Nuclear Power Reactor Operators
51-8011.00Median wage $122,890/yr5,150 employed (US)Rank #909 of 923 scored · top 98% by substitution
Operate or control nuclear reactors. Move control rods, start and stop equipment, monitor and adjust controls, and record data in logs. Implement emergency procedures when needed. May respond to abnormalities, determine cause, and recommend corrective action.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution7
Exposure10
Augmentation43
Substitution — the headline: capability discounted by cost, barriers and adoption.
Exposure — technical capability alone, regardless of whether anyone deploys it.
Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.
Tasks on the substitution scale
19 rated tasks, binned by substitution score.
Position among all scored occupations
Distribution of 923 occupation scores; the marker is this occupation.
Tasks with substitution ≥ 70
0%
Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.
Why this score
The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.
Task automatabilityw 35%10
panel mean rating 1.4/5 → substitution pressure 10/100
Technical feasibility todayw 20%9
panel mean rating 1.4/5 → substitution pressure 9/100
Cost vs. human wagew 15%8
panel mean rating 1.3/5 → substitution pressure 8/100
Adoption barriersw 20%inverted — strong barriers lower the score1
panel mean rating 5.0/5 (barrier strength) → substitution pressure 1/100
Sector adoption velocityw 10%2
panel mean rating 1.1/5 → substitution pressure 2/100
Task breakdown (19 tasks)
Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.
Record operating data, such as the results of surveillance tests.
41CI 26–55 · exposure 53 · augmentation 63 · importance 4.5/5 · click for rater detail
Record operating data, such as the results of surveillance tests.
41| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Nuclear power is a heavily regulated, risk-averse, slow-moving sector with long operational cycles and limited digitization pressure; adoption of autonomous data-logging systems is lagging far behind other industries despite technical feasibility. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is a highly regulated, safety-critical, low-digitization-adoption sector where AI deployment for core operational recordkeeping remains rare and cautious. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can assist operators by auto-populating forms, flagging anomalies in test results, and organizing data for human review, substantially reducing manual logging burden while the operator maintains oversight and validates entries. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-assisted data capture, anomaly flagging, and auto-population of logs could meaningfully speed up documentation, though a licensed operator must still review and validate entries. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Recording operating data from surveillance tests is highly structured and repetitive—existing AI systems can reliably capture, parse, and log numeric/categorical data from displays, instruments, or digital systems with high accuracy. However, some contextual interpretation and manual verification may still be needed in edge cases, preventing a full 5. |
| Task automatability | claude-sonnet-5 | 3/5 | Recording standardized surveillance test results is largely structured data entry that AI/automation could handle, but integration with plant instrumentation and validation adds complexity beyond simple transcription. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Nuclear plants operate under stringent NRC regulations requiring documented chain of custody, human accountability, and specific sign-off procedures for safety-critical data; automated logging may not satisfy legal recording requirements without human operator verification and signature. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear operations are heavily regulated (NRC oversight), requiring licensed operators to verify and record data with strict chain-of-custody and accountability, making full automation legally and operationally restricted. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | The marginal cost of AI-based automated logging (software, integration, minimal oversight) is substantially lower than the operator labor time spent on manual data entry and transcription, though integration and regulatory compliance costs are non-trivial. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While automated logging tools could be cheap to run, the certification, validation, and integration costs in a nuclear regulatory environment keep all-in costs comparable to or above current human/operator-supported systems. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Data entry automation products and sensor-to-database systems exist and are deployed in some industrial settings, but nuclear power operations typically require human-validated logging and maintained audit trails; no mature off-the-shelf system is standard practice in U.S. nuclear plants today. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some industrial data logging and OCR/voice-to-text systems exist, but nuclear plants require validated, auditable systems and few if any deployed AI products handle this specific regulated logging task in production. |
Monitor all systems for normal running conditions, performing activities such as checking gauges to assess output or the effects of generator loading on other equipment.
21CI 18–25 · exposure 30 · augmentation 75 · importance 4.7/5 · click for rater detail
Monitor all systems for normal running conditions, performing activities such as checking gauges to assess output or the effects of generator loading on other equipment.
21| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | The nuclear power sector is highly conservative in automation adoption due to safety mandates, long licensing cycles, and regulatory scrutiny; AI-driven autonomous monitoring has not achieved significant production deployment in reactors despite decades of operation. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is a highly regulated, safety-critical, low-digitization-of-decision-authority sector where AI adoption for core operational control is minimal and slow due to regulatory conservatism. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-enhanced dashboards, predictive analytics for equipment wear, and anomaly detection can substantially assist operators in interpreting complex gauge data and detecting subtle system interactions, raising situational awareness and decision speed. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-based anomaly detection, predictive analytics, and automated alarm prioritization can meaningfully help operators notice trends and anomalies faster, improving situational awareness while humans retain control authority. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While gauge monitoring and data reading can be partially automated through sensors and dashboards, the task requires judgment about system interdependencies and anomaly detection in complex, safety-critical contexts where full autonomous operation is not trusted today. Current AI cannot reliably replace the human oversight responsibility in this safety-critical domain. |
| Task automatability | claude-sonnet-5 | 2/5 | Continuous monitoring of instrumentation can be partially supported by automated alarm and sensor systems, but final interpretation, cross-checking, and response authority must remain with a licensed operator, limiting true end-to-end automation today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear reactor operation is heavily licensed and regulated; a licensed nuclear operator must legally supervise all systems, and regulatory bodies require human sign-off on critical safety functions, making autonomous substitution legally prohibited. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear Regulatory Commission rules require licensed reactor operators to be present and responsible for monitoring reactor systems; this is a hard legal/regulatory barrier to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Implementing robust AI monitoring infrastructure with necessary redundancy, validation, and regulatory compliance is expensive; the cost of integration and ongoing oversight often exceeds the savings from reducing operator time on routine checks. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While sensor and SCADA systems are cheap to run, the need for licensed human oversight and redundant verification means the all-in cost of replacing the operator role with AI is not clearly cheaper than current staffing models. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Automated monitoring systems and sensor dashboards exist in production reactors, but they augment rather than replace human operators; AI-driven anomaly detection has limited real-world deployment at scale in nuclear contexts due to regulatory and validation constraints. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Plants already use extensive automated instrumentation and control systems, but these are decision-support tools rather than autonomous AI systems performing full operator monitoring duties in production. |
Note malfunctions of equipment, instruments, or controls and report these conditions to supervisors.
21CI 18–25 · exposure 30 · augmentation 63 · importance 4.5/5 · click for rater detail
Note malfunctions of equipment, instruments, or controls and report these conditions to supervisors.
21| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Nuclear power is a highly conservative, heavily regulated sector with slow technology adoption cycles. While some plants use digital monitoring systems, full AI-based malfunction detection remains largely in pilot phases, with strong organizational and regulatory resistance to replacing human operator surveillance. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative, heavily regulated, and slow to adopt new technology for safety-critical monitoring functions, with extensive certification cycles limiting AI deployment in control rooms. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered monitoring dashboards and anomaly alerts can significantly augment operator situational awareness and speed malfunction detection, allowing operators to focus on response rather than continuous scanning. Current systems in nuclear facilities demonstrate real augmentation value while keeping humans accountable for final judgment and reporting. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based anomaly detection and predictive analytics can help operators identify subtle patterns or flag potential issues earlier, providing decision support while the licensed human operator remains responsible for verification and reporting. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI systems can monitor sensor data and detect anomalies in real-time, the task requires judgment about safety significance, discrimination between malfunctions and instrument drift, and contextual understanding of reactor state. Current AI cannot reliably replace the operator's integrated assessment needed for safe reporting, though it could flag potential issues for human review. |
| Task automatability | claude-sonnet-5 | 2/5 | While AI can assist in flagging anomalies from sensor data, the actual observation, judgment, and reporting of malfunctions in a nuclear reactor context requires human presence, safety accountability, and integration with control room protocols that current systems can't fully replace. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear reactor operation is heavily regulated by the NRC and similar bodies; operators must hold licenses and be physically present at controls. Regulatory frameworks explicitly require licensed humans to monitor and report on reactor conditions, and liability for missed or misreported malfunctions creates high legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear reactor operations are among the most heavily regulated activities, requiring licensed operators (NRC-certified) to monitor and report equipment status; this is a hard legal and safety requirement that cannot be delegated to AI. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Implementing and maintaining AI monitoring systems requires significant capital investment, specialized nuclear domain expertise, and continuous validation against false positives. The fully-loaded cost per malfunction detection event likely exceeds the cost of a trained operator performing the task, especially given oversight and integration costs. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Sensor-based monitoring systems have some cost advantages, but the extensive validation, certification, and redundancy required for nuclear-grade monitoring plus mandatory human oversight keep all-in costs comparable to or higher than trained operators. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Anomaly detection systems exist in some nuclear facilities and can identify sensor deviations, but deployed systems typically generate high false-positive rates and require operator verification. No fully autonomous AI system currently performs end-to-end malfunction detection and reporting without human oversight in production nuclear plants. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Anomaly detection systems and predictive maintenance tools exist in industrial contexts, but no deployed AI product autonomously monitors and reports nuclear reactor malfunctions to supervisors in production nuclear plants today. |
Review and edit standard operating procedures.
15CI 13–18 · exposure 20 · augmentation 50 · importance 4.1/5 · click for rater detail
Review and edit standard operating procedures.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | The nuclear energy sector is highly regulated and conservative, with slow adoption of unproven automation in safety-critical tasks. Few if any operational nuclear plants are deploying AI agents for procedure review; pilots are minimal and adoption remains near zero. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is a highly regulated, safety-critical, low-digitization-adoption sector where AI tools are used cautiously if at all for procedural documentation, with virtually no production-scale AI adoption for this task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist operators by flagging inconsistencies, suggesting formatting improvements, or drafting procedural language from outlines, thereby saving some editorial time. However, the human expert must remain central to judgment and accountability, limiting the productivity multiplier to moderate assistance on drafting and formatting rather than transformative augmentation. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help draft, format, cross-reference, and flag inconsistencies in SOP language, offering useful support to operators/engineers who retain final review and sign-off responsibility. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can draft or suggest edits to procedural text, the safety-critical nature of nuclear operating procedures demands human expert judgment to verify accuracy, compliance with regulations, and operational feasibility. Current AI cannot reliably handle the domain-specific technical requirements and liability considerations independently, limiting time savings to minor editorial assistance rather than end-to-end automation. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can draft or summarize procedural text, but reviewing/editing SOPs for a nuclear reactor requires deep verification against plant-specific engineering, safety analyses, and regulatory codes that current systems cannot reliably validate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Extreme barriers exist: nuclear facilities operate under strict NRC (or equivalent international) regulatory frameworks that require licensed operators to take responsibility for procedure accuracy and safety compliance. Liability, legal sign-off requirements, and the human expert sign-off mandate for safety-critical procedures create hard legal and organizational barriers to unsupervised or minimally-supervised automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear procedures are governed by strict NRC (or equivalent) regulations requiring licensed, qualified personnel to review and approve SOPs, with severe liability for errors—an essentially hard regulatory barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted text generation and editing can reduce some labor costs, but the need for qualified nuclear operators to review, verify, and sign off on edits means overhead remains substantial. The cost of integrating AI tools and human oversight is not yet lower than traditional expert review. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI drafting/editing assistance is cheap, the mandatory human expert review, verification, and regulatory sign-off dominate the cost, keeping overall cost comparable to or only slightly below the human-only process. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed commercial AI product is authorized or actually used in production to independently review and edit nuclear reactor operating procedures. Such systems would require specialized domain knowledge, regulatory approval, and verified performance that current general-purpose or even specialized AI systems have not demonstrated at scale. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Generic LLM tools can assist with document editing and consistency checks, but no deployed product performs authoritative SOP review/approval in nuclear plants; this remains a human, credentialed process. |
Direct measurement of the intensity or types of radiation in work areas, equipment, or materials.
13CI 9–18 · exposure 20 · augmentation 50 · importance 3.7/5 · click for rater detail
Direct measurement of the intensity or types of radiation in work areas, equipment, or materials.
13| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power generation is a heavily regulated, capital-intensive sector with low digitization velocity for safety-critical operator tasks; regulatory inertia, safety certification requirements, and the specialized expertise needed create strong resistance to AI-driven automation of radiation measurement direction. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly regulated, safety-conservative, and slow to adopt automation for core safety functions, with minimal evidence of AI displacing human radiation monitoring roles. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI could assist operators by analyzing historical radiation patterns, flagging anomalies, or recommending measurement locations, improving situational awareness; however, the task's real-time, safety-critical nature limits the depth of augmentation relative to human-directed measurement. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI and data analytics can help interpret radiation trend data, flag anomalies, or optimize scheduling of surveys, offering moderate assistive value while humans remain responsible for direction and measurement. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Radiation measurement devices can automatically collect intensity data and record it, but directing measurement (selecting locations, adjusting for environmental factors, interpreting results in operational context) requires human judgment and physical positioning that current AI cannot perform autonomously in safety-critical nuclear settings. |
| Task automatability | claude-sonnet-5 | 2/5 | Radiation measurement involves handling specialized detection equipment and directing physical monitoring procedures in a safety-critical environment, which requires physical presence and regulatory-mandated human judgment that current AI cannot replace end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear Regulatory Commission regulations and industry safety standards require licensed operators to conduct and certify radiation measurements; liability for detection failures is severe, and direct human supervision of radiation monitoring is a hard regulatory requirement, not merely a preference. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear regulatory frameworks (e.g., NRC licensing) require certified, licensed operators to direct radiation safety measurements, making this a hard-barrier task with legal liability and safety-critical human sign-off requirements. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Automated radiation monitoring equipment carries high capital and maintenance costs; the loaded cost of a nuclear operator performing this task remains competitive with or cheaper than retrofitting autonomous measurement systems in safety-critical facilities where liability and redundancy requirements are stringent. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI system performing this task independently, so no meaningful cost comparison favors AI; human-operated detection equipment and judgment remain necessary. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While radiation detection instruments with automated logging exist, no deployed AI system independently directs measurement activities or interprets radiation patterns in nuclear reactor environments without human oversight; current systems support but do not replace the operator's directional and interpretive role. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product autonomously directs radiation surveys in nuclear plants; this remains a human-directed, instrument-based physical task with no commercial autonomous substitute in production. |
Dispatch orders or instructions to personnel through radiotelephone or intercommunication systems to coordinate auxiliary equipment operation.
9CI 0–18 · exposure 13 · augmentation 25 · importance 4.3/5 · click for rater detail
Dispatch orders or instructions to personnel through radiotelephone or intercommunication systems to coordinate auxiliary equipment operation.
9| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is a highly regulated, safety-critical sector with long operational lifespans and extreme conservatism around automation. Deployment of AI-driven dispatch in this domain is essentially non-existent and faces fundamental regulatory and institutional resistance. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is a highly regulated, safety-critical, low-digitization-for-control-functions industry with essentially no AI agent adoption for real-time operational dispatch. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could potentially assist by logging communications or suggesting routine message templates, but the core task—exercising authority over personnel to coordinate safety-critical equipment—depends on human presence and real-time judgment that AI augmentation alone cannot meaningfully enhance without creating new risks. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could potentially assist with logging, scheduling reminders, or drafting standard communications, but the live coordination and judgment calls central to this task see minimal AI augmentation today. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Dispatching orders to personnel requires real-time decision-making in a safety-critical environment with human judgment about personnel availability, equipment status, and contextual priorities. Current AI cannot reliably handle the dynamic coordination and interpersonal nuance required, nor can it legally or safely replace the human operator's direct authority in this safety-critical role. |
| Task automatability | claude-sonnet-5 | 2/5 | While generating routine dispatch messages is trivial, coordinating real-time auxiliary equipment operations in a nuclear plant requires situational awareness, judgment, and rapid response to changing conditions that current AI cannot reliably handle end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear reactor operations are among the most heavily regulated activities; dispatch authority is a licensed operator function, and regulatory frameworks (NRC, IAEA) require human accountability and sign-off on all safety-critical communications. Legal liability and mandatory human operator control create hard barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear reactor operation is one of the most heavily regulated professions, requiring NRC licensure, and control room communications/decisions must legally be made by certified human operators. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Even if AI could draft dispatch messages, integration into reactor control systems and regulatory oversight would be expensive and complex. The cost of human error in nuclear operations is extraordinarily high, making AI oversight itself costly and likely not economical compared to the experienced operator wage. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Given the near-total absence of viable AI systems for this safety-critical coordination task, there is no meaningful AI cost basis to compare against the human operator's wage. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate text or draft messages, no deployed product reliably replaces a reactor operator's real-time dispatch decisions. AI might assist with message formatting or logging, but actual personnel coordination demands human judgment that current systems cannot demonstrate reliably in production nuclear facilities. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed products autonomously dispatch operational instructions to nuclear plant personnel; this remains firmly in the domain of licensed human operators with no commercial AI substitute in production. |
Monitor or operate boilers, turbines, wells, or auxiliary power plant equipment.
7CI 0–14 · exposure 13 · augmentation 50 · importance 4.7/5 · click for rater detail
Monitor or operate boilers, turbines, wells, or auxiliary power plant equipment.
7| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is a heavily regulated, conservative sector with minimal AI adoption for core operational tasks. Safety and compliance culture, combined with strict operator licensing, result in very slow experimental adoption of AI-driven autonomy. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is an extremely conservative, safety-critical, heavily regulated sector with minimal AI deployment in control-room operations to date. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist operators through real-time monitoring dashboards, anomaly detection, predictive maintenance alerts, and decision support—improving situational awareness and reducing cognitive load while the human operator remains in control. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based anomaly detection, predictive maintenance, and decision-support tools can assist operators in monitoring equipment trends, though the core monitoring and operation remains manual. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can monitor sensor data and flag anomalies, the task requires real-time operational control decisions and immediate response to complex equipment failures. Current systems lack the safety-critical reliability and human judgment needed for end-to-end autonomous operation of nuclear plant equipment. |
| Task automatability | claude-sonnet-5 | 1/5 | Physical monitoring and hands-on operation of reactor equipment requires embodied presence, sensor interpretation, and manual control actions that current AI cannot perform end-to-end without robotics not yet deployed in this domain. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear reactor operation is subject to strict NRC licensing and regulatory requirements that mandate trained, certified human operators physically present and in control. Legal, safety, and liability frameworks create hard barriers to full automation or unsupervised AI control. |
| Adoption barriers | claude-sonnet-5 | 5/5 | NRC licensing mandates certified human reactor operators be present and in control at all times, making this one of the most heavily regulated human-authorization requirements in any occupation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of AI systems, integration, redundant hardware, validation, and continuous human oversight in nuclear plants exceeds the wage cost of reactor operators. High liability and regulatory validation costs make AI economically uncompetitive. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this task at all, so cost comparison favors the human operator by default; any AI-assisted monitoring adds cost on top of required human staffing. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Monitoring and alerting systems exist in nuclear plants, but no deployed AI system reliably operates boilers, turbines, or auxiliary equipment autonomously. Existing supervisory systems remain human-supervised; autonomous control in production is not standard practice. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product operates nuclear power plant equipment autonomously; control rooms remain fully human-operated with software only providing monitoring dashboards and alarms. |
Respond to system or unit abnormalities, diagnosing the cause, and recommending or taking corrective action.
4CI 0–9 · exposure 8 · augmentation 50 · importance 4.7/5 · click for rater detail
Respond to system or unit abnormalities, diagnosing the cause, and recommending or taking corrective action.
4| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power plant operations remain highly regulated with strong safety and human-oversight cultures. Adoption of autonomous AI for core safety diagnostics is negligible; plants continue to invest in operator training and decision-support rather than replacement systems. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative, safety-regulated, and slow to adopt novel automation for control-room decision-making, with minimal evidence of production AI deployment in this specific task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist operators by flagging anomalies, suggesting likely diagnoses based on historical patterns, and summarizing system states in real time, meaningfully aiding human decision-making without replacing operator judgment or legal responsibility. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based anomaly detection, diagnostic support systems, and predictive analytics can assist operators by flagging abnormal conditions and suggesting likely causes, improving situational awareness without replacing operator judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with pattern recognition in system data and anomaly detection, responding to reactor abnormalities requires real-time diagnostic decision-making under safety-critical conditions with legal accountability. Current AI cannot reliably diagnose causes and recommend corrective actions without human oversight, and regulatory requirements mandate human operator authority and judgment. |
| Task automatability | claude-sonnet-5 | 1/5 | Diagnosing and correcting abnormalities in a nuclear reactor requires real-time physical situational awareness, safety-critical judgment, and accountability that current AI systems cannot perform end-to-end without human control at every step. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Federal law and NRC regulations mandate that licensed human reactor operators must directly supervise safety systems and maintain legal responsibility for abnormality response. Automation of this task faces hard legal and liability barriers that prevent substitution without fundamental regulatory change. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear reactor operation is one of the most heavily regulated professions, requiring NRC licensing, mandatory human oversight, and legal liability that make full automation of this task essentially prohibited. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The integration, validation, cybersecurity, liability coverage, and continuous oversight required for AI in a nuclear safety-critical context would exceed the cost of human operator labor, particularly given minimal existing automation infrastructure in legacy reactor systems. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Any AI system would need to run alongside, not instead of, licensed human operators due to regulatory mandates, so it adds cost rather than substituting for the wage of the operator. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system reliably operates as the primary responder to nuclear reactor abnormalities in production. Diagnostic aids exist in research and pilot stages, but no commercial product demonstrates independent, verified capability to diagnose and recommend corrective actions meeting nuclear regulatory standards. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product autonomously diagnoses and corrects nuclear plant abnormalities in production; decision support tools exist but final diagnosis and action remain fully human-controlled and regulator-mandated. |
Implement operational procedures, such as those controlling start-up or shut-down activities.
3CI 0–5 · exposure 5 · augmentation 38 · importance 4.5/5 · click for rater detail
Implement operational procedures, such as those controlling start-up or shut-down activities.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear energy is a highly regulated, safety-first industry with extremely slow technology adoption cycles. No significant production deployment of autonomous AI reactor control has occurred, and organizational and regulatory friction strongly discourage rapid substitution. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative and slow-moving regarding control-room automation due to safety regulation and physical infrastructure constraints. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist operators by monitoring system parameters in real-time, flagging anomalies, and providing decision support, but the operator must remain in active control and responsible for all procedural steps. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with monitoring, anomaly detection, or procedure documentation support, but the core execution of start-up/shutdown sequences remains a manual, tightly controlled human task with limited AI augmentation in the control room itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Nuclear reactor operations require real-time human judgment, safety verification, and legal sign-off at every critical step. Current AI cannot autonomously execute start-up/shut-down procedures that involve reactor physics, emergency response decision-making, and compliance with strict regulatory protocols. |
| Task automatability | claude-sonnet-5 | 1/5 | Executing start-up/shutdown procedures on a nuclear reactor requires certified human operators making real-time safety-critical decisions in a physical control room; no AI system today performs this end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear Regulatory Commission (NRC) licenses require a certified operator to physically perform and legally certify all start-up and shut-down activities. Strong regulatory barriers, liability asymmetry, and mandatory human sign-off create hard constraints on automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | NRC licensing mandates certified human reactor operators for these exact procedures, with strict regulatory, legal, and safety oversight preventing any AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI systems for nuclear operations require extensive validation, regulatory approval, and ongoing human oversight that add significant cost, making them more expensive than the human operator they would theoretically replace. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute cost to compare; any automation would require extensive certified redundant safety systems, making replacement far more costly than retaining licensed operators. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with procedure documentation and checklist generation, no deployed system reliably performs independent start-up or shut-down execution for nuclear reactors. The safety-critical nature and liability exposure mean even research systems are not tested in production environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed commercial product autonomously implements nuclear reactor start-up/shutdown; this remains firmly in human-operated, heavily regulated control rooms. |
Conduct inspections or operations outside of control rooms as necessary.
2CI 0–4 · exposure 0 · augmentation 25 · importance 3.9/5 · click for rater detail
Conduct inspections or operations outside of control rooms as necessary.
2| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is a highly conservative, heavily regulated sector with slow technology adoption. Physical plant operations remain almost entirely human-dependent despite decades of automation advances elsewhere. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative, safety-regulated, and slow to adopt automation for physical safety-critical tasks, with minimal evidence of AI displacing human inspectors in this domain. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could support inspections through predictive analytics or remote monitoring dashboards, but the core task of physical presence and hands-on assessment cannot be meaningfully augmented by current AI systems. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Some sensor-based monitoring, predictive maintenance analytics, or drone-assisted imaging could support inspection planning, but the physical inspection itself sees limited AI augmentation currently. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Physical inspection and operation outside control rooms requires presence in hazardous environments, manipulation of equipment in variable conditions, and real-time environmental assessment. Current AI systems cannot autonomously navigate nuclear facilities, perform hands-on equipment checks, or respond to unexpected physical contingencies. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physically walking through the plant, visually inspecting equipment, and making real-time judgment calls in a hazardous industrial environment—far beyond current AI or robotic capabilities to perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear facilities are heavily regulated; operators must be licensed, and safety-critical inspections require trained humans to take legal and safety responsibility. Regulatory frameworks (NRC, IAEA) mandate human presence and decision-making for field operations. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear operations are heavily regulated by bodies like the NRC, requiring licensed operators for many safety-critical tasks, and liability for missed inspection issues is extremely high, creating strong legal and safety barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Specialized robotic systems for nuclear inspection are expensive to develop, deploy, and maintain, while human operators are relatively cost-effective. The all-in cost of robotic systems would likely exceed the loaded human wage. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Deploying robotics or AI systems capable of navigating a nuclear facility, interpreting equipment conditions, and ensuring safety compliance would be far more costly than a trained human operator performing routine walkthroughs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can reliably conduct independent physical inspections in nuclear facilities. Robotics for nuclear environments exist in research contexts but lack the flexibility, judgment, and certification to operate autonomously in unstructured plant environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously conducts physical nuclear plant inspections outside control rooms; robotics in nuclear settings remain research-stage or narrowly used for radiation surveys, not general inspection duties. |
Operate nuclear power reactors in accordance with policies and procedures to protect workers from radiation and to ensure environmental safety.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.9/5 · click for rater detail
Operate nuclear power reactors in accordance with policies and procedures to protect workers from radiation and to ensure environmental safety.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is a highly regulated, risk-averse sector with strict operator licensing and safety requirements. Adoption of AI-driven automation for core reactor operations is effectively zero, and regulatory culture actively resists autonomous control in favor of human oversight. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative, safety-regulated, and slow to adopt automation for core control functions, with no signs of AI displacing licensed operators. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | While AI could assist with data analysis, anomaly detection, and procedure-lookup, the core task of operating the reactor requires human judgment, regulatory sign-off, and accountability. Augmentation potential is limited because the human operator must remain fully responsible and in active control. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with monitoring, anomaly detection, predictive maintenance, and decision support tools, but the operator remains fully in control of reactor operation decisions. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Operating nuclear reactors requires real-time monitoring, dynamic decision-making in complex, safety-critical environments, and immediate response to anomalies. Current AI systems cannot reliably handle the full scope of reactor operation, emergency protocols, and the irreversible consequences of errors, making autonomous end-to-end automation infeasible. |
| Task automatability | claude-sonnet-5 | 1/5 | Real-time control of a nuclear reactor requires physical presence, licensed judgment, and continuous safety-critical decision-making that current AI cannot perform end-to-end even with significant setup. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Extremely high barriers exist: nuclear operators must be licensed by the NRC (or equivalent international regulators), legal liability rests on the licensed human operator, and regulations explicitly require trained human supervision of reactor operation. Automation is legally and structurally prohibited. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear reactor operation is one of the most heavily regulated tasks in existence, requiring NRC licensure, extensive training, and legal accountability that cannot be delegated to AI. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Nuclear reactor operators earn substantial salaries ($80k–$100k+) with specialized training. The cost of AI systems capable of handling safety-critical monitoring, plus extensive verification, testing, and regulatory approval, would far exceed labor replacement savings. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this task, so cost comparison is moot; any partial AI tooling would add cost on top of required human oversight, not replace it. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product autonomously operates nuclear reactors in production. Regulatory frameworks (NRC, IAEA) mandate human operators with licenses, and the safety-critical nature precludes reliance on current AI systems for core operational control. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product operates a nuclear reactor; this remains firmly in the domain of licensed human operators with no production-grade autonomous alternative. |
Adjust controls to position rod and to regulate flux level, reactor period, coolant temperature, or rate of power flow, following standard procedures.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.9/5 · click for rater detail
Adjust controls to position rod and to regulate flux level, reactor period, coolant temperature, or rate of power flow, following standard procedures.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power plants operate in a heavily regulated, safety-critical environment where adoption of AI automation is extremely slow; the sector has not moved toward AI-driven control systems due to regulatory, liability, and safety concerns. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative, heavily regulated, and slow to adopt automation for core safety-control functions, with control room modernization proceeding cautiously over decades. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | While AI could assist with monitoring and alerting (e.g., flagging anomalies), the core task of adjusting controls requires human judgment and legal accountability; assistance is limited to non-critical decision support rather than transformative productivity gains. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based monitoring, predictive analytics, and decision-support tools can help operators interpret sensor data and anticipate parameter changes, but the actual control adjustments remain manual and procedure-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Nuclear reactor operation requires real-time physical intervention, continuous monitoring of safety-critical parameters, and compliance with strict regulations that mandate human decision-making and accountability; current AI cannot perform these control adjustments end-to-end with the required reliability and legal authority. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a real-time, safety-critical physical control task inside a licensed nuclear facility requiring certified human operators; current AI cannot legally or reliably perform end-to-end control actions on reactor systems. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear Regulatory Commission (NRC) regulations explicitly require a licensed Senior Reactor Operator or Reactor Operator to directly control the reactor; this is a hard legal barrier that prevents automation or substitution of human decision-making authority. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear Regulatory Commission rules mandate licensed human operators for reactor control actions, with strict human-in-the-loop and human-authority requirements plus catastrophic liability exposure for errors. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The licensing, training, and salary of a nuclear reactor operator are substantial, but the cost of implementing and maintaining AI systems with the required safety certifications, redundancy, and regulatory compliance would exceed the cost of retaining human operators. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Even if AI assistance existed, the cost of certifying, validating, and insuring an autonomous control system for nuclear safety would vastly exceed current licensed operator wages given extreme liability and redundancy requirements. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system operates nuclear reactor controls independently; reactor operation is heavily regulated and requires licensed human operators with legal responsibility, making independent AI deployment infeasible in production. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed commercial AI product autonomously adjusts control rods or regulates reactor parameters in operating power plants; this remains research-stage (e.g., advisory/monitoring systems only). |
Develop or implement actions such as lockouts, tagouts, or clearances to allow equipment to be safely repaired.
0CI 0–0 · exposure 0 · augmentation 25 · importance 4.8/5 · click for rater detail
Develop or implement actions such as lockouts, tagouts, or clearances to allow equipment to be safely repaired.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is the most heavily regulated industrial sector with the slowest automation adoption in safety-critical procedures. Regulatory conservatism and the catastrophic cost of errors make adoption of autonomous safety systems virtually non-existent. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is a highly regulated, safety-critical, low-digitization-for-physical-tasks sector with minimal AI adoption for hands-on safety procedures. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by logging clearance checklists or monitoring equipment status displays, but the core task—physically executing and certifying lockouts—remains a human responsibility. Assistance here is minimal and cannot substitute for operator judgment. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could help track clearance documentation, checklists, or scheduling, but offers minimal assistance to the core physical and procedural safety act itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical intervention (applying lockouts/tagouts to equipment), real-time safety judgment in complex environments, and direct responsibility for preventing catastrophic failures. Current AI systems cannot perform the physical actions or make the safety-critical decisions autonomously today. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical presence, hands-on isolation of equipment, and safety-critical judgment in a highly regulated nuclear environment that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear safety regulations (10 CFR, NRC rules) legally require a licensed operator to perform and certify lockout/tagout and equipment clearance procedures. This is a hard regulatory and liability barrier that prevents substitution regardless of technical capability. |
| Adoption barriers | claude-sonnet-5 | 5/5 | NRC regulations, plant licensing requirements, and safety procedures mandate certified human operators to perform and verify lockout/tagout and clearance actions, with severe liability for errors. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of deploying AI to physically execute lockouts and oversee safety clearances would exceed the loaded wage of a trained reactor operator, and liability costs would be prohibitive if errors occurred during the most safety-critical work in the industry. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI has no viable substitute pathway here, so any AI cost is irrelevant against the human-performed, regulation-mandated procedure. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product independently executes lockout/tagout procedures or safety clearances in nuclear facilities. These require licensed operators to physically verify and sign off on safety-critical steps; no production system has demonstrable capability to do this reliably or legally. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product performs or authorizes lockout/tagout procedures in nuclear plants; this remains a fully human, procedurally governed task. |
Direct reactor operators in emergency situations, in accordance with emergency operating procedures.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.8/5 · click for rater detail
Direct reactor operators in emergency situations, in accordance with emergency operating procedures.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is one of the most heavily regulated, safety-conscious sectors; adoption of AI for safety-critical human direction would move at regulatory speed, which is glacial. No utilities are piloting or deploying AI operators for emergency command. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is among the most heavily regulated, safety-conservative, low-digitization sectors, with essentially no movement toward AI-directed emergency operations. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could provide decision support (scenario modeling, procedure lookup, communication logging) to assist a human director, but the core task—directing people in real time with accountability—remains human. The augmentation is narrow and supplementary. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based decision support and monitoring tools can help operators track plant status and flag procedural steps during emergencies, but authority and final direction remain fully human. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Directing human operators during emergencies requires real-time judgment, authority delegation, situational assessment under extreme uncertainty, and accountability that current AI systems cannot assume. Emergency response involves novel scenarios, conflicting priorities, and safety-critical decisions where human expertise and legal responsibility remain essential. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires real-time, high-stakes judgment under emergency conditions with legal accountability; no current AI system can direct human operators in a nuclear emergency end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task is hardcoded into nuclear regulations: the shift supervisor or control room director must be a licensed human who legally bears responsibility for emergency direction and decisions. Regulatory frameworks (10 CFR 26, NRC rules) explicitly require human authority in the control room during emergencies. |
| Adoption barriers | claude-sonnet-5 | 5/5 | NRC licensing requires certified Senior Reactor Operators to direct emergency response; this is a hard legal and safety-critical barrier that cannot be delegated to AI. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Even if technically feasible, the cost of AI oversight, validation, and liability insurance would far exceed the loaded wage of a shift supervisor. The stakes are too high to lower costs through automation. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Given the catastrophic liability exposure and lack of any viable AI substitute, cost comparison is moot—human oversight is mandatory and AI cannot substitute regardless of cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product can reliably direct nuclear reactor operators during emergencies or act as a substitute for the licensed shift supervisor/control room director. This task requires legal authority, regulatory compliance, and real-time human command presence that AI has not demonstrated in production nuclear environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs this function in production nuclear plants; this remains firmly in the domain of certified human supervisors following strict regulatory procedures. |
Participate in nuclear fuel element handling activities, such as preparation, transfer, loading, or unloading.
0CI 0–0 · exposure 0 · augmentation 25 · importance 4.4/5 · click for rater detail
Participate in nuclear fuel element handling activities, such as preparation, transfer, loading, or unloading.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is a heavily regulated, risk-averse sector with long equipment lifecycles and strict certification requirements; automation of critical fuel handling has seen minimal real-world adoption despite decades of regulatory stability. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative and heavily regulated, with minimal AI/robotic adoption in core safety-critical physical operations like fuel handling. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist with monitoring fuel condition data or predicting handling logistics, but the direct physical handling task requires continuous human judgment and hands-on control, limiting meaningful augmentation scope. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with monitoring, procedural checklists, or predictive diagnostics around fuel handling logistics, but it offers little direct assistance to the physical handling task itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Nuclear fuel element handling requires direct physical manipulation in heavily shielded, radioactive environments with strict safety protocols and real-time human judgment about positioning and containment. Current AI systems cannot perform end-to-end physical manipulation in such constrained, high-consequence settings. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a hands-on physical task involving handling of hazardous nuclear materials with specialized equipment; no current AI system can perform the physical manipulation, and it requires certified human presence and judgment throughout. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear fuel handling is heavily regulated under NRC and international atomic energy authorities; operators must be licensed, and liability for mishandling is severe. Regulatory frameworks explicitly require human expertise and sign-off for fuel element operations. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear fuel handling is strictly regulated by the NRC and equivalent bodies, requiring licensed operators, chain-of-custody documentation, and human accountability for safety-critical actions with severe liability exposure. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Purpose-built robotic systems for nuclear fuel handling are extremely expensive to develop, certify, and maintain, and would require substantial capital investment compared to the wages of trained operators performing the task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so any AI cost comparison is moot; specialized robotics for this purpose are far more costly than human operators given low volume and extreme precision/safety requirements. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI or robotic system reliably performs unsupervised nuclear fuel handling in production reactors today. Specialized robotics exist in research contexts, but nuclear power plants continue to rely on trained human operators for these critical handling tasks. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI or robotic product performs nuclear fuel handling autonomously in production; this remains a highly regulated, manual/semi-automated process controlled by licensed operators. |
Authorize maintenance activities on units or changes in equipment or system operational status.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.3/5 · click for rater detail
Authorize maintenance activities on units or changes in equipment or system operational status.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | No adoption of AI for independent authorization is occurring or permitted, as this task is subject to strict federal nuclear safety regulation. Adoption velocity remains zero by design. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly regulated, safety-conservative, and slow to adopt automation for control-room authority functions, with minimal AI agent deployment in production control decisions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could potentially assist by surfacing compliance checklists, historical maintenance records, or safety considerations for the human operator to review before authorizing, but the authorization decision itself remains entirely human and the assistance value is limited. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist by analyzing maintenance logs, predicting equipment issues, or summarizing operational data to inform the operator's decision, though the authorization itself remains a human function. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires formal legal authorization and accountability for critical safety systems in a federally regulated industry. No AI system can or should independently authorize maintenance or operational changes without explicit human responsibility and licensure. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires legally authorized human judgment and accountability for safety-critical decisions; no AI system today can perform this end-to-end with equal quality and time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task is protected by hard legal barriers: the NRC requires licensed reactor operators to authorize such changes, liability falls on the license holder, and regulatory frameworks explicitly mandate human oversight and sign-off on all maintenance and operational status changes. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear reactor operations are governed by strict regulatory licensing (e.g., NRC requirements) mandating that only certified human operators authorize such changes, creating a hard legal barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI cannot perform this task at all; the cost comparison is meaningless. The task requires a licensed human operator whose presence and decision-making are non-substitutable regulatory requirements. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Given licensing, liability, and safety requirements, AI cannot substitute for the human authorizer, so there is no viable cost comparison—the human cost is mandatory regardless of AI capability. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product performs independent authorization of nuclear reactor maintenance or equipment changes. This remains entirely a human-performed, legally-mandated task requiring a licensed reactor operator to sign off. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product authorizes maintenance or operational status changes at nuclear reactors; this remains firmly in the domain of licensed human operators under regulatory oversight. |
Supervise technicians' work activities to ensure that equipment is operated in accordance with policies and procedures that protect workers from radiation and ensure environmental safety.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.3/5 · click for rater detail
Supervise technicians' work activities to ensure that equipment is operated in accordance with policies and procedures that protect workers from radiation and ensure environmental safety.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is among the most heavily regulated and conservative sectors; deployment of autonomous or AI-led supervision of reactor operations is prohibited by law and industry practice. Adoption velocity for AI replacement is effectively zero. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is a highly regulated, safety-critical, low-digitization-of-decision-authority sector with minimal AI adoption for supervisory/safety roles. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could provide real-time monitoring dashboards, anomaly detection on equipment readings, and procedure compliance alerts to assist a human supervisor, but the core supervisory authority and decision-making must remain with the licensed operator. Augmentation potential is limited by the requirement that humans retain full accountability. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with monitoring dashboards, anomaly detection, and compliance documentation, but the core supervisory judgment and accountability remain human-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Supervising technicians and ensuring regulatory compliance with radiation safety requires real-time human judgment, situational awareness, and accountability that current AI cannot reliably handle. AI cannot take legal responsibility for safety-critical decisions or adapt to novel failure modes in a nuclear environment. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical presence, real-time judgment, and legal accountability for radiation safety supervision that current AI cannot perform end-to-end or independently. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear Regulatory Commission (NRC) regulations explicitly require licensed reactor operators to supervise operations and maintain direct responsibility for safety. Only a certified human operator can legally hold this position and make binding decisions regarding radiation protection and environmental safety. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear reactor operation is one of the most heavily regulated fields, requiring NRC-licensed personnel for exactly this kind of safety supervision, making substitution legally prohibited. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | A nuclear power reactor operator with full qualifications and licensing costs tens of thousands annually in salary and benefits. AI supervision systems do not exist at production scale, and any partial monitoring system would require substantial human oversight, making the total cost-of-ownership comparable to or higher than the human. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot substitute for the licensed human role here, so there is no viable cost comparison; the human is mandatory regardless of AI cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can autonomously supervise nuclear reactor operations or make binding safety decisions. This task fundamentally requires a licensed, qualified human operator with legal authority and situational context that no current product reliably provides. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product supervises nuclear technicians or bears responsibility for radiation/environmental safety compliance in production settings today. |
Authorize actions to correct identified operational inefficiencies or hazards so that operating efficiency is maximized and potential environmental issues are minimized.
0CI 0–0 · exposure 0 · augmentation 50 · importance 4.1/5 · click for rater detail
Authorize actions to correct identified operational inefficiencies or hazards so that operating efficiency is maximized and potential environmental issues are minimized.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is one of the most heavily regulated and conservative sectors; adoption of AI for autonomous authorization is virtually non-existent and faces fundamental legal and safety prohibitions. Velocity is near-zero. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | The nuclear power sector is highly conservative and slow to adopt autonomous decision-making systems due to safety regulation and catastrophic error costs. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by providing rapid diagnostics, anomaly detection, and analytical support for inefficiencies and hazards, helping operators make faster, better-informed decisions. However, the final authorization remains the operator's responsibility, so augmentation is moderate rather than transformative. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based monitoring and predictive analytics can help operators identify inefficiencies or hazards faster, improving decision support even though final authorization remains human-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires human judgment, safety accountability, and real-time decision-making authority in a tightly regulated environment. Current AI systems cannot legally or reliably authorize actions in nuclear operations—they may assist analysis, but the authorization itself is a human responsibility that cannot be automated end-to-end today. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires real-time authorization of safety-critical corrective actions in a highly regulated nuclear environment; current AI cannot legally or practically perform end-to-end authorization of such actions. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task is protected by hard regulatory barriers: nuclear operators must be licensed, trained, and legally responsible for authorizing corrective actions. Regulatory agencies explicitly require human operators in the loop for authorization decisions; automation is not permitted. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear operations are governed by strict regulatory frameworks (e.g., NRC licensing) requiring certified human operators to authorize safety and efficiency actions, creating a hard legal barrier to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The task is fundamentally about human sign-off and accountability; there is no cost-replacing alternative because the legal and safety liability cannot be transferred to an AI system. The human operator cost remains irreplaceable. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Given licensing, liability, and safety requirements, any AI system would require extensive human oversight and validation, making it more costly than human authorization in practice today. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system independently authorizes actions in nuclear power plants; regulatory frameworks (NRC, international) mandate human operators hold explicit authority and accountability for operational decisions. AI tools for monitoring exist, but authorization remains human-performed. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI products autonomously authorize corrective operational actions in nuclear plants; decision-support tools exist but authorization remains strictly human-controlled. |
Direct the collection and testing of air, water, gas, or solid samples to determine radioactivity levels or to ensure appropriate radioactive containment.
0CI 0–0 · exposure 0 · augmentation 38 · importance 3.9/5 · click for rater detail
Direct the collection and testing of air, water, gas, or solid samples to determine radioactivity levels or to ensure appropriate radioactive containment.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Nuclear power is a highly regulated, conservative industry with strong safety-first culture and slow technology adoption cycles. No meaningful adoption of autonomous sample collection AI is occurring in production nuclear facilities. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Nuclear power is among the most heavily regulated, safety-critical, low-digitization-for-autonomy sectors, with minimal AI agent adoption for core safety-directing tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by analyzing historical test data or flagging anomalies in results after human collection and testing, but the core sampling and testing activities require direct human involvement and judgment, limiting augmentation potential. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with data logging, trend analysis of radioactivity readings, and flagging anomalies, but the directive and judgment-based coordination of sampling remains human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical sample collection at specific locations, hands-on testing apparatus operation, and real-time judgment about containment integrity—all of which demand human presence and manual dexterity in a hazardous environment. Current AI systems cannot independently collect physical samples or operate lab equipment. |
| Task automatability | claude-sonnet-5 | 1/5 | Directing physical sample collection and radioactivity testing in a nuclear plant requires hands-on coordination, safety judgment, and physical presence that current AI cannot replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Nuclear facilities operate under strict NRC and DOE regulations requiring licensed operators to directly supervise and authorize sample collection and testing. Liability for radioactive containment failures is severe, and human certification and accountability are legally mandated. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Nuclear plant operations are governed by NRC licensing requiring certified human reactor operators to directly oversee radiological safety and containment procedures, making substitution legally prohibited. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI systems cannot replace the physical and manual components of sample collection and testing. The cost of integrating automated sampling equipment, combined with regulatory compliance and validation, would far exceed the loaded wage of a trained reactor operator performing this task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this directive/oversight role, so cost comparison favors the human operator by default since AI cannot deliver the output. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | While AI could assist with data analysis of test results, no deployed system performs autonomous sample collection and testing in nuclear facilities. The task fundamentally requires physical presence and human oversight in a highly regulated, safety-critical setting. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product directs sampling operations or radioactive containment verification in nuclear plants; this remains firmly a human operator function under strict regulatory oversight. |
Related occupations — Production
How to read this
A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.
What would change this score
New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.