Meter Readers, Utilities
43-5041.00Median wage $48,150/yr19,430 employed (US)Rank #84 of 923 scored · top 9% by substitution
Read meter and record consumption of electricity, gas, water, or steam.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution49
Exposure44
Augmentation44
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
12 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
17%
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%45
panel mean rating 2.8/5 → substitution pressure 45/100
Technical feasibility todayw 20%43
panel mean rating 2.7/5 → substitution pressure 43/100
Cost vs. human wagew 15%55
panel mean rating 3.2/5 → substitution pressure 55/100
Adoption barriersw 20%inverted — strong barriers lower the score62
panel mean rating 2.5/5 (barrier strength) → substitution pressure 62/100
Sector adoption velocityw 10%41
panel mean rating 2.6/5 → substitution pressure 41/100
Task breakdown (12 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.
Upload into office computers all information collected on hand-held computers during meter rounds, or return route books or hand-held computers to business offices so that data can be compiled.
91CI 84–97 · exposure 92 · augmentation 50 · importance 4.3/5 · click for rater detail
Upload into office computers all information collected on hand-held computers during meter rounds, or return route books or hand-held computers to business offices so that data can be compiled.
91| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Utilities are information-heavy, digitized organizations with established IT infrastructure and strong financial incentives to reduce labor costs in routine back-office tasks. Mobile data synchronization and automated upload are already common in deployed field-service systems across the sector. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Utilities have broadly adopted automated meter reading and data telemetry systems, though some legacy manual-read routes and route-book processes persist in smaller utilities. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by validating data completeness, flagging anomalies, or automatically routing uploads to appropriate systems, reducing the cognitive load on meter readers preparing end-of-day submissions. However, the task itself is already highly structured, limiting transformation potential. |
| Augmentation potential | claude-sonnet-5 | 3/5 | For remaining manual processes, AI/software can streamline validation and error-checking during upload, though the core transfer task is already largely automated rather than merely augmented. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Uploading data from handheld devices into office computers is primarily a data transfer and integration task that current systems can fully automate with minimal human intervention. The task involves mechanical data movement with minimal judgment, allowing current AI tools to handle it reliably and save well over 50% of manual time. |
| Task automatability | claude-sonnet-5 | 5/5 | Uploading or transferring data from handheld devices to office systems is a straightforward, structured data-transfer task that is already fully automatable via sync software, APIs, or automated meter reading (AMR) systems. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While some organizations have legacy system integration requirements or prefer human oversight of data collection completeness, there are no legal licensing barriers or mandatory human sign-off required for this data-handling step. Integration friction exists but is not insurmountable. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or human-judgment requirements attach to data upload/transfer; it's a purely administrative/technical step. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated data synchronization and upload incurs minimal inference cost (primarily API calls and data pipeline execution), likely orders of magnitude cheaper than paying a meter reader to manually upload data or return equipment to an office. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated data upload/sync costs are negligible (software/hardware infrastructure) compared to manual return-and-compile labor time. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Data upload and integration from field devices to enterprise systems is a well-solved problem with mature, production-deployed solutions across utilities and field-service industries. Systems handling this at scale exist in major utility companies today. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Deployed handheld meter reading devices and utility billing software routinely auto-sync or upload data via docking stations, wireless transfer, or cloud sync in production today. |
Update client address and meter location information.
80CI 76–84 · exposure 75 · augmentation 63 · importance 3.8/5 · click for rater detail
Update client address and meter location information.
80| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Utilities are mature adopters of digital customer portals and automated billing systems; address and location updates are routinely collected and processed automatically in modern utility operations. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Utilities are moderately digitized with growing smart-meter and CRM integration, but many still rely on manual processes and legacy systems, placing adoption in the middle tier. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by flagging inconsistent or suspicious address changes for manual review, cross-referencing with external data sources, and auto-populating forms—useful productivity aids without full replacement of human judgment on edge cases. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-assisted data validation, address standardization, and GIS mapping tools significantly speed up and reduce errors in this task even when humans remain involved for verification. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Address and meter location updates can be largely automated through data extraction from forms, OCR, and database management systems. While some ambiguity resolution may require human review, the core task of parsing and updating records achieves >50% time savings with current tools. |
| Task automatability | claude-sonnet-5 | 4/5 | Updating structured records like addresses and meter locations is a routine data-entry task that AI/software systems can handle end-to-end with substantial time savings, given integration with utility databases and GPS/mapping data. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or liability barriers exist for updating address/location records in utility databases; the task is administrative data management with minimal legal sign-off requirements, though some customer verification policies may add minor friction. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There is no licensing, liability, or regulatory requirement mandating human performance of this administrative record-keeping task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated database updates cost orders of magnitude less than manual data entry by field workers; a single integration system serves thousands of updates at near-zero marginal cost after deployment. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated data entry and database update systems cost a small fraction of a human worker's time for the same repetitive administrative task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed systems in utility companies already automate address/location data ingestion through customer portals, IVR systems, and data integration pipelines. Error rates are generally low for well-structured input, though edge cases and unstructured submissions still occur. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Utility companies widely use automated meter infrastructure (AMI) and CRM/GIS systems that update customer and location data automatically or via simple validation workflows, though some edge cases still require human verification. |
Read electric, gas, water, or steam consumption meters and enter data in route books or hand-held computers.
67CI 55–79 · exposure 70 · augmentation 38 · importance 4.5/5 · click for rater detail
Read electric, gas, water, or steam consumption meters and enter data in route books or hand-held computers.
67| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Utility companies are typically large but conservative, risk-averse entities with entrenched legacy processes and regulatory scrutiny; pilot adoption exists but sector-wide production displacement remains slow and limited, especially in non-digital meter infrastructure. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Utilities have been rapidly deploying smart meters over the past 15 years, with high penetration in many developed markets, though full transition is not complete everywhere. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | Mobile apps and computer vision can assist meter readers by auto-capturing images and validating readings in real-time, improving accuracy and reducing manual data entry, though the core task remains largely human-driven in most deployments today. |
| Augmentation potential | claude-sonnet-5 | 2/5 | For the remaining manual meter-reading tasks, hand-held computers and route optimization software provide some assistance, but the core reading action itself gains little from AI-based augmentation beyond basic digitization. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Modern computer vision and mobile systems can autonomously read digital/analog meters and log data directly to databases with high accuracy, achieving substantial time savings over manual entry. The core bottleneck remains physical access to meters and occasional problematic meter conditions, preventing a full 5. |
| Task automatability | claude-sonnet-5 | 4/5 | Meter reading is largely being replaced by automated meter reading (AMR) and smart meter infrastructure that transmits data wirelessly, eliminating the need for manual reads in most modernized utility networks.It falls short of a 5 because a substantial number of legacy meters and rural/older infrastructure still require manual reading. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory requirements, utility commission approval, and liability concerns around billing accuracy create moderate-to-strong friction; utilities are risk-averse about autonomous billing systems, and safety protocols around hazardous gas/electrical equipment require procedural oversight. |
| Adoption barriers | claude-sonnet-5 | 2/5 | There is no licensing or legal requirement for a human to physically read meters, though utility regulatory approval and capital investment for meter replacement create some adoption friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Computer vision-based meter reading with mobile integration is orders of magnitude cheaper than the fully-loaded cost of sending human meter readers to every location, especially considering vehicle costs, labor, and scheduling overhead. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Once smart meters are installed, automated data collection costs a small fraction of a human reader's wage per read, as the marginal cost of transmission is near zero compared to physical route-based reading. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI-powered meter reading solutions exist and are deployed in pilot programs and some utility companies, but adoption is mixed; many systems still require human verification or handle only digital meters, indicating material limitations in real-world reliability. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Smart meter and AMI (advanced metering infrastructure) systems are mature, widely deployed products used by major utilities at scale, reliably capturing consumption data without human readers.Full replacement is not universal since many utilities still operate mixed fleets with legacy meters. |
Walk or drive vehicles along established routes to take readings of meter dials.
64CI 41–87 · exposure 58 · augmentation 13 · importance 4.3/5 · click for rater detail
Walk or drive vehicles along established routes to take readings of meter dials.
64| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Utilities have strong financial incentives to automate meter reading (cost reduction, accuracy, coverage), and large regulated utilities have already invested in automated meter infrastructure (AMI) and AI-based reading systems. Adoption is accelerating, particularly among well-capitalized utilities in digitized sectors, though lagging in small municipal systems. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Utilities have adopted smart meters at a steady but uneven pace over the past two decades, with substantial variation by region and utility size; this is a physical infrastructure sector, not a fast-moving software adoption pattern. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Meter reading leaves little room for human-AI augmentation once image capture is automated; the task is inherently about accurate conversion of a dial reading to a number. Any AI assistance is marginal compared to full automation, as the human in the loop adds cost without meaningful judgment enhancement. |
| Augmentation potential | claude-sonnet-5 | 1/5 | There is little role for AI to augment a human physically walking or driving a route to read meter dials; navigation apps exist but are not AI-specific productivity tools for this task. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Meter reading is among the most automatable utility tasks: routes are established and fixed, readings are quantitative dial-to-digit conversions, and computer vision systems already achieve high accuracy on meter imagery. Current AI-powered meter reading solutions (optical character recognition on meter photos, autonomous vehicle routing) achieve well over 50% time savings at equal or better accuracy compared to human walkers. |
| Task automatability | claude-sonnet-5 | 2/5 | The physical navigation and reading portion cannot be automated by generative AI systems; true automation here requires smart meters/AMI hardware infrastructure rather than 'AI' per se, and adoption depends on utility capital investment cycles.To the extent AI applies, it's minimal for this specific mobile task. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Meter reading is lightly regulated; there are no statutory licensing or sign-off requirements for the reading task itself. Barriers are primarily organizational inertia and legacy workforce contracts rather than legal or compliance mandates, making adoption friction moderate but surmountable. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for meter reading, but utility infrastructure upgrades face regulatory approval, capital budgeting cycles, and customer property access issues that slow full transition to automated systems. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI meter reading (combining computer vision, routing optimization, and automated data processing) costs substantially less per meter than dispatching a human worker (vehicle, fuel, wages, benefits, routing overhead). The cost differential is roughly an order of magnitude or greater when labor, logistics, and equipment are fully loaded. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Smart meter infrastructure has high upfront capital cost but low marginal cost per reading, making it cheaper long-term than human labor, though initial deployment cost can rival or exceed labor savings in the short term. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products for automated meter reading exist and are in production use by major utilities (computer vision-based image capture, AI dial-reading engines). Some systems still require human oversight or have edge cases with obscured/damaged dials, but the core task is reliably performed at scale by off-the-shelf or vendor-integrated solutions. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Automated meter reading (AMR) and smart meter systems exist and are deployed widely, but they are hardware/telemetry solutions, not AI-driven; for locations still requiring manual walk/drive routes, no AI product performs this task today. |
Answer customers' questions about services and charges, or direct them to customer service centers.
61CI 52–70 · exposure 55 · augmentation 63 · importance 3.9/5 · click for rater detail
Answer customers' questions about services and charges, or direct them to customer service centers.
61| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Utility and energy companies are actively deploying AI-powered chatbots and virtual assistants in production; this reflects the information/service sector's faster adoption of customer-facing automation. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Utility field operations are a low-digitization, physically-oriented sector where AI adoption for this specific interpersonal task lags far behind office-based customer service automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist human customer service representatives by drafting responses, retrieving account information, and suggesting solutions, enabling faster and more accurate handling of customer inquiries while the human remains in control. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Meter readers could be equipped with AI-powered mobile apps or chat assistants to quickly look up account/billing info to answer customer questions more efficiently, though this isn't yet standard practice. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI chatbots and virtual assistants can handle routine inquiries about billing, service status, and basic account questions, but complex or sensitive issues (disputes, special accommodations, account adjustments) typically require human intervention, limiting automation to roughly 50% of interactions. |
| Task automatability | claude-sonnet-5 | 3/5 | Answering routine questions about services/charges can be handled by chatbots or IVR systems today, but the field context and need to route complex billing disputes limits full end-to-end automation for a meter reader's ad hoc interactions. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While customer preference for human contact and regulatory oversight of billing disclosures create some friction, no hard licensing or legal requirement mandates human performance of routine customer service answering; utilities deploy automation readily. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement blocks automation of general billing Q&A, though utilities may have some liability concerns about inaccurate information being given regarding charges. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI chatbot inference and maintenance costs are orders of magnitude cheaper than the loaded wage of a meter reader or customer service representative answering phone/chat inquiries, especially when answering repetitive questions. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | While AI-driven customer service phone lines are cheap per interaction, replacing the meter reader's ad hoc face-to-face Q&A requires additional infrastructure (kiosks, apps) making the cost comparison less favorable than pure call-center automation. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed chatbot systems across utilities companies reliably handle common customer service queries at scale, though some organizations still report material fallback rates for edge cases and prefer human escalation for high-value or sensitive matters. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed customer service chatbots and phone systems handle utility billing questions at scale, but this task occurs in-field via meter readers, where no product currently substitutes for that real-time human interaction. |
Report lost or broken keys.
61CI 24–97 · exposure 58 · augmentation 38 · importance 3.7/5 · click for rater detail
Report lost or broken keys.
61| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Utility and facility management companies are actively digitizing incident reporting and asset management workflows, with significant adoption of automated ticketing and mobile reporting systems in production environments. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | placeholder |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can assist humans by auto-populating reports with meter reader location data, date, time, and key details, then routing to the correct department, substantially reducing manual data entry and administrative burden. |
| Augmentation potential | claude-sonnet-5 | 1/5 | placeholder |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Reporting a lost or broken key is a straightforward administrative task requiring only documentation of an incident. Current AI systems can fully automate this via form submission, email, or API integration with facility management systems, achieving 100% time savings at equal or better quality (legibility, completeness). |
| Task automatability | claude-sonnet-5 | 2/5 | Reporting a lost/broken key is a trivial administrative act but still requires a human to notice, recall, and initiate the report; AI could log or transcribe it but not perform the underlying observation and decision to report.','rating_note':''}, |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While there are minimal legal barriers to automating routine reporting, some organizations may require human sign-off or verification for asset tracking purposes, creating light but present friction. |
| Adoption barriers | claude-sonnet-5 | 1/5 | placeholder |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automating key-loss reporting costs only seconds of compute time per incident, plus minimal integration overhead, making it orders of magnitude cheaper than paying a human meter reader to manually file a report. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | placeholder |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Deployed products in facility management, helpdesk ticketing systems, and business process automation already perform this task reliably at scale through automated incident reporting and form processing. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | placeholder |
Leave messages to arrange different times to read meters in cases in which meters are not accessible.
51CI 43–59 · exposure 38 · augmentation 50 · importance 4.0/5 · click for rater detail
Leave messages to arrange different times to read meters in cases in which meters are not accessible.
51| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Utilities are mid-digitization: larger utilities have pilot IVR and SMS systems for notifications, but production adoption of autonomous scheduling agents for meter access is still emerging rather than widespread. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Utilities are generally slower-moving, capital intensive, and less digitized than professional services, though customer communication systems are a common early automation target. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by drafting messages, suggesting alternative time slots based on historical patterns, and flagging recurring access issues, thereby reducing the mental load on schedulers while they retain decision authority over final outreach. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-generated scheduling messages and automated dialing/texting tools can help meter readers coordinate access more efficiently, but the underlying access problem still requires human coordination. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Leaving messages involves asynchronous communication, which AI can do (e.g., generating or sending texts/emails), but this task requires understanding specific meter accessibility issues, scheduling constraints, and customer contact preferences—factors that are often context-dependent and would require significant human judgment to handle reliably at scale. |
| Task automatability | claude-sonnet-5 | 3/5 | Message drafting and even automated call/text scheduling can be handled by AI voice/SMS systems, though physical follow-up remains human, so only the communication portion is automatable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Utilities are regulated but not heavily licensed for message-leaving itself; however, meter-access disputes can involve customer service standards and dispute resolution requirements that create organizational friction against full automation without human oversight. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for leaving a message, though customer-facing communication may require some quality control and utilities may prefer consistent human-vetted messaging. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Sending automated messages via API is extremely cheap compared to a human wage ($15–30/hour loaded), though integration and oversight of scheduling failures would add modest cost; AI-driven messaging systems operate at roughly 1–5% of human labor cost. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated messaging/scheduling systems cost pennies per interaction versus a human worker's time to call or leave notes, making AI substantially cheaper for this narrow communication task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI systems can generate messages and send communications, no deployed product reliably handles the full end-to-end task of diagnosing why a meter is inaccessible, determining appropriate alternative times, and managing the two-way scheduling negotiation that typically follows in production utility operations. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Automated appointment-reminder and scheduling systems (IVR, SMS bots) are deployed in utilities today, but many meter-reading operations still rely on manual note-leaving or human dispatch calls. |
Verify readings in cases where consumption appears to be abnormal, and record possible reasons for fluctuations.
44CI 32–55 · exposure 38 · augmentation 75 · importance 4.2/5 · click for rater detail
Verify readings in cases where consumption appears to be abnormal, and record possible reasons for fluctuations.
44| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Utilities are mid-stage in AI adoption for meter-reading and anomaly detection, with pilots and early production deployments common, but full end-to-end automation of verification remains rare due to liability and operational complexity. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Utilities have moderately adopted smart-meter analytics and AI-driven anomaly detection, but broader deployment lags information/finance sectors due to legacy infrastructure and slower digitization. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI anomaly detection can substantially assist meter readers by automatically prioritizing cases with unusual patterns, reducing manual sampling burden and guiding investigation—a clear productivity boost while the human remains responsible for final verification and root-cause determination. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-based anomaly detection significantly aids meter readers/analysts by pre-flagging suspicious readings and suggesting likely causes, substantially speeding up the verification and documentation process. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Only a small fraction of the verification work can be automated. While AI can flag abnormal consumption patterns against historical data, determining the actual cause of fluctuations (equipment failure, customer behavior change, billing error) requires human judgment, site knowledge, and investigation—most of which remains manual. |
| Task automatability | claude-sonnet-5 | 3/5 | Flagging anomalous consumption via statistical comparison to historical data is straightforward for AI/software, but verifying the physical cause often requires site visits or contextual judgment that AI cannot fully replace end-to-end.meter access, weather, or occupancy changes. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Utilities face regulatory requirements for meter accuracy and consumer protection; human verification is often required before billing adjustments. However, AI-assisted flagging and triage face moderate friction rather than hard legal blocks on deployment. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement blocks automation, though utility companies may have some organizational friction and customer-facing verification norms that slow full replacement of human judgment. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Anomaly-detection infrastructure and human verification oversight together are not substantially cheaper than a meter reader's loaded hourly wage; the task still requires significant human intervention to validate and act on AI flags. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated anomaly detection software is cheap to run, but the verification step requiring physical inspection or customer contact still incurs comparable labor costs, keeping overall cost roughly on par with human-driven processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Anomaly detection systems exist in utility platforms, but no deployed end-to-end product reliably diagnoses *reasons* for abnormal readings at scale. AI can identify statistical outliers but struggles with causal attribution in production utility environments. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Smart meter analytics platforms already flag abnormal consumption in production, but the 'verify' and 'record reasons' portions often still require human follow-up or field investigation, limiting full deployment reliability. |
Inspect meters for unauthorized connections, defects, and damage, such as broken seals.
28CI 25–30 · exposure 25 · augmentation 50 · importance 4.2/5 · click for rater detail
Inspect meters for unauthorized connections, defects, and damage, such as broken seals.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Utility companies are large, risk-averse, and heavily regulated; adoption of AI-driven meter inspection remains in pilot phases with minimal production displacement. Digital meter rollouts and remote reading prioritize data collection over automated defect detection. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Utilities are a traditionally slow-adopting, capital-intensive sector; while AMI adoption is growing, dedicated AI-driven tamper/damage detection is still in pilot stages. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI image analysis could assist meter readers by flagging potential defects or damage in photos for human review, improving detection consistency and field efficiency without removing the inspector from the decision loop. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled analytics and computer vision can help flag anomalies or prioritize meters for inspection, meaningfully assisting human inspectors without replacing on-site judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI could assist with image analysis of meter photos to detect some defects and damage patterns, but unauthorized connections and broken seals require physical inspection, spatial reasoning in complex environments, and visual context that current systems struggle with reliably. This task cannot achieve 50% time saving end-to-end without significant human involvement. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires physical presence to visually and sometimes manually inspect meters and seals for tampering, which off-the-shelf AI cannot yet perform end-to-end without robotics or human-operated cameras.imestamp |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Utilities face regulatory requirements for meter integrity verification, potential liability for missed defects that enable theft or safety hazards, and customer contact/physical access requirements that cannot be fully automated. Regulatory bodies often mandate human sign-off on inspection records. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement mandates a human specifically, but liability for missed theft/damage, property access issues, and physical verification needs create moderate friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | On-site meter inspection requires physical presence, and the cost of image capture devices, storage, processing, and human oversight for verification approaches or exceeds the cost of a meter reader, especially when false negatives carry liability risk. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Automated meter infrastructure (AMI) reduces routine reading costs, but detecting physical tampering/damage still requires human inspection or expensive sensor retrofits, keeping costs comparable to human labor for this specific subtask. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Computer vision can identify some defects in controlled meter images, but deployed products lack the robustness to handle varied lighting, angles, meter types, and environmental conditions encountered in field work. No production system reliably performs full inspection and defect detection at scale in real utility operations. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Smart meter systems and some computer-vision-equipped handheld devices exist for anomaly flagging, but reliable autonomous physical inspection for tampering/defects is not deployed at scale. |
Report to service departments any problems, such as meter irregularities, damaged equipment, or impediments to meter access, including dogs.
21CI 7–35 · exposure 13 · augmentation 38 · importance 4.1/5 · click for rater detail
Report to service departments any problems, such as meter irregularities, damaged equipment, or impediments to meter access, including dogs.
21| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Utilities are moderately digitized, but this particular task—physical inspection and hazard assessment—remains labor-intensive and human-dependent. Automation pilots exist for meter reading itself, but field problem reporting lags. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Utility field services are a low-digitization, physical-labor sector where AI adoption for these micro-tasks remains in pilot stages (e.g., smart meter analytics) rather than widespread production use for anomaly/hazard reporting. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Mobile apps and dashboards could assist meter readers in logging and routing reports, but AI cannot meaningfully augment the core judgment task of identifying physical irregularities and safety hazards in the field. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist by auto-generating structured reports from a reader's voice or photo input, and by using computer vision to help flag irregularities, improving speed and consistency of reporting. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical site inspection, judgment about equipment damage/irregularities, and assessment of hazards like dogs. Current AI cannot navigate physical spaces, inspect meter conditions in person, or make safety determinations in the field without human intervention. |
| Task automatability | claude-sonnet-5 | 2/5 | This task requires physical presence to observe meter conditions and site hazards, though the reporting/logging step once an observation is made could be text-generated by AI; the perception component itself is not automatable with current general-purpose systems.dial-based meters. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Utility companies operate under regulatory oversight, and equipment inspection for safety and liability reasons typically requires licensed/certified personnel to perform and sign off on reported irregularities and hazards. |
| Adoption barriers | claude-sonnet-5 | 2/5 | There are no licensing requirements for this specific sub-task, though safety liability (e.g., failing to report a dangerous animal) creates some organizational caution around fully automated reporting without human verification. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Current AI lacks the embodied presence needed for this task. The cost of robotic systems capable of field inspection, combined with ongoing human oversight, would exceed the cost of human meter readers for this function. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI-assisted flagging tools have low marginal cost but still require a human physically present to observe and often to input data, so overall cost savings versus a human meter reader doing this sub-task are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can autonomously visit meter locations, assess physical damage, and report problems to service departments. Computer vision alone cannot replace on-site human inspection, and regulatory compliance requires human certification. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some smart meter systems and computer vision on handheld devices can flag anomalies, but broad deployed products that reliably detect diverse physical impediments like dogs or damaged equipment in the field are not mainstream today. |
Perform preventative maintenance or minor repairs on meters.
12CI 5–19 · exposure 8 · augmentation 25 · importance 3.7/5 · click for rater detail
Perform preventative maintenance or minor repairs on meters.
12| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Meter reading and maintenance remain dominated by human field workers; adoption of robotic or AI-driven automation is negligible in the utilities sector. Physical, on-site work with regulatory oversight limits digitization-driven velocity. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Utility field maintenance is a physical, low-digitization sector with slow AI/robotics adoption for hands-on repair tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Current AI offers minimal assistance to meter technicians; some diagnostic software and predictive analytics on meter data exist, but hands-on repairs remain human-dependent. Limited augmentation potential without embodied AI systems. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with diagnostics, scheduling, or predictive maintenance alerts, but offers minimal direct assistance during the actual physical repair or cleaning process. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Meter maintenance and minor repairs require physical manipulation, access control, and diagnosis of mechanical/electrical faults. Current AI lacks embodied robotics at the precision and cost-effectiveness needed; only conceptual (not deployed) robot systems exist for this work, well below the 50% time-saving bar. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical dexterity to inspect, clean, and repair meters in the field, which current AI systems (software-based, non-embodied) cannot perform end-to-end.rk Robots/AI cannot yet perform this physical hands-on task.rk |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Utility companies operate under regulatory frameworks requiring licensed technicians to certify meter installations and repairs. Safety, customer access, and liability concerns create strong legal and organizational barriers to full automation without human sign-off. |
| Adoption barriers | claude-sonnet-5 | 3/5 | While not strictly licensed work in all jurisdictions, utility companies impose safety training, certification, and liability requirements for handling metering equipment and electrical/gas connections. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Specialized robotic systems capable of meter maintenance do not exist at competitive cost. AI/robotic development, integration, and maintenance would far exceed the loaded wage of a meter reader, making automation economically infeasible. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing the physical repair work, so AI cost cannot be compared favorably; a human technician remains necessary. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI or robotic systems reliably perform meter maintenance at scale in production environments today. This remains almost entirely manual human work, requiring physical inspection, troubleshooting, and hands-on repair. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously performs physical minor repairs or preventative maintenance on utility meters; this remains a human field-technician task. |
Connect and disconnect utility services at specific locations.
8CI 0–16 · exposure 0 · augmentation 25 · importance 4.0/5 · click for rater detail
Connect and disconnect utility services at specific locations.
8| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Utility companies operate in heavily regulated, physical-infrastructure sectors with minimal digital-first adoption patterns. No meaningful adoption of autonomous systems for this task is visible in the industry. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Utilities sector is adopting smart grid and remote metering technology gradually, but physical service connection/disconnection automation is still nascent and slow-moving relative to information-sector AI adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could marginally assist with route planning or work-order scheduling, but the core task of physically connecting/disconnecting services offers limited augmentation potential beyond logistics optimization. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with scheduling, routing, and remote diagnostics to support technicians, but offers limited direct assistance to the physical act of connecting/disconnecting services. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Connecting and disconnecting utility services requires physical manipulation at specific real-world locations, including handling of infrastructure (gas, water, electric lines) that current AI systems cannot perform. This task is fundamentally outside the scope of deployed autonomous systems today. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical manipulation of valves, switches, or meters at a customer location, which current AI systems cannot perform without a robotic embodiment; no off-the-shelf system does this end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Utility service connections are heavily regulated and require licensed technicians in most jurisdictions. Legal liability, safety certification, and formal authorization create hard barriers to any form of substitution or even unsupervised automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Utility connection/disconnection often involves safety regulations, liability for gas/electric mishandling, and sometimes requires certified technicians, creating moderate regulatory and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI cannot perform this task at any cost today, so human labor remains the only option. The cost comparison is moot, but human technicians will always be required for the foreseeable future. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Where remote-controlled smart meters exist, switching costs are trivially low, but for the broader task involving physical site work, human labor remains necessary and AI offers no cost advantage. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No current AI product reliably performs physical utility connections/disconnections at customer sites. This remains a task requiring human technicians with specialized training and on-site presence. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | There are no deployed AI/robotic products that autonomously connect or disconnect utility services in the field at scale; this remains manual technician work, sometimes aided by smart meter remote switching but not general connection/disconnection tasks. |
Related occupations — Office & Administrative Support
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.