The Job Description of the Diagnostic Radiologist in 2035
Three experts wrote it three ways, and they don't agree. Five questions tell you which future you're betting on.
How to take part
The idea
Nobody knows what this job becomes.
So we wrote it three ways.
Three people wrote three versions of the same job description: the diagnostic radiologist in 2035. Two are radiologists. One is a health-tech professional with 20+ years in medical imaging IT. The vantage points are different on purpose, because the future of this job is not only a clinical question. It is also an economic one.
One version is optimistic: the radiologist grows into an orchestrator of AI. One is the trade-off view, written from the medical imaging IT side: the clinical job gets better, but there will be fewer radiologists. One is pessimistic: the profession splits into tiers, from a scarce on-site elite to a large, low-paid remote audit workforce.
Answer five questions about what you believe will be true. We show you the version that matches your outlook. Then you can read all three and compare them side by side.
The test
Which 2035 do you believe in?
Five questions. Pick the answer closest to what you expect. There are no right answers, only the future you find most likely.
The versions
Three job descriptions for 2035.
Thanks — counted.
01 · Summary
Position summary.
By 2035, the radiologist's role has evolved from traditional worklist-based image reporting to that of a diagnostic orchestrator, leveraging AI as a co-pilot to augment image interpretation, integrate multimodal data, and turn complexity into clear, actionable insights for clinicians and patients.
A new generation of AI tools supports radiologists by enabling faster image acquisition and lower doses while maintaining high quality, enhancing image analysis through automated lesions detection, quantification, and characterization, and generating structured preliminary reports ready for review and validation, including patient-friendly summaries. These tools also provide prognostic and predictive insights based on multimodal data.
Radiologists remain the primary decision-makers, retaining full control over AI outputs and their use, and leading communication with both clinicians and patients. As automation expands in routine cases, their focus increasingly shifts toward complex diagnostics, edge cases, multidisciplinary decision-making, and interventional procedures.
As AI handles more of the routine read, the radiologist has an opportunity, and a responsibility, to become more visible: known by name to patients and referrers, not just as a signature on a report. The human touch becomes a differentiator, not a given.
02 · Responsibilities
Key responsibilities.
01Timely diagnostic imaging reporting and communication, with AI as a co-pilot, and expansion of interventional radiology
- Near-immediate turnaround for emergency cases, and within hours for non-emergent studies
- High diagnostic accuracy with no critical misses
- AI-enabled workflow optimization, including: worklist prioritization, detection and alerting of emergent findings, automated image analysis, generation of preliminary draft reports, AI-augmented voice recognition
- Flexible reading models, combining subspecialty expertise and generalist coverage to optimize precision and efficiency
- Contribute to patient-facing radiology, including direct communication, education, and shared decision-making when appropriate
- Expansion into image-guided procedures and minimally invasive therapies as a core domain of human-added value
02Active participation in radiology practice organization and business strategy
- Attend monthly/quarterly multidisciplinary team meetings with radiologists, technologists, administrative staff, and management
- Contribute to continuous improvement of workflows, operations, and organizational processes to enhance efficiency, quality, and workplace well-being
- Leverage data and performance metrics (e.g., turnaround times, quality indicators) to inform decision-making and optimize resource allocation
- Collaborate in the integration and evaluation of new technologies, including AI tools, to ensure clinical relevance and operational value
- Contribute to strategic discussions on service development, growth opportunities, and patient-centered care models
03AI stewardship, oversight, and co-development
- Participate in the selection, validation, testing, and implementation of AI solutions
- Monitor AI performance in real-world settings and identify, escalate, and document any drift or degradation
- Contribute to continuous post-deployment surveillance, including quality assurance and regulatory compliance
- Collaborate with industry partners to suggest new features, improvements, and clinical use cases, and engage in co-development initiatives when relevant
- Build, with institutional support, AI-augmented workflows across defined levels of automation (L0–L4), with clear go/no-go criteria, guardrails, and rollback protocols
- Translate automation gains into reduced turnaround times and backlog, while stewarding dose optimization, scanner time, computational efficiency, and appropriate imaging utilization
- Ensure ethical, transparent, and fair use of AI, including awareness of bias, explainability, and patient consent
04Multidisciplinary collaboration and data integration
- Participate in multidisciplinary meetings (MDMs) as a diagnostic orchestrator
- Take ownership of the organization and efficiency of MDMs, including multimodal data flow and integration of AI tools
- Synthesize imaging with pathology, genomics, laboratory, and clinical data to support comprehensive diagnosis
- Deliver AI-augmented insights combining radiology and multimodal data
- Act as a bridge between data scientists and clinicians, supporting precision medicine and cost-effective care
05Quality assurance, education, and leadership
- Participate in peer review and discrepancy learning loops to support continuous improvement and a strong safety culture
- Engage in structured peer learning through case sharing, multidisciplinary discussions, and presentations
- Mentor and support colleagues across experience levels, fostering knowledge transmission and professional development
- Contribute to the development of best practices, guidelines, and quality standards within the organization
- Commit to continuous learning, including AI fluency, data science fundamentals, and a high-level understanding of adjacent disciplines (e.g., pathology, genomics)
- Promote a culture of openness, feedback, and collective intelligence, embracing innovation and lifelong learning
- Contribute to research, publications, and thought leadership in AI and imaging
06Professional attitude and work culture
- Foster a positive, respectful, and collaborative attitude toward colleagues, partners, and patients
- Contribute to a healthy, inclusive, and supportive work environment, promoting teamwork and psychological safety
- Take responsibility for personal well-being, including physical and mental health, and make appropriate use of available resources
- Demonstrate collegiality, mutual support, and accountability within the team
- Embrace professionalism, adaptability, and resilience in a rapidly evolving clinical and technological environment
03 · Required
Required qualifications.
- MD (or equivalent) with board certification in Diagnostic Radiology; eligible for licensure in the country of practice
- Demonstrated AI fluency, evidenced by recognized coursework or micro-credential in clinical AI or imaging informatics (or commitment to complete within 12 months, with institutional support)
- Excellent communication skills (both clinician- and patient-facing), with proficiency in structured and quantitative reporting
04 · Preferred
Preferred qualifications.
- Fellowship training (e.g., oncologic, neuro, cardiothoracic, pediatric radiology) and/or procedural expertise as locally required
- Fluency with DICOMweb, HL7 FHIR, and structured vocabularies (e.g., RadLex; indication-specific systems such as BI-RADS, LI-RADS, PI-RADS where relevant)
- Experience in AI governance and post-market surveillance, including dataset curation, validation, and performance monitoring
- Exposure to quantitative imaging (e.g., QIBA), multi-site or teleradiology workflows, and secure remote reading environments
- Leadership experience in multidisciplinary settings, including collaboration with IT and enterprise architecture teams on PACS, VNA, RIS, and reporting integrations
- Familiarity with value-based care models and health economics, including the impact of imaging and AI on outcomes and cost-effectiveness
05 · Environment
Work environment and conditions.
- Hybrid and remote work options, with institutional support for home workstations
- Teleradiology across sites, with opportunities for subspecialty reading based on expertise and preference
- Protected time and silent zones for deep reporting work
- AI-enabled safety net ("second read") operating in the background to detect potential errors and provide timely alerts
- Access to advanced data and AI infrastructure
- Continuous learning supported by AI-curated CME, with protected time (e.g., 3–4 weeks/year) and participation in innovation programs
- Dedicated protected time for future skills development, including cross-training toward higher-complexity diagnostics and procedural pathways
- Financial and time support for leadership, business, and professional development
- Access to mental health resources and initiatives promoting a healthy workplace
- Dedicated time for non-reporting and non-clinical tasks
06 · Success
Success measures.
What "good" looks like.
- Turnaround time (TAT)
- <2 hours for routine cases; near-real-time for emergencies
- Diagnostic quality
- Low rate of clinically significant errors, as assessed through peer review and AI-supported monitoring
- Referrer engagement
- Growth in referring physicians and high retention rates
- Satisfaction
- High patient and clinician satisfaction, measured through NPS or equivalent metrics
- AI adoption and impact
- Successful integration and consistent use of AI tools, with demonstrated improvements in efficiency and quality
- Multidisciplinary engagement
- High participation and leadership in MDTs
- Workforce well-being
- Low absenteeism and strong indicators of team engagement
- Skills development
- Continuous expansion of expertise across clinical, technological (AI/data), and leadership/business domains
- Professional impact
- Active involvement in institutional, local, national, or international committees, working groups, or task forces
07 · Benefits
Benefits and perks.
- Competitive compensation aligned with clinical excellence, leadership, innovation, and measurable impact
- Performance-based bonuses tied to the achievement of defined objectives
- Flexible hybrid work model, with support for professional development (AI-curated CME and protected learning time)
- Access to mental health support, including digital tools and counseling services
- Growth stipend and dedicated time for cross-training in complex diagnostics, procedures, and adjacent skills
- Additional benefits may include healthy meals and beverages, and flexible vacation policies based on performance, where permitted
08 · Equal opportunity
Equal opportunity statement.
We are committed to equal opportunity principles. All qualified applicants will be considered for employment without regard to legally protected characteristics, in accordance with applicable laws in the country of employment.
Global Head of AI Advocacy, GE HealthCare. 20+ years in medical imaging IT. Not a radiologist; writes from the imaging-tech side.
The trade-off view
01 · Summary
Position summary.
By 2035, the radiologist's role is still defined by clinical judgment. What has changed is the threshold for when that judgment is needed. AI manages the routine read across defined modalities and indications. What reaches the radiologist's desk is harder: rare presentations, multi-system complexity, studies where imaging and clinical context don't align, and cases where the cost of being wrong is highest.
The radiologist is the final clinical authority. Not the first reader on most studies, but the accountable one. That means carrying responsibility for AI outputs validated under their name alongside direct interpretation of the cases AI can't handle. Both are real functions. Neither is optional.
As routine reporting contracts, the consultative role expands. Radiologists in 2035 spend more time in clinical conversations, multidisciplinary meetings, and direct patient communication than their 2030 counterparts. The profession is more visible, more integrated, and more exposed to the consequences of getting it wrong.
The economics of imaging AI point toward more output per radiologist, which means fewer radiologists per department. The job described here is a better version of the 2030 role for the people who hold it. There will be fewer of them.
02 · Responsibilities
Key responsibilities.
01Diagnostic reporting and complex case interpretation
- High-accuracy interpretation of complex, ambiguous, and rare presentations across modalities, with no critical misses
- Final clinical authority on AI-generated reports for routine indications, with full accountability for validated outputs
- Maintenance of independent search patterns and clinical vigilance on AI-processed studies; AI-positive results require scrutiny, not acceptance
- Direct interpretation of cases outside AI training distributions: unusual anatomy, rare diagnoses, multi-system presentations, and populations underrepresented in training data
- Near-real-time turnaround for emergency cases; defined windows for AI-processed routine studies
- Structured, quantitative reporting with direct communication to referring clinicians on complex and high-stakes findings
- Expansion into image-guided procedures and minimally invasive therapies as a core domain of clinical added value
02Multidisciplinary collaboration and clinical integration
- Active participation in MDMs as the primary imaging voice, with ownership of imaging data quality and presentation
- Synthesis of imaging findings with pathology, genomics, laboratory, and clinical data to support diagnosis and treatment decisions
- Direct patient communication in cases that require it, including findings escalated from the AI pipeline
- Contribution to precision medicine workflows where multimodal integration adds value beyond the imaging report alone
- Bridge between imaging data and clinical decision-making, with enough context to be useful in both directions
03Practice organization, AI governance, and quality
- Participation in the evaluation and deployment decisions for AI tools in clinical use, with clinical judgment as the primary input
- Ongoing monitoring of AI output quality: detection of performance drift, systematic errors, and failure modes in real-world deployment
- Escalation of AI degradation events with documented evidence to governance committees and vendors
- Contribution to workflow design with awareness of departmental cost, throughput, and staffing constraints
- Participation in peer review, discrepancy learning, and quality assurance processes
- Contribution to post-market surveillance reporting as required by applicable regulatory frameworks
04Education, research, and professional development
- Mentorship of colleagues across experience levels, with emphasis on clinical judgment in AI-assisted environments
- Continuous learning across complex diagnostics, procedural pathways, and AI governance
- Contribution to a culture of accountability, including open review of errors and near-misses
- Participation in research, publications, or institutional and national working groups where relevant
03 · Required
Required qualifications.
- MD (or equivalent) with board certification in Diagnostic Radiology; eligible for licensure in the country of practice
- Subspecialty depth sufficient to handle the complexity mix that reaches the radiologist's desk when routine cases are AI-managed
- Demonstrated AI fluency, evidenced by recognized coursework or micro-credential in clinical AI or imaging informatics (or commitment to complete within 12 months, with institutional support)
- Capacity to carry legal and clinical responsibility for AI-validated outputs, including cases where errors originated upstream
04 · Preferred
Preferred qualifications.
- Fellowship training in a domain where AI performance remains limited, including complex oncologic staging, rare neurological presentations, or pediatric imaging
- Experience with post-market surveillance: drift detection, degradation documentation, and audit trail construction
- Familiarity with liability frameworks governing AI-assisted clinical sign-off in the country of practice
- Familiarity with health economics and value-based care models, including the cost and outcome implications of AI deployment in imaging
- Leadership experience in multidisciplinary or institutional settings
- Fluency with DICOMweb, HL7 FHIR, and structured vocabularies (e.g., RadLex; BI-RADS, LI-RADS, PI-RADS where relevant)
05 · Environment
Work environment.
- Reading queue weighted toward complex and flagged cases, with routine studies processed by the AI pipeline before they arrive
- Productivity benchmarked against AI-validated throughput baselines; volume per radiologist is higher than 2030
- Smaller departments than 2030 equivalents, with shared accountability across a tighter team
- More direct clinical contact than traditional reading room models, with protected time for MDM participation and clinician communication
- Teleradiology across sites, with subspecialty reading matched to case complexity and expertise
- Protected time for continuing education, skills development, and research; supported by AI-curated CME
06 · Success
Success measures.
What "good" looks like.
- Diagnostic accuracy
- Low rate of clinically significant errors, with particular weight on complex cases where AI provides no safety net; assessed through peer review and AI-supported monitoring
- Turnaround time
- Near-real-time for emergencies; within defined windows for routine AI-processed studies
- Clinical integration
- MDM participation, referrer engagement, and direct patient communication; growth in both volume and quality over time
- Cost per exam
- Throughput efficiency relative to departmental targets; a real constraint, not a secondary metric
- Referrer satisfaction
- High retention and engagement among referring clinicians; the radiologist is known by name, not just by report
- Workforce sustainability
- Team retention, absenteeism, and the ability to maintain a functioning department at reduced FTE count
07 · Benefits
Benefits.
- Competitive compensation reflecting diagnostic complexity, throughput, and clinical leadership contribution
- A worklist weighted toward complex cases; routine volume is handled before it reaches you
- Greater clinical visibility than 2030 reading room roles, with expanded MDM and patient-facing responsibilities
- Protected time for skills development across complex diagnostics, procedural pathways, and AI governance
- Access to advanced imaging and AI infrastructure
- Smaller, specialized teams with higher per-person accountability and less administrative overhead
08 · Equal opportunity
Equal opportunity statement.
We are committed to equal opportunity principles. All qualified applicants will be considered for employment without regard to legally protected characteristics, in accordance with applicable laws in the country of employment.
The structure
Three tiers, not one job.
Christoph Agten's version does not describe a single role. It describes a profession that has split into tiers: a scarce, elite on-site subspecialist layer; a mid-tier hands-on procedural layer; and a large, low-paid remote workforce auditing the output of autonomous AI. The three job descriptions below are the three tiers.
Tier 1
Senior Subspecialist & AI Workflow Consultant
Elite tier · 100% on-site · high-scarcity premium base salary
Tier 1
Senior Subspecialist & AI Workflow Consultant
Elite tier · 100% on-site · high-scarcity premium base salary
01Position summary
The Tier 1 Senior Subspecialist functions as the ultimate diagnostic adjudicator, clinical consultant, and legal signatory for the hospital's autonomous imaging pipelines. This role focuses exclusively on highly complex, ambiguous subspecialty "edge cases" (primarily advanced MRI and complex CT) that have been flagged by AI as borderline, rare, or high-risk. Tier 1 radiologists provide visible clinical authority in face-to-face multidisciplinary teams and hold ultimate responsibility for the governance of the entire departmental workflow.
02Core responsibilities
- Ambiguous case adjudication
- Interpret and definitively resolve high-complexity, multi-parametric MRI and advanced CT cases flagged by AI systems as ambiguous, atypical, or borderline.
- On-site multidisciplinary leadership
- Take full ownership of face-to-face Multidisciplinary Team Meetings (MDTMs), acting as the definitive diagnostic voice alongside surgeons, oncologists, and other clinical peers.
- AI pipeline governance
- Hold ultimate legal and clinical responsibility for autonomous, zero-click AI reporting. Direct and review the quality-control data compiled by the remote Radiology Auditor pool.
- Optional intervention
- Perform high-level, advanced image-guided therapeutic procedures as locally required.
03Required qualifications and skills
- MD (or equivalent) with Board Certification in Diagnostic Radiology and valid licensure
- Completed subspecialty fellowship (e.g., musculoskeletal radiology, neuroradiology, pediatric radiology, or oncologic imaging)
- Demonstrated AI fluency and governance competency, evidenced by recognized credentials in imaging informatics or algorithmic auditing
- Exceptional communication and high-EQ social skills, with a proven track record of leading discussions and standing one's ground in high-pressure clinical environments
04Environment, compensation, and benefits
- Work structure
- 100% on-site for clinical collaboration and face-to-face MDT meetings. Normal office hours, with background subspecialty on-call rotation for nights and weekends.
- Compensation
- Exceptionally high, fixed base salary reflecting immense systemic responsibility. Non-performance based.
- Vacation and education
- 5–6 weeks of annual leave; 2 weeks of paid educational leave.
Tier 2
Clinical Radiology Specialist
Mid-tier · 100% on-site · market-standard fixed salary
Tier 2
Clinical Radiology Specialist
Mid-tier · 100% on-site · market-standard fixed salary
01Position summary
The Clinical Radiology Specialist is the physical engine of the on-site department. Because AI cannot physically manipulate a needle or hold an ultrasound probe, this role focuses entirely on operator-dependent diagnostic imaging and minor image-guided interventions. Clinical Radiology Specialists manage the heavy daily volume of patient-facing clinical workflows, serving as the hands-on execution layer of the hospital, with complex diagnostic escalations routed to Tier 1.
02Core responsibilities
- High-volume ultrasound operations
- Perform and interpret real-time diagnostic ultrasound examinations (abdominal, pelvic, vascular, small parts).
- Minor image-guided interventions
- Independently perform routine needle-based procedures on-site, including musculoskeletal infiltrations, therapeutic joint injections, superficial biopsies (breast, thyroid, lymph node), and fluid aspirations.
- Front-line emergency and trauma triage
- Provide mandatory, shift-based on-site coverage for nights and weekends, handling initial emergency trauma triage alongside AI decision-support tools.
- MDT support
- Attend local multidisciplinary team meetings to provide procedural and ultrasound-specific perspectives on active patient cases.
03Required qualifications and skills
- MD (or equivalent) with Board Certification in Diagnostic Radiology
- Demonstrated technical competency in interventions: documented logbook proof of proficiency in ultrasound- and CT-guided injections, infiltrations, and core and fine-needle biopsies
- High manual dexterity: absolute comfort with real-time, operator-dependent diagnostic scanning and needle tracking
04Environment, compensation, and benefits
- Work structure
- 100% on-site due to the physical nature of ultrasound and needle interventions. Shift-based schedule including regular nights, weekends, and holidays.
- Compensation
- Moderate, market-standard fixed salary driven by a stable but highly populated applicant pool of procedurally-focused radiologists.
- Vacation and education
- 5–6 weeks of annual leave; 2 weeks of paid educational leave, focused on advanced procedural techniques.
Tier 3
Radiology Auditor
High-supply tier · 100% remote teleradiology · low salary, just above resident level
Tier 3
Radiology Auditor
High-supply tier · 100% remote teleradiology · low salary, just above resident level
01Position summary
The Radiology Auditor forms the massive, distributed human-in-the-loop safety net for the hospital enterprise. Operating entirely via remote teleradiology workstations, this workforce manages the continuous data streams produced by autonomous AI models. Because the global supply of radiologists drastically exceeds the number of elite on-site roles, this position is highly competitive yet commands a very low salary meant strictly for high-volume quality policing.
02Core responsibilities
- Massive-scale algorithmic auditing
- Conduct continuous, high-speed manual audits of the AI pipeline's "high-confidence normals." Review a randomized daily quota (2–5%) of autonomous, AI-signed plain films, routine chest CTs, and screening mammograms to catch silent model degradation or drift.
- First-line error logging and triage
- Screen automated system flags and discrepancies. Route verified errors, software hallucinations, or unexpected anatomical variances out of the automated pipeline and escalate them to the appropriate tier.
- Preliminary draft verification
- For non-normal, non-edge cases, review, edit, and approve AI-generated structured preliminary drafts to prepare them for final systemic routing.
03Required qualifications and skills
- MD (or equivalent) with basic Board Certification or eligibility in Radiology
- Basic AI systems literacy: familiarity with DICOMweb viewer platforms, algorithmic error logging, and high-volume structured template workflows
- High visual endurance: ability to maintain strict diagnostic focus over prolonged periods of high-volume, screen-based data review
04Environment, compensation, and benefits
- Work structure
- 100% remote teleradiology. Flexible, contract-based or shift-based hour models managed via digital worklists. No on-site hospital or patient interaction required.
- Compensation
- Very low, volume-insulated base salary, positioned just slightly above traditional resident and trainee pay scales due to the massive surplus of remote applicants.
- Vacation
- 5–6 weeks of standard unpaid or pro-rated leave depending on regional labor-law baselines.
Side by side
Where the three versions disagree.
More dots mean more of that dimension. Read across each row to see how the three futures differ.
| Dimension | OptimisticKorchi | Trade-offBeger | PessimisticAgten |
|---|---|---|---|
| AI autonomy on routine reads | |||
| Radiologist clinical authority | |||
| Direct patient contact | |||
| Interventional and procedural work | |||
| AI governance and oversight load | |||
| Radiologist headcount | |||
| Career security | |||
| Back a view |
Highlighted column = the version your answers matched.
All three columns are read from the authors' full texts. Agten's version describes a tiered profession rather than one job, so his column shows the typical radiologist's experience averaged across the three tiers. The tiers themselves are set out in full in his version above.
From the authors
This is a draft. Tell us which future you would bet on.
Three versions, three vantage points. We do not agree with each other, and that is the point. The most useful thing you can do is read the one the test gave you, then read the one you disagree with most.
Practising radiologists, educators, technologists, health-system leaders: tell us what is missing, what is wrong, and which version you think will turn out closest to true.
Take the argument to LinkedIn.
Post where you disagree, tag the three of us, and use #radiologist2035. We read every post and we reply. The disagreement is the point.
