonstrated consistent

performance. However, it should be questioned when its recommendations diverge from

the clinical picture, when there are signs not captured by the algorithm, or when the case

departs from the patterns on which the system was trained (Hoff; Bashir, 2015).

Thus, the challenge is not to choose between trusting or distrusting artificial

intelligence, but to learn how to trust it appropriately. Excessive trust may transform AI

into an undue authority; absolute distrust may prevent the use of its potential benefits.

Between these extremes, calibrated trust allows artificial intelligence to be used as a

supportive instrument, while maintaining human judgment as the final instance of

interpretation, decision-making, and responsibility.

Explainability, professional responsibility, and humanized care

The discussion on artificial intelligence in dentistry cannot be limited to

algorithmic performance. In clinical contexts, it is as important to know whether a system

produces an adequate response as it is to understand how that response was generated,

which data influenced the recommendation, and to what extent the professional can

interpret it critically. This need makes explainable artificial intelligence, or Explainable

Artificial Intelligence (XAI), a central issue for the safe integration of AI into dental

practice.

Explainable artificial intelligence refers to a set of approaches aimed at making

algorithmic systems more understandable, interpretable, and auditable. In healthcare, this

requirement is particularly important because clinical decisions cannot be based solely on

opaque outputs, even when they are statistically accurate. When a system indicates a

possible carious lesion, bone loss, periapical lesion, or tomographic alteration, the dental

professional must have at least a minimal understanding of which elements contributed to

that indication, in order to confront it with the clinical examination, the patient’s history,

and his or her own professional reasoning (Amann et al., 2020; Antoniadi et al., 2021).

Explainability contributes to reducing opacity, favoring calibrated trust, supporting

clinical auditing, and preserving the critical role of the dental professional. However, it

should not be treated as a definitive solution to all problems related to AI in healthcare.

Visually convincing explanations may produce a false sense of understanding, especially

when the professional lacks adequate training to interpret the limits of the model.

Therefore, explainability must be associated with clinical validation, critical assessment,

professional training, and appropriate governance (Rudin, 2019; Ghassemi;

Oakden-Rayner; Beam, 2021).

The ethical and regulatory dimension reinforces this need. International documents

on AI in healthcare emphasize principles such as human supervision, transparency, safety,

equity, data protection, accountability, and preservation of autonomy. In dentistry, these

principles are relevant because clinical decisions may involve irreversible interventions

and aesthetic, functional, emotional, and financial impacts. Regulatory approval of a

system, however important, does not eliminate the need for professional interpretation in

the concrete clinical case (World Health Organization, 2021; UNESCO, 2021; European

Parliament and Council of the European Union, 2024).

Professional responsibility remains a central axis. Even when an AI system assists

in diagnosis or planning, the dental professional remains the one who establishes the

clinical relationship, interprets findings, communicates alternatives, and assumes

therapeutic decision-making before the patient. An algorithmic recommendation may

participate in the decision-making process, but it does not possess moral responsibility,

does not establish a fiduciary relationship, and does not answer to the patient (Char; Shah;

Magnus, 2018; Price; Gerke; Cohen, 2019; Gerke; Minssen; Cohen, 2020; Naik et al.,

2022). This responsibility includes transparency in the use of technology, data protection,

attention to equity, and informed consent. Algorithms trained on insufficiently diverse

datasets may show unequal performance across population groups, reproducing health

disparities. In addition, radiographic images, tomographic scans, clinical photographs,

digital models, and therapeutic records require adequate protection, legitimate purpose, and

secure storage. Algorithmic efficiency cannot be achieved at the expense of the patient’s

informational vulnerability (World Health Organization, 2021; Beauchamp; Childress,

2019).

Humanized care represents the axis that articulates these elements. AI may

contribute to reorganizing clinical workflows, reducing repetitive tasks, and allowing

greater attention to listening, communication, and shared deliberation. However, this

possibility is not automatic. If technology reduces the patient to data, images, and

probabilities, it may favor distancing and mechanization of the clinical relationship. If used

as a supportive tool, subordinated to clinical judgment, it may contribute to better-founded

decisions and greater transparency in care (Topol, 2019a; Sassi et al., 2025).

The preservation of humanized care depends directly on professional education.

Digital dentistry requires a dental professional capable of understanding the basic logic of

intelligent systems, critically interpreting their results, recognizing limitations, identifying

risks of bias, and integrating algorithmic recommendations into clinical reasoning. This

competence is not limited to the operational use of software; it involves digital literacy,

critical thinking, applied ethics, clinical communication, and the ability to make decisions

in scenarios of uncertainty (Schwendicke et al., 2023; Harte et al., 2025; Mun et al., 2025).

Therefore, digital dentistry should not move toward the replacement of clinical

practice by automation, but toward a clinical practice expanded by technology and guided

by human responsibility. Professional education must prepare the dental professional to act

as a critical interpreter of artificial intelligence, a guardian of the fiduciary relationship,

and an ethical mediator between data, evidence, and patient values. On this horizon, true

innovation will not consist merely in adopting more intelligent systems, but in building a

dental practice capable of remaining human amid technological sophistication.

of artificial intelligence, its associated risks, and the preserved role of the dental

professional in clinical decision-making.

professional

Recognizing patterns,

Confusing technical

Integrating findings with

organizing data, and

efficiency with clinical

clinical examination and the

supporting diagnosis

sufficiency

patient’s context

Transforming algorithmic

Interpreting information,

uncertainties

probability into clinical

weighing risks, and

certainty

deciding under uncertainty

Accepting automated

Maintaining vigilance,

standardized

recommendations without

questioning discrepancies,

recommendations

critical verification

and clinically validating

findings

Producing a false sense of

Assessing whether the

more transparent and

understanding or security

explanation is clinically

auditable

plausible and relevant

Supporting documentation,

Diluting or displacing

Preserving accountability

analysis, and clinical

responsibility for care

and final professional

decision-making

decision-making

Reducing the patient to

Maintaining the bond, clear

communication, and shared

data, images, or

communication, and

deliberation

probabilities

patient-centered care

Discussion

This critical narrative review analyzed the incorporation of artificial intelligence in

dentistry based on four interdependent axes: clinical efficiency, human judgment,

professional responsibility, and humanized care. Its main contribution is to shift the debate

from the mere assessment of algorithmic performance to a broader reflection on the place

This point is relevant because much of the enthusiasm surrounding artificial

The discussion of clinical judgment therefore remains central to understanding the

hypothesis formulation, critical review of information, communication of uncertainties,

and decision-making in frequently ambiguous scenarios. In many situations, the dental

professional does not decide between correct or incorrect alternatives, but among

possibilities that require weighing risks, benefits, prognosis, biological cost, and patient

expectations. From this perspective, uncertainty is not a failure of clinical practice, but a

condition inherent to healthcare (Croskerry, 2009; Norman; Eva, 2010; Simpkin;

Schwartzstein, 2016).

AI may contribute to reducing specific technical uncertainties, especially when it

assists in detecting subtle findings, comparing images, triaging exams, and organizing

clinical information. However, it may also produce an illusion of certainty when its results

are received as final conclusions. This risk is particularly important in relation to

automation bias and automation complacency, phenomena in which professionals may

place excessive trust in automated systems or reduce their critical vigilance when faced

with algorithmic recommendations perceived as reliable (Parasuraman; Riley, 1997;

Goddard; Roudsari; Wyatt, 2012; Lyell; Coiera, 2017).

In dental practice, the problem lies not only in the possibility of algorithmic failure,

but also in the cognitive relationship established between the professional and the system.

When AI is treated as a decision-making authority, clinical judgment tends to be displaced

to a secondary position, weakening professional autonomy and responsibility toward the

patient (Gaube et al., 2021; Asan; Bayrak; Choudhury, 2020). For this reason, calibrated

trust emerges as an essential competence: appropriately trusting an intelligent system

means recognizing its usefulness without ignoring its limitations, validation context, and

domain of application (Hoff; Bashir, 2015).

Explainable artificial intelligence contributes to this calibrated trust, but it does not

solve all challenges in isolation. Systems capable of presenting heat maps, regions of

interest, confidence levels, or interpretable justifications may help professionals better

understand algorithmic recommendations. However, visually convincing explanations do

not necessarily guarantee real understanding of the model’s functioning or the clinical

validity of the inference. Therefore, explainability must be associated with clinical

validation, professional education, data governance, and the preservation of human

judgment as the final instance of decision-making (Rudin, 2019; Amann et al., 2020;

Ghassemi; Oakden-Rayner; Beam, 2021).

The ethical and regulatory dimension reinforces this conclusion. International

documents on AI in healthcare emphasize principles such as human supervision,

transparency, safety, equity, data protection, accountability, and preservation of autonomy.

These principles are particularly relevant in dentistry, since clinical decisions may involve

irreversible interventions and aesthetic, functional, emotional, and financial impacts for the

patient. Regulatory approval of a system, however important, does not eliminate the need

for professional interpretation in the concrete clinical case (World Health Organization,

2021; UNESCO, 2021; European Parliament and Council of the European Union, 2024).

In this sense, responsibility for AI-mediated decisions should be understood in an

expanded manner, involving developers, manufacturers, institutions, regulators, and

professionals. However, in the direct clinical relationship, the dental professional remains

the agent responsible for integrating information, communicating alternatives, clarifying

uncertainties, and conducting therapeutic decision-making. An algorithmic

recommendation may participate in the decision-making process, but it does not possess

moral responsibility, does not establish a fiduciary relationship, and does not answer to the

patient (Char; Shah; Magnus, 2018; Price; Gerke; Cohen, 2019; Gerke; Minssen; Cohen,

2020; Naik et al., 2022).

Humanized care represents another decisive axis. The promise that artificial

intelligence may make clinical practice more human depends on how it is incorporated. If

technology reduces the patient to data, images, and probabilities, it may favor distancing

and mechanization of the clinical relationship. On the other hand, if it is used to reduce

repetitive tasks, organize information, and support better-founded decisions, it may free

time and attention for listening, communication, and shared deliberation (Topol, 2019a;

Sassi et al., 2025).

Professional education therefore becomes a strategic element for the safe

integration of artificial intelligence into dentistry. The dental professional of the future

should not be merely an operator of software, but a critical interpreter of intelligent

systems. This requires competencies in digital literacy, basic understanding of machine

learning, recognition of algorithmic biases, interpretation of results, communication of

uncertainties, applied ethics, and patient-centered decision-making (Schwendicke et al.,

2023; Harte et al., 2025; Mun et al., 2025).

As a limitation, this narrative review did not aim to perform a meta-analysis,

quantify the performance of specific systems, or statistically compare AI tools used in

different areas of dentistry. Its contribution lies in the critical integration of clinical

evidence, ethical frameworks, regulatory discussions, and concepts related to professional

judgment. Future research should evaluate how artificial intelligence systems influence

dental decisions in real-world scenarios, including diagnostic errors, adherence to

recommendations, communication with patients, informed consent, equity, professional

education, and distribution of responsibility.

Thus, artificial intelligence should be understood as one of the most relevant

transformations in contemporary dentistry, but its value will depend less on the promise of

replacing human decisions and more on its capacity to expand clinical practice responsibly.

Algorithmic efficiency may qualify diagnosis and planning; explainability may promote

transparency; regulation may establish safety parameters; and professional education may

reduce cognitive risks. However, clinical decision-making will continue to require

judgment, prudence, communication, and ethical commitment.

Conclusion

Artificial intelligence represents a relevant transformation in contemporary

dentistry, with the potential to enhance clinical efficiency, qualify information analysis, and

support diagnostic and therapeutic decisions. However, its value should not be measured

solely by its ability to recognize patterns or process large volumes of data, but by the way

it is integrated into dental care.

The findings of this critical narrative review indicate that AI should be understood

as an instrument of augmented intelligence, rather than as a substitute for clinical

judgment. Dental decision-making remains dependent on contextual interpretation,

management of uncertainty, communication with the patient, and the ethical responsibility

of the dental professional.

The safe adoption of AI requires calibrated trust, explainability, human supervision,

data governance, adequate professional education, and clarity regarding responsibility for

clinical acts mediated by technology. It is concluded that the future of digital dentistry will

not be defined only by the sophistication of intelligent systems, but by the profession’s

ability to integrate them in a critical, ethical, and humanized manner.

REFERENCES AMANN, J. et al. Explainability for artificial intelligence in healthcare: a multidisciplinary perspective. BMC Medical Informatics and Decision Making, London, v. 20, n.1, 2020. Available at: https://doi.org/10.1186/s12911-020-01332-6. Accessed on: 28 Sep. 2026. ANTONIADI, A. M. et al. Current challenges and future opportunities for XAI in machine learning-based clinical decision support systems: a systematic review. Applied Sciences, [S. l.], v. 11, n. 11, 2021. Available at: https://doi.org/10.3390/app11115088. Accessed on: 28 Sep. 2026. ASAN, O.; BAYRAK, A. E.; CHOUDHURY, A. Artificial intelligence and human trust in healthcare: focus on clinicians. Journal of Medical Internet Research, Toronto, v. 22, n. 6, e15154, 2020. Available at: https://doi.org/10.2196/15154. Accessed on: 28 Sep. 2026. BAETHGE, C.; GOLDBECK-WOOD, S.; MERTENS, S. SANRA - a scale for the quality assessment of narrative review articles. Research Integrity and Peer Review, [S. l.], v. 4, n. 5, 2019. Available at: https://doi.org/10.1186/s41073-019-0064-8. Accessed on: 28 Sep. 2026. BEAUCHAMP, T. L.; CHILDRESS, J. F. Principles of Biomedical Ethics. 8th. ed. Oxford: Oxford University Press, 2019. CHAR, D. S.; SHAH, N. H.; MAGNUS, D. Implementing machine learning in healthcare: addressing ethical challenges. The New England Journal of Medicine, Boston, v. 378, n. 11, p. 981-983, 2018. Available at: https://doi.org/10.1056/NEJMp1714229. Accessed on: 28 Sep. 2026. CHEN, Y. W.; STANLEY, K.; ATT, W. Artificial intelligence in dentistry: current applications and future perspectives. Quintessence International, Berlin, v. 51, n. 3, p. 248-257, 2020. Available at: https://doi.org/10.3290/j.qi.a43952. Accessed on: 28 Sep. 2026. CROSKERRY, P. A universal model of diagnostic reasoning. Academic Medicine, Philadelphia, v. 84, n. 8, p. 1022-1028, 2009. Available at: https://doi.org/10.1097/ACM.0b013e3181ace703. Accessed on: 28 Sep. 2026. EUROPEAN PARLIAMENT AND COUNCIL OF THE EUROPEAN UNION. Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal of the European Union, Luxembourg, L 2024/1689, 12 July 2024. Available at: http://data.europa.eu/eli/reg/2024/1689/ojs. Accessed on: 28 Sep. 2026.

GAUBE, S. et al. Do as AI say: susceptibility in deployment of clinical decision-aids. npj Digital Medicine, London, v. 4, n.1, 2021. Available at: https://doi.org/10.1038/s41746-021-00385-9. Accessed on: 28 Sep. 2026. GERKE, S.; MINSSEN, T.; COHEN, I. G. Ethical and legal challenges of artificial intelligence-driven health care. In: BOHR, A.; MEMARZADEH, K. (ed.). Artificial Intelligence in Healthcare. [S. l.]: Academic Press, 2020. Available at: https://doi.org/10.1016/B978-0-12-818438-7.00012-5. Accessed on: 28 Sep. 2026. GHASSEMI, M.; OAKDEN-RAYNER, L.; BEAM, A. L. The false hope of current approaches to explainable artificial intelligence in health care. The Lancet Digital Health, London, v. 3, n. 11, p. e745-e750, 2021. Available at: https://doi.org/10.1016/S2589-7500(21)00208-9. Accessed on: 28 Sep. 2026. GODDARD, K.; ROUDSARI, A.; WYATT, J. C. Automation bias: a systematic review of frequency, effect mediators, and mitigators. Journal of the American Medical Informatics Association, [S. l.], v. 19, n. 1, p. 121-127, 2012. Available at: https://doi.org/10.1136/amiajnl-2011-000089. Accessed on: 28 Sep. 2026. HARTE, M. et al. Transforming undergraduate dental education: the impact of artificial intelligence. British Dental Journal, London, v. 238, p. 57-60, 2025. Available at: https://doi.org/10.1038/s41415-024-7788-7. Accessed on: 28 Sep. 2026. HOFF, K. A.; BASHIR, M. Trust in automation: integrating empirical evidence on factors that influence trust. Human Factors, [S. l.], v. 57, n. 3, p. 407-434, 2015. Available at: https://doi.org/10.1177/0018720814547570. Accessed on: 28 Sep. 2026. JODA, T.; BRÄGGER, U.; GALLUCCI, G. Systematic literature review of digital three-dimensional superimposition techniques to create virtual dental patients. The International Journal of Oral & Maxillofacial Implants, Carol Stream, v. 30, n. 2, p. 330-337, 2015. Available at: https://doi.org/10.11607/jomi.3852. Accessed on: 28 Sep. 2026. KAHNEMAN, D. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux, 2011. LEE, S. J. et al. Artificial intelligence in dentistry: exploring emerging applications and future prospects. Journal of Dentistry, Oxford, v. 155, p. 105648, 2025. Available at: https://doi.org/10.1016/j.jdent.2025.105648. Accessed on: 28 Sep. 2026. LYELL, D.; COIERA, E. Automation bias and verification complexity: a systematic review. Journal of the American Medical Informatics Association, [S. l.], v. 24, n. 2, p. 423-431, 2017. Available at: https://doi.org/10.1093/jamia/ocw105. Accessed on: 28 Sep. 2026. MANGANO, F. et al. Intraoral scanners in dentistry: a review of the current literature. BMC Oral Health, London, v. 17, n. 1, p. 149, 2017. Available at: https://doi.org/10.1186/s12903-017-0442-x. Accessed on: 28 Sep. 2026.

MILLER, T. Explanation in artificial intelligence: insights from the social sciences. Artificial Intelligence, Amsterdam, v. 267, p. 1-38, 2019. Available at: https://doi.org/10.1016/j.artint.2018.07.007. Accessed on: 28 Sep. 2026. MUN, M. et al. Digital dentists: a curriculum for the 21st century. JMIR Medical Education, Toronto, v. 11, e54153, 2025. Available at: https://doi.org/10.2196/54153. Accessed on: 28 Sep. 2026. NAIK, N. et al. Legal and ethical consideration in artificial intelligence in healthcare: who takes responsibility? Frontiers in Surgery, [S. l.], v. 9, p. 862322, 2022. Available at: https://doi.org/10.3389/fsurg.2022.862322. Accessed on: 28 Sep. 2026. NORMAN, G. R.; EVA, K. W. Diagnostic error and clinical reasoning. Medical Education, Oxford, v. 44, n. 1, p. 94-100, 2010. Available at: https://doi.org/10.1111/j.1365-2923.2009.03507.x. Accessed on: 28 Sep. 2026. PARASURAMAN, R.; RILEY, V. Humans and automation: use, misuse, disuse, abuse. Human Factors, [S. l.], v. 39, n. 2, p. 230-253, 1997. Available at: https://doi.org/10.1518/001872097778543886. Accessed on: 28 Sep. 2026. PRICE, W. N., 2nd; GERKE, S.; COHEN, I. G. Potential liability for physicians using artificial intelligence. JAMA, [S. l.], v. 322, n. 18, p. 1765-1766, 2019. Available at: https://doi.org/10.1001/jama.2019.15064. Accessed on: 28 Sep. 2026. RUDIN, C. Stop explaining black box machine learning models for high-stakes decisions and use interpretable models instead. Nature Machine Intelligence, [S. l.], v. 1, n. 5, p. 206-215, 2019. Available at: https://doi.org/10.1038/s42256-019-0048-x. Accessed on: 28 Sep. 2026. SASSI, Z. et al. Human-centered AI in healthcare: empowering patients and support persons in clinical decision-making. BMC Medical Informatics and Decision Making, London, v. 25, n. 1, 2025. Available at: https://doi.org/10.1186/s12911-025-03298-9. Accessed on: 28 Sep. 2026. SCHWENDICKE, F.; SAMEK, W.; KROIS, J. Artificial intelligence in dentistry: chances and challenges. Journal of Dental Research, Alexandria, v. 99, n. 7, p. 769-774, 2020. Available at: https://doi.org/10.1177/0022034520915714. Accessed on: 28 Sep. 2026. SCHWENDICKE, F. et al. Artificial intelligence for oral and dental healthcare: core education curriculum. Journal of Dentistry, Oxford, v. 128, 2023. Available at: https://doi.org/10.1016/j.jdent.2022.104363. Accessed on: 28 Sep. 2026. SHUJAAT, S. et al. FDA-approved AI solutions in dental imaging: a narrative review of applications, evidence, and outlook. International Dental Journal, [S. l.], v. 76, n. 1, 2026. Available at: https://doi.org/10.1016/j.identj.2025.109315. Accessed on: 28 Sep. 2026.

SIMPKIN, A. L.; SCHWARTZSTEIN, R. M. Tolerating uncertainty: the next medical revolution? The New England Journal of Medicine, Boston, v. 375, n. 18, p. 1713-1715, 2016. Available at: https://doi.org/10.1056/NEJMp1606402. Accessed on: 28 Sep. 2026. THURZO, A. et al. Where is the artificial intelligence applied in dentistry? Systematic review and literature analysis. Healthcare, Basel, v. 10, n. 7, 2022. Available at: https://doi.org/10.3390/healthcare10071269. Accessed on: 28 Sep. 2026. TOPOL, E. J. Deep Medicine: how artificial intelligence can make healthcare human again. New York: Basic Books, 2019a. TOPOL, E. J. High-performance medicine: the convergence of human and artificial intelligence. Nature Medicine, London, v. 25, n. 1, p. 44-56, 2019b. Available at: https://doi.org/10.1038/s41591-018-0300-7. Accessed on: 28 Sep. 2026. UNESCO. Recommendation on the ethics of artificial intelligence. Paris: UNESCO, 2021. Available at: https://unesdoc.unesco.org/ark:/48223/pf0000381137. Accessed on: 28 Sep. 2026. VAN NOORT, R. The future of dental devices is digital. Dental Materials, Oxford, v. 28, n. 1, p. 3-12, 2012. Available at: https://doi.org/10.1016/j.dental.2011.10.014. Accessed on: 28 Sep. 2026. WORLD HEALTH ORGANIZATION. Ethics and governance of artificial intelligence for health: WHO guidance. Geneva: WHO, 2021. Available at: https://iris.who.int/handle/10665/341996. Accessed on: 28 Sep. 2026.