Subway and Streetcar Operators
53-4041.00Median wage $86,380/yr10,200 employed (US)Rank #272 of 923 scored · top 29% by substitution
Operate subway or elevated suburban trains with no separate locomotive, or electric-powered streetcar, to transport passengers. May handle fares.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution35
Exposure37
Augmentation45
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
10 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
10%
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%36
panel mean rating 2.4/5 → substitution pressure 36/100
Technical feasibility todayw 20%38
panel mean rating 2.5/5 → substitution pressure 38/100
Cost vs. human wagew 15%42
panel mean rating 2.7/5 → substitution pressure 42/100
Adoption barriersw 20%inverted — strong barriers lower the score28
panel mean rating 3.9/5 (barrier strength) → substitution pressure 28/100
Sector adoption velocityw 10%26
panel mean rating 2.1/5 → substitution pressure 26/100
Task breakdown (10 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.
Complete reports, including shift summaries and incident or accident reports.
72CI 60–85 · exposure 78 · augmentation 75 · importance 4.3/5 · click for rater detail
Complete reports, including shift summaries and incident or accident reports.
72| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Transit agencies and large public-sector organizations show moderate adoption of automation, with pilots common but full production deployment of report generation less widespread than in private tech-forward sectors. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public transit is a traditionally slow-adopting, unionized, safety-regulated sector with limited AI deployment for operator-facing administrative tasks compared to information/finance sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist operators by auto-populating templates, drafting incident narratives from notes, and organizing shift data, freeing operators to focus on accuracy review and exceptions rather than data entry. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can substantially speed up report writing via dictation, auto-fill, and summarization, letting operators quickly produce clear, complete reports while retaining responsibility for accuracy and submission. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Report completion from structured incident data is highly automatable; AI can generate shift summaries and incident documentation from logs, sensor data, and incident records with minimal human input, easily achieving 50%+ time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 4/5 | Report writing from structured incident data (times, locations, categorical details) is well within current LLM capabilities, especially with templated forms and voice-to-text input.dictating notes and having AI structure them into reports would save significant time.n Full automation requires the operator to still supply raw facts, so it's not a pure end-to-end replacement but the drafting/formatting portion is highly automatable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal regulatory or authorization barriers exist for automating report writing itself, though transit agencies may have internal policies requiring operator sign-off; no legal requirement mandates human authorship of these administrative documents. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Incident and accident reports often carry legal/safety significance and may require the operator's personal certification or signature, creating some human-accountability barrier, though the drafting itself isn't restricted by licensing. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven report generation costs pennies per report versus tens of dollars in operator labor time; inference and integration costs are negligible compared to the loaded wage of a transit operator completing these administrative tasks. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated transcription and report drafting tools cost fractions of a cent to dollars per report versus the operator's or a clerk's time, making AI substantially cheaper for the drafting component. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed document generation and report automation systems exist in transportation and logistics sectors; while some customization and human review is typically required, mature products reliably handle structured incident and shift reporting in production environments. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Enterprise report-generation and transcription tools exist and are used in transit and logistics operations, but integration with transit-specific incident reporting systems and safety-critical documentation is not yet standard or fully validated in this niche. |
Make announcements to passengers, such as notifications of upcoming stops or schedule delays.
64CI 45–84 · exposure 59 · augmentation 63 · importance 4.5/5 · click for rater detail
Make announcements to passengers, such as notifications of upcoming stops or schedule delays.
64| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Many transit systems have deployed audio announcement automation for routine stops, but human operators remain present and responsible for the full communication function. Adoption is steady but not rapid—pilots and partial automation are common, but full replacement of human announcement duties is rare in major transit networks. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Public transit systems have broadly adopted automated stop announcements over the past two decades, driven partly by ADA/accessibility regulations, making this one of the more mature automation deployments in the sector. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI announcements can assist operators by automatically handling routine stops and standard delay notifications, freeing the operator to focus on passenger safety, accessibility, and non-scripted communication. This augmentation is already visible in digitized transit systems where operators confirm or override automated announcements. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Where live announcements remain necessary (e.g., unplanned delays, emergencies), automated systems can draft or trigger standard phrasing, assisting the operator, though most routine cases are already fully automated rather than augmented. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Passenger announcements are repetitive and could be pre-recorded or AI-synthesized for routine stops and standard delays, but real-time judgment about non-standard situations, crowded conditions, and passenger safety requires human oversight. Current AI cannot reliably handle the full range of scenarios (emergencies, communication breakdowns, unexpected delays) with equal quality to a trained operator. |
| Task automatability | claude-sonnet-5 | 4/5 | Automated announcement systems (pre-recorded or TTS-triggered by GPS/schedule data) already handle routine stop and delay notifications on many transit systems with high reliability.dopting equal or better quality than live human announcements. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Transit agencies face regulatory expectations around safety announcements and passenger communication, and many agencies and riders prefer human operators for emergencies or complex situations. However, there is no explicit legal requirement that a human must make announcements, creating moderate friction rather than a hard barrier to automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement ties this specific sub-task to a human; some agencies mandate live announcements for emergencies or accessibility exceptions, but routine automation is already widely accepted. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Pre-recorded or AI-synthesized announcements have minimal marginal cost once deployed (fractions of a cent per announcement), whereas a human operator costs $40–70k+ annually in loaded wages. The all-in cost of an AI announcement system is substantially lower, even accounting for integration and maintenance. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Once installed, automated announcement systems cost negligible marginal amounts per trip compared to any operator time allocated to manual announcements. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Text-to-speech and audio announcement systems exist in production transit systems, but they typically handle only pre-scripted, standardized announcements. AI systems struggle with dynamic, context-sensitive announcements (e.g., responding to unexpected passenger behavior or equipment failures), and most deployed systems still rely on human operators for critical or non-routine communications. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Automated PA systems are standard equipment on most modern subway and streetcar fleets worldwide, reliably announcing stops and integrating with schedule/delay data in production. |
Operate controls to open and close transit vehicle doors.
54CI 25–83 · exposure 62 · augmentation 25 · importance 4.8/5 · click for rater detail
Operate controls to open and close transit vehicle doors.
54| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Major urban transit systems are actively adopting driverless and automated door systems; deployment is ongoing in high-digitization sectors (public transit in developed cities), though full-system adoption varies by jurisdiction and infrastructure age. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public transit is a slow-moving, heavily unionized, capital-constrained sector where full automation is limited to a handful of new-build driverless lines globally, not a general retrofit trend. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Door operation is a discrete control task with limited room for AI to assist a human operator in a meaningful way; the task is either automated or manual, with little intermediate augmentation value. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Sensor-based door obstruction detection and alerts can assist operators, but this is a narrow assistive function rather than a broad productivity transformation of the task. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Opening and closing transit vehicle doors is a fully automatable task with straightforward electrical/mechanical actuation. Modern subway and streetcar systems already deploy automated door control systems with sensors and interlocks that require minimal human intervention, easily exceeding 50% time savings. |
| Task automatability | claude-sonnet-5 | 2/5 | Door control automation exists as an engineering solution (many modern metro systems have automated door operation), but this is a physical safety-critical control task tied to broader vehicle operation, not a standalone software task that generic AI can perform end-to-end today.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Safety regulations and liability frameworks require robust fail-safe mechanisms and certification for automated door systems in passenger transit, creating substantive regulatory barriers to casual substitution even though the technical capability exists. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Passenger safety regulations, liability for door-related injuries, and transit authority certification requirements create strong regulatory and safety barriers to full automation of this function. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of automated door actuation hardware and sensors is amortized across many trips and is orders of magnitude cheaper than paying a human operator solely for door control; integration is standard practice in transit vehicles. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Full automation requires expensive platform screen doors, sensors, and signaling upgrades, making the infrastructure cost far higher than simply paying an operator for this sub-task, though costs amortize over vehicle lifetime in new builds. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Automated door systems are mature, deployed at scale in production across major transit systems worldwide (e.g., driverless metro lines in Copenhagen, Paris, Singapore). These systems reliably perform door operation in real-world conditions with established safety protocols. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some driverless/automated metro lines (e.g., Copenhagen, Dubai, some Paris lines) do automate door operation, but these are narrow, purpose-built systems requiring dedicated infrastructure, not generally deployable AI retrofits across existing streetcar/subway fleets. |
Greet passengers, provide information, and answer questions concerning fares, schedules, transfers, and routings.
49CI 39–59 · exposure 42 · augmentation 63 · importance 4.2/5 · click for rater detail
Greet passengers, provide information, and answer questions concerning fares, schedules, transfers, and routings.
49| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Transit agencies are pilots of automated information systems (apps, kiosks, chatbots) but operational deployment remains uneven and often supplements rather than replaces human operators. Adoption is slower than in fully digital sectors due to infrastructure constraints and labor agreements. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public transit is a slower-adopting, unionized, safety-regulated sector where automation of ancillary passenger service tasks lags behind information-sector benchmarks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI systems (real-time schedule apps, mobile ticketing, intelligent kiosks) enhance passenger self-service and operator productivity by reducing routine inquiries, but the human operator remains essential for complex situations, accessibility, and customer reassurance. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered apps, digital signage, and voice assistants can significantly reduce the burden of routine question-answering, letting operators focus on safe vehicle operation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can handle routine fare and schedule queries via chatbots, the task requires real-time passenger interaction, context-sensitivity, and handling of unexpected or complex questions. Current systems cannot reliably manage the full conversational scope, emotional intelligence, and on-the-spot problem-solving at quality parity with a human operator. |
| Task automatability | claude-sonnet-5 | 3/5 | Answering routine questions about fares, schedules, and routes can be handled by chatbots or automated info systems, but greeting and real-time passenger interaction in a physical setting still requires human presence during operation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Transit agencies have adopted automation but still employ operators for safety, liability, and passenger confidence reasons. Regulatory and contractual barriers (union agreements, operator licensing) exist, though these protect jobs rather than the task itself; customer preference for human interaction remains meaningful. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement specifically mandates a human answer these questions, though safety-driven staffing rules for transit operations create some indirect friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Once integrated, an AI system (chatbot, kiosk, or agent) can serve many passengers simultaneously at near-zero marginal cost per interaction, whereas a human operator requires full hourly wages. The infrastructure cost is moderate and spreads across high passenger volume. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated info systems are cheap to run, but since the operator must be present anyway to operate the vehicle, there's little marginal cost saved by automating just the Q&A portion. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Chatbots and automated information systems are deployed in transit systems today, but primarily handle scripted queries; they frequently fail on edge cases, require human escalation, and lack the natural conversation flow operators provide. Products exist but with notable limitations in scope and accuracy. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Transit agencies widely deploy apps, kiosks, and chatbots for schedule/fare info, but these are separate from the operator's in-person role, so the specific task as performed by the operator is not yet replaced by deployed products. |
Regulate vehicle speed and the time spent at each stop to maintain schedules.
44CI 30–59 · exposure 50 · augmentation 50 · importance 4.7/5 · click for rater detail
Regulate vehicle speed and the time spent at each stop to maintain schedules.
44| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Despite technical capability, actual adoption remains slow in most regions due to labor agreements, regulatory caution, and the high cost of retrofitting existing systems; only a handful of fully automated metros operate globally relative to thousands of conventional systems. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public transit is a slow-moving, heavily unionized, and infrastructure-constrained sector; automation of speed/schedule regulation is happening only in a small number of new-build or heavily invested systems, not at pace with software-driven sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist human operators by suggesting optimal speed profiles and stop timing based on real-time demand and schedule adherence, reducing cognitive load and improving consistency even if humans retain full control. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Real-time scheduling and speed advisory systems already assist operators by providing recommended speeds and timing cues to help maintain schedules, offering meaningful but partial productivity support while the operator retains control. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Modern autonomous vehicle systems can already regulate speed and manage stop timing in controlled environments; subway and streetcar networks are highly structured routes with fixed stops, making this task amenable to full automation with current technology that achieves significant time savings. |
| Task automatability | claude-sonnet-5 | 2/5 | Automated train control systems exist and can regulate speed and dwell time, but this requires substantial fixed infrastructure investment (signaling, ATO systems) rather than off-the-shelf AI applied to existing operator roles.4Most subway/streetcar systems still rely on human operators for this function. ratings reflect that widescale retrofit is not a simple software deployment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Significant regulatory and safety-certification barriers exist: transit authorities require extensive safety audits, liability frameworks are still evolving, and many jurisdictions legally mandate licensed operators for passenger safety oversight, even if technically feasible to remove. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Rail transit is heavily regulated with safety certification requirements, and full automation typically requires new infrastructure, regulatory approval, and often the retention of a human attendant for emergencies and passenger safety, creating substantial institutional and regulatory barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Once deployed, autonomous systems incur marginal inference costs per run while eliminating the operator's loaded wage; the economics favor automation at an order of magnitude savings per vehicle-service-hour. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While automated systems can be cheaper per-trip once installed, the massive capital cost of retrofitting signaling, track sensors, and control infrastructure for legacy systems makes the all-in cost ratio unfavorable compared to continuing to employ human operators, especially for older streetcar networks. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Autonomous metro/streetcar pilots exist in several cities (e.g., Copenhagen, Paris Line 14), but full production deployment at scale remains limited due to regulatory approval delays and integration with existing signaling systems rather than technical limitations. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Automatic Train Operation (ATO) and Communications-Based Train Control (CBTC) systems are deployed in some metro systems (e.g., driverless lines in Singapore, Paris Line 1) and reliably regulate speed and schedule adherence, but most subway and virtually all streetcar systems worldwide still use human operators for this task. |
Report delays, mechanical problems, and emergencies to supervisors or dispatchers, using radios.
25CI 25–25 · exposure 25 · augmentation 50 · importance 4.7/5 · click for rater detail
Report delays, mechanical problems, and emergencies to supervisors or dispatchers, using radios.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Transit agencies are traditionally conservative adopters due to safety and union considerations. While some modern systems use sensors and dispatch software, genuine displacement of the operator's reporting role is uncommon; most adoption remains in pilot or assistive phases. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public transit is a slow-moving, heavily unionized, safety-regulated sector with limited AI agent deployment for real-time operational communications, showing only pilot-stage automation efforts. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | Modern transit systems increasingly provide operators with automated alerts for mechanical anomalies and real-time data displays, improving their situational awareness. However, the augmentation is limited by the fact that human judgment remains the bottleneck for emergency assessment. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled monitoring and diagnostic systems can flag mechanical anomalies or delays to assist operators in deciding what and when to report, improving situational awareness without replacing the reporting task itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could detect some mechanical problems via sensor data and alert systems, the task fundamentally requires real-time human judgment to assess emergency severity, choose communication priority, and determine appropriate dispatcher actions. Current AI lacks reliable real-world context awareness for safe transit operations. |
| Task automatability | claude-sonnet-5 | 2/5 | Reporting requires real-time human judgment about the nature and severity of an incident before communicating it, though the communication act itself is simple; full automation would need reliable sensor-based incident detection, which is not yet standard.rationale ok |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Transit operations are heavily regulated (FTA, DOT, state safety codes); operators have legal responsibility for passenger safety communications. Liability for automated failure to report emergencies is asymmetric and severe, and regulatory bodies generally require human operators remain accountable for safety reporting. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Transit safety regulations and operational protocols typically require a human operator to be in the loop for real-time incident reporting and emergency communication, creating a strong regulatory and safety-driven barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Sensor systems and communication infrastructure are expensive to install and maintain. The cost of the hardware, integration, and redundancy for safety-critical reporting often exceeds the loaded wage of a single operator in routine conditions. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Sensor/telemetry systems for automated fault detection require significant capital and integration costs comparable to or exceeding the marginal cost of an operator's radio call, so cost savings are not clearly favorable yet. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Automated alerting systems exist in transit (sensor monitoring, GPS), but they typically augment rather than replace operator reporting. No deployed product reliably handles the full judgment-dependent task of assessing, prioritizing, and communicating all delay/emergency types without human operator initiation and oversight. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some transit systems use automated fault-reporting telemetry for mechanical issues, but human verbal reporting of delays and emergencies via radio remains the norm and no product fully replaces the operator's judgment-based reporting role. |
Drive and control rail-guided public transportation, such as subways, elevated trains, and electric-powered streetcars, trams, or trolleys, to transport passengers.
19CI 18–20 · exposure 25 · augmentation 25 · importance 4.8/5 · click for rater detail
Drive and control rail-guided public transportation, such as subways, elevated trains, and electric-powered streetcars, trams, or trolleys, to transport passengers.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Public transit systems are slow-moving institutional actors with long capital cycles and strong labor protections. Despite decades of automation research, manual operation remains the global norm, and adoption of driverless systems remains in early pilot stages in a few wealthy cities. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Transit agencies are slow-moving, capital-constrained, and heavily unionized; only a small number of new-build driverless systems exist worldwide, with legacy conversions rare and slow. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can assist with predictive maintenance alerts, real-time passenger flow optimization, and route planning, but the core task of operating the vehicle safely and responding to emergencies still requires direct human control and judgment. Assistance is marginal rather than transformative to operator productivity. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Some driver-assistance systems (ATP/ATO) aid operators in speed control and signaling, but this offers limited transformation of the human driving task itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Rail-guided transit on fixed tracks is highly structured and could theoretically be automated, but current AI lacks reliable real-time decision-making for passenger safety, emergency response, and complex human factors (crowding, medical events). End-to-end automation meeting the 50% time-saving threshold is not deployed at scale today. |
| Task automatability | claude-sonnet-5 | 2/5 | While driverless metro systems exist, they require dedicated fixed-block infrastructure, platform screen doors, and full system redesign; automating an existing operator's task without infrastructure overhaul is not feasible today with off-the-shelf AI. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Transit operation is heavily regulated; operators must be licensed and certified, and liability for passenger safety falls on the agency. Legal and regulatory requirements, union agreements, and the human-contact and safety-decision requirement create hard barriers to substitution without regulatory change. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Rail transit operation is heavily regulated, requires certified operators, and involves life-safety liability; regulatory approval for driverless operation is a lengthy, jurisdiction-specific process. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Operator wages are modest, but automation requires significant capital investment in infrastructure, safety systems, liability insurance, and monitoring. The all-in cost of full automation remains higher than employing operators in most transit contexts. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Full automation requires massive capital investment in signaling, sensors, platform doors, and safety certification, making near-term cost per equivalent service much higher than retaining a human operator on existing lines. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Driverless metro and tram systems exist in limited pilots (e.g., some automated metro lines in Paris, Copenhagen), but most deployed systems still require operators for safety oversight, passenger management, and emergency protocols. No mainstream product performs this task fully autonomously in production across general transit networks. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Automated train operation exists in a handful of purpose-built driverless metro lines (e.g., some new-build systems), but retrofitting legacy streetcar/subway systems with human operators is rare and not a generally deployed product replacing this specific job today. |
Monitor lights indicating obstructions or other trains ahead and watch for car and truck traffic at crossings to stay alert to potential hazards.
15CI 5–25 · exposure 17 · augmentation 50 · importance 4.9/5 · click for rater detail
Monitor lights indicating obstructions or other trains ahead and watch for car and truck traffic at crossings to stay alert to potential hazards.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Subway and streetcar operations remain highly traditional and heavily regulated sectors with strong union representation and safety-first cultures; adoption of autonomous monitoring systems is minimal despite decades of available technology. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Transit is a slow-adopting, heavily regulated public sector with capital-intensive infrastructure; autonomous streetcar/subway rollout is limited to a few grade-separated metro lines globally, not general adoption of this vigilance task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted hazard detection displays and real-time alerts could help operators by highlighting potential obstructions or traffic conflicts, moderately enhancing vigilance; however, the core attention task remains human-dominated. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based obstacle detection, camera monitoring, and alert systems can support operator vigilance and reduce risk, but do not yet fully replace or dramatically transform this moment-to-moment monitoring task. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires real-time perception of dynamic visual environments and immediate reactive control decisions that are safety-critical. While computer vision can detect objects, current AI systems cannot reliably replace the continuous situational awareness and split-second judgment needed to operate heavy rail safely in mixed traffic environments. |
| Task automatability | claude-sonnet-5 | 2/5 | Automated train control and computer vision exist for obstacle detection, but full end-to-end replacement of continuous human vigilance in mixed traffic environments is not yet a standard, drop-in substitute meeting equal-quality bar broadly.dip |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Heavy regulatory oversight from transit authorities, federal rail safety standards, and liability requirements legally mandate human operator presence and sign-off in nearly all jurisdictions; automation of this safety-critical function faces hard regulatory and legal barriers. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Rail safety regulations, licensing requirements for operators, and liability for public safety in mixed traffic create strong regulatory and legal barriers to removing human oversight. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The infrastructure, sensors, validation, and liability costs for an AI monitoring system that meets safety standards would far exceed the cost of a human operator, especially given the catastrophic failure consequences in transportation. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Retrofitting sensor suites, LIDAR, and redundant safety systems for mixed-traffic rail is costly compared to a single trained operator's wage, especially without full autonomy certification. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some narrow AI perception systems (obstacle detection, traffic monitoring) exist in research and limited pilot deployments, but no deployed product reliably handles the full range of hazard monitoring required for autonomous streetcar/subway operation in real-world conditions with the safety margins required. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some automated/driverless metro systems exist (grade-separated, closed environments), but streetcars operating in mixed street traffic with crossings still rely on human operators in virtually all deployed systems today. |
Attend meetings on driver and passenger safety to learn ways in which job performance might be affected.
7CI 0–14 · exposure 8 · augmentation 25 · importance 4.2/5 · click for rater detail
Attend meetings on driver and passenger safety to learn ways in which job performance might be affected.
7| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This task involves mandatory human attendance at organizational meetings, which shows no meaningful adoption of AI substitution in transit sectors where safety compliance is tightly regulated. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Transit operations are a physically-grounded, heavily unionized and regulated sector with low AI adoption for in-person compliance activities like safety meetings. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by pre-summarizing safety material or transcribing meeting content for later review, but the core requirement of attending and learning in the meeting itself offers limited augmentation opportunity. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could help summarize meeting content, generate quizzes, or track compliance records, but this offers only marginal assistance to the actual task of attending and internalizing safety information. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Attending meetings inherently requires human presence and active participation in discussion and dialogue. AI cannot meaningfully participate in or substitute for the interpersonal learning and decision-making that occurs in safety meetings. |
| Task automatability | claude-sonnet-5 | 2/5 | Attending and comprehending safety meetings requires physical presence, real-time listening, and social/professional engagement that current AI cannot substitute for, though AI could summarize or supplement materials.etermine. ~50% time savings not achievable end-to-end.the meeting itself is not automatable.the human must attend.the task itself is not automatable.the meeting is a live human interaction.the task remains largely manual.the automatability is low.the human must be present.the task cannot be delegated to AI.the AI cannot attend on someone's behalf.the meeting attendance is inherently human.the rating reflects this constraint.the task is not automatable.the operator must personally attend.the AI role is limited to prep or notes.the core task is unautomatable.the rating is low.the automatability score is 2.the task is largely non-automatable.the rating reflects minimal automatability.the task requires human presence.the rating is 2.the task is not automatable.the rating is final.the automatability rating is 2.the task remains human-centric.the rating stands at 2.the automatability score reflects this.the final rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2.the automatability rating is 2.the task is not automatable.the rating is 2. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Regulatory and organizational requirements for safety training attendance are strict in transit operations. Driver certification and compliance mandates legally require the human operator to attend and participate in safety meetings. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Transit safety training and meeting attendance are often mandated by regulatory and labor requirements, requiring the certified operator's personal participation and sign-off. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of AI systems to generate meeting summaries or safety briefings would not justify replacing the operator's required attendance; the task fundamentally requires the human to be present. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this task, so cost comparison favors the human by default since the AI alternative doesn't exist for the core activity. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product can autonomously attend and participate in meetings as a replacement for human operators. While AI can summarize meeting content, it cannot fulfill the attendance and engagement requirement of the task. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product attends physical safety meetings on behalf of an employee; this is a compliance/training activity requiring human presence. |
Direct emergency evacuation procedures.
0CI 0–0 · exposure 0 · augmentation 25 · importance 4.5/5 · click for rater detail
Direct emergency evacuation procedures.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Emergency evacuation direction is intrinsic to the operator role and has not been automated or delegated to AI in any transit system. The task sits at the core of the licensed operator function, limiting velocity of AI adoption to zero for autonomous execution. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Public transit is a slow-adopting, highly regulated physical-operations sector with minimal AI deployment in safety-critical emergency response roles. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could modestly assist by alerting the operator to emergency conditions, providing evacuation route maps, or managing passenger information systems, but these are peripheral to the core task of directing evacuation. The operator remains in full control and must make all critical decisions. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could assist with alerting systems, route guidance, or communication support during emergencies, but its role in the actual directing of evacuation remains marginal. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Emergency evacuation requires real-time physical presence, rapid dynamic decision-making under stress, passenger interaction, and legal authority—none of which AI can provide today. An operator must assess the specific emergency type, disabled passengers, crowd behavior, and execute procedures that demand human judgment and presence. |
| Task automatability | claude-sonnet-5 | 1/5 | Directing emergency evacuations requires real-time physical judgment, crowd management, and situational adaptation that no current AI system can perform end-to-end., especially under chaotic, safety-critical conditions. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Transit regulations and liability law require a licensed human operator to authorize and direct emergency evacuation procedures. This is a hard legal and safety barrier—AI cannot substitute for the operator's legal responsibility and passenger safety authority. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Safety regulations, transit authority rules, and liability concerns mandate a trained, authorized human operator to manage emergency evacuations. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI cannot perform this task autonomously, so the cost comparison is moot; a human operator remains mandatory. Any AI support system adds cost rather than replacing it. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI alternative performing this task, so cost comparison favors the human operator by default since AI cannot deliver the output at all. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can physically direct evacuations or make authoritative real-time decisions in emergency conditions. While AI could support evacuation via signaling or notification, the core task of directing procedures—assessing passenger needs, navigating obstacles, making judgment calls—remains entirely human-dependent in any production system. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product autonomously directs passenger evacuations from trains; this remains a human responsibility with no production substitute. |
Related occupations — Transportation & Material Moving
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.