Radiologists

Review or transmit images and information using picture archiving or communications systems.

AI: Fully automatable - Reviewing, routing, and transmitting images via PACS is a software task that can be fully automated using existing AI/IT integrations and RPA tools.

Document the performance, interpretation, or outcomes of all procedures performed.

AI: Fully automatable - Automated reporting and structured documentation from imaging and procedure metadata are well within current AI capabilities and can fully produce records for review.

Interpret images using computer-aided detection or diagnosis systems.

AI: Fully automatable - Modern computer‑aided detection/diagnosis systems and AI models are capable of performing image interpretation tasks at clinically useful levels and are deployed to generate automated reads in many contexts.

Schedule examinations and assign radiologic personnel.

AI: Fully automatable - AI-driven scheduling and workforce-optimization systems can automatically schedule examinations and assign radiologic personnel according to constraints and priorities.

Provide advice on types or quantities of radiology equipment needed to maintain facilities.

AI: Fully automatable - AI can analyze utilization, maintenance, and throughput data to provide accurate recommendations on types and quantities of radiology equipment needed for facilities.

Obtain patients' histories from electronic records, patient interviews, dictated reports, or by communicating with referring clinicians.

AI: Partial - AI can reliably extract histories from electronic records and perform preliminary interviews, but obtaining nuanced clinical histories and resolving inconsistencies generally requires clinician interaction.

Prepare comprehensive interpretive reports of findings.

AI: Partial - AI can automatically interpret many imaging findings and draft reports, but producing comprehensive interpretive reports that integrate complex clinical context and assume legal responsibility still requires radiologist oversight.

Perform or interpret the outcomes of diagnostic imaging procedures including magnetic resonance imaging (MRI), computer tomography (CT), positron emission tomography (PET), nuclear cardiology treadmill studies, mammography, or ultrasound.

AI: Partial - By 2025 AI can accurately interpret many imaging studies and assist heavily, but it cannot physically perform procedures or fully replace radiologist judgment and responsibility in complex or ambiguous cases.

Communicate examination results or diagnostic information to referring physicians, patients, or families.

AI: Partial - AI can generate and deliver report summaries and notifications to clinicians and patients, but sensitive conversations and final communications typically require human oversight and empathy.

Evaluate medical information to determine patients' risk factors, such as allergies to contrast agents, or to make decisions regarding the appropriateness of procedures.

AI: Partial - AI tools can screen EHR data for allergies and suggest appropriateness of procedures, but incomplete records, clinical nuance, and liability constraints limit full automation.

Provide counseling to radiologic patients to explain the processes, risks, benefits, or alternative treatments.

AI: Partial - AI can provide tailored explanations of procedures, risks, and alternatives, yet counseling that requires nuance, consent validation, and emotional support still needs human involvement.

Perform interventional procedures such as image-guided biopsy, percutaneous transluminal angioplasty, transhepatic biliary drainage, or nephrostomy catheter placement.

AI: Partial - AI can assist with planning, image guidance, and robotic navigation for interventional procedures but cannot autonomously perform complex, hands‑on interventions or manage unexpected complications and sterile technique.

Administer or maintain conscious sedation during and after procedures.

AI: Partial - AI can monitor physiologic parameters and suggest dosing adjustments for conscious sedation but cannot legally or safely assume full responsibility for administering and managing sedation without human oversight.

Instruct radiologic staff in desired techniques, positions, or projections.

AI: Partial - AI and AR guidance can instruct staff on positioning and technique, but hands-on training, adaptation to unusual patients, and supervisory responsibility remain human tasks.

Confer with medical professionals regarding image-based diagnoses.

AI: Partial - AI can prepare analyses and differential diagnoses to support interdisciplinary discussion, but it cannot fully substitute for clinician-to-clinician conferencing and joint decision-making.

Coordinate radiological services with other medical activities.

AI: Partial - Routine coordination (scheduling, triage, notifications) can be largely automated, but complex coordination across services and exception handling still require human judgment.

Establish or enforce standards for protection of patients or personnel.

AI: Partial - AI can monitor compliance and assist enforcement of safety protocols, but establishing standards and assuming ultimate enforcement responsibility remains a human and regulatory function.

Develop or monitor procedures to ensure adequate quality control of images.

AI: Partial - AI can analyze image quality and continuously monitor QC metrics to support procedures, but cannot fully replace human oversight, regulatory responsibility, and protocol development.

Recognize or treat complications during and after procedures, including blood pressure problems, pain, oversedation, or bleeding.

AI: Partial - AI can detect physiological changes and alert clinicians or suggest interventions, but cannot perform hands-on treatment or emergent interventions required to manage complications.

Serve as an offsite teleradiologist for facilities that do not have on-site radiologists.

AI: Partial - AI can provide reliable preliminary teleradiology reads and triage but cannot fully replace an offsite radiologist's clinical responsibility, final sign‑off, and medico‑legal accountability in most settings.

Develop treatment plans for radiology patients.

AI: Partial - AI can generate and optimize radiology treatment recommendations and protocols, but developing and owning a patient’s comprehensive treatment plan requires multidisciplinary judgment and clinician responsibility.

Participate in continuing education activities to maintain and develop expertise.

AI: Partial - AI can generate, personalize, and deliver continuing-education content and simulations, but cannot replace the human professional judgment, reflection, and credentialing aspects of continuing education participation.

Treat malignant internal or external growths by exposure to radiation from radiographs (x-rays), high energy sources, or natural or synthetic radioisotopes.

AI: Partial - AI substantially automates radiation therapy planning and optimization but cannot autonomously deliver or oversee therapeutic radiation and manage treatment‑level decisions and acute complications without human clinicians.

Participate in quality improvement activities including discussions of areas where risk of error is high.

AI: Partial - AI can analyze data to identify high-risk areas and produce actionable recommendations for quality improvement, but cannot fully substitute for human-led multidisciplinary discussions and decision-making.

Conduct physical examinations to inform decisions about appropriate procedures.

AI: Partial - AI can augment remote assessments and interpret sensor data, but it cannot perform full in‑person physical examinations (palpation, specialized maneuvers) or substitute for the nuanced exam findings clinicians use to decide procedures.

Supervise and teach residents or medical students.

AI: Partial - AI can provide instruction, case review, and formative feedback to trainees, but cannot fully replicate the supervisory, evaluative, and mentorship responsibilities of attending physicians.

Implement protocols in areas such as drugs, resuscitation, emergencies, power failures, or infection control.

AI: Partial - AI can draft protocols and coordinate logistical plans for drugs, resuscitation, and emergencies, but cannot assume institutional authority or perform the physical/administrative implementation and accountability.

Participate in research projects involving radiology.

AI: Partial - AI can perform literature reviews, image analysis, stats, and draft manuscripts to substantially participate in radiology research, but cannot independently carry out clinical research activities that require human oversight, consent, and IRB leadership.

Administer radiopaque substances by injection, orally, or as enemas to render internal structures and organs visible on x-ray films or fluoroscopic screens.

AI: Not automatable - Administering contrast agents is a hands-on, licensed clinical procedure that AI cannot physically perform or be authorized to carry out.