Data Entry Keyers
43-9021.00Median wage $41,340/yr127,080 employed (US)Rank #10 of 923 scored · top 1% by substitution
Operate data entry device, such as keyboard or photo composing perforator. Duties may include verifying data and preparing materials for printing.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution74
Exposure72
Augmentation55
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
9 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
67%
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%74
panel mean rating 3.9/5 → substitution pressure 74/100
Technical feasibility todayw 20%70
panel mean rating 3.8/5 → substitution pressure 70/100
Cost vs. human wagew 15%80
panel mean rating 4.2/5 → substitution pressure 80/100
Adoption barriersw 20%inverted — strong barriers lower the score82
panel mean rating 1.7/5 (barrier strength) → substitution pressure 82/100
Sector adoption velocityw 10%63
panel mean rating 3.5/5 → substitution pressure 63/100
Task breakdown (9 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.
Maintain logs of activities and completed work.
99CI 97–100 · exposure 100 · augmentation 63 · importance 4.2/5 · click for rater detail
Maintain logs of activities and completed work.
99| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Data-driven organizations across finance, tech, logistics, and professional services have already adopted automated logging at scale through workflow automation, RPA, and system integrations; this is a mature, widely deployed pattern. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Automated logging and audit trails are already deeply embedded in most digitized business software and workflows, representing widespread, mature adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can significantly assist human data entry keyers by auto-populating fields, suggesting log entries from activity streams, and flagging missing information, substantially raising their productivity while keeping oversight in human hands. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Where humans still maintain logs manually, AI tools can auto-fill, summarize, or format entries, offering moderate productivity gains for any residual manual logging. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Maintaining logs of completed work is highly automatable. Current AI can extract, structure, and record work activities from various sources (emails, project management tools, timestamps) with high accuracy, easily achieving 50%+ time savings at equal or better quality through intelligent data capture and logging systems. |
| Task automatability | claude-sonnet-5 | 5/5 | Logging activities and completed work is a structured, repetitive record-keeping task that can be fully handled by automated systems, scripts, or workflow tools that timestamp and log actions without human intervention. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | There are no meaningful legal, regulatory, or authorization barriers to automating activity logging. No licensed professional sign-off is required, and organizations have full discretion to implement automation without compliance concerns. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There are no licensing, liability, or regulatory requirements mandating human maintenance of internal activity logs; nothing prevents full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of automated logging (cloud APIs, RPA, or simple integrations) is orders of magnitude cheaper than paying a data entry keyer to manually maintain logs, especially at scale with high-volume activity streams. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated log generation via software is essentially free at scale, costing a tiny fraction of paying a human keyer to manually record the same activities. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Deployed products for automated activity logging and work tracking (RPA tools, API-driven integrations, workflow automation platforms) demonstrably perform this task reliably in production across many organizations, with minimal error rates on routine logging activities. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Automated logging is already standard in production systems via ERP, ticketing, and workflow software that generates activity logs automatically as a byproduct of digital work processes. |
Store completed documents in appropriate locations.
97CI 97–97 · exposure 100 · augmentation 50 · importance 4.4/5 · click for rater detail
Store completed documents in appropriate locations.
97| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Document management and RPA adoption is widespread in large organizations, financial services, healthcare, and government; mid-market and smaller enterprises are catching up. This task is among the earliest and most commonly automated in back-office operations. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office and administrative functions across many sectors have widely adopted document management and workflow automation tools already. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist humans by auto-classifying or suggesting document locations before filing, reducing manual decision-making and human error on ambiguous or complex documents. |
| Augmentation potential | claude-sonnet-5 | 3/5 | For any remaining manual filing needs, AI can suggest storage locations or auto-tag documents, though the task itself is largely already automated rather than augmented. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Document storage and filing based on metadata or predetermined rules is fully automatable. Current systems can classify documents, rename them according to standards, and move them to the correct digital or physical storage locations with high accuracy and easily >50% time savings. |
| Task automatability | claude-sonnet-5 | 5/5 | Filing and storing digital documents in appropriate folders/systems is a rule-based task easily handled by automated document management workflows and scripts.implify. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | There are no regulatory, legal, or licensing requirements mandating human involvement in document storage. Organizations can freely automate this task without compliance risk. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or human-contact requirements apply to storing completed documents; it's purely administrative. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automation cost (RPA licensing, OCR API calls, integration overhead) is orders of magnitude cheaper than the loaded cost of a human data entry keyer performing filing tasks full-time. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated file storage/routing costs a tiny fraction of a cent per document compared to manual labor time for filing. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature production systems (RPA, document management platforms, workflow automation) reliably perform document filing and storage at scale in enterprise environments today. Optical character recognition and intelligent document classification are well-established in deployed products. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Document management systems, RPA tools, and cloud storage automation already reliably route and store files at scale in production environments. |
Compare data with source documents, or re-enter data in verification format to detect errors.
95CI 92–97 · exposure 100 · augmentation 50 · importance 4.5/5 · click for rater detail
Compare data with source documents, or re-enter data in verification format to detect errors.
95| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | RPA and automation adoption in data-heavy industries (finance, insurance, healthcare) is rapid and measurable; many organizations have already displaced human verification staff with automated workflows, indicating deep adoption in information/processing sectors. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Data entry and verification automation is widespread in finance, healthcare, and administrative sectors, with RPA and OCR deployment being common and mature, though some legacy manual processes remain. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can highlight suspect records or flag anomalies for human review, improving a keyer's error-detection rate, but the task is inherently suited to full automation rather than human-in-loop assistance. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-assisted validation tools can flag discrepancies for human reviewers, improving speed and accuracy on complex or messy source documents, though pure automation often replaces rather than augments this task. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Data comparison and verification can be fully automated using OCR, regex matching, and AI-driven error detection; modern systems can identify discrepancies between source documents and entered data with >50% time savings and equal or superior accuracy. |
| Task automatability | claude-sonnet-5 | 5/5 | Comparing structured data against source documents (OCR/validation matching) is a well-solved automation problem using OCR, RPA, and validation scripts, meeting the 50% time-saving bar easily. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Light regulatory oversight exists for most data entry contexts; some sectors (healthcare, finance) may require human sign-off on critical records, and legacy systems integration can create friction, but no hard legal barrier prevents automation of the comparison task itself. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or human-contact requirements exist for this back-office verification task; organizations readily adopt automated validation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated verification costs pennies per transaction (API calls, compute) versus human data entry/verification wages ($15–25/hour loaded), yielding cost ratios of 1:100 or better for high-volume work. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated OCR/validation software costs pennies per document compared to a human keyer's wage for the same verification task, an order-of-magnitude cost difference. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Production-grade systems for data validation, document comparison, and automated error detection are widely deployed across finance, healthcare, and administrative sectors, with mature tools like UiPath, Blue Prism, and native database validation systems operating reliably at scale. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Mature products (RPA tools, OCR/ICR systems, double-entry verification software) already perform automated data comparison and error detection reliably at scale in production environments like finance and healthcare data processing. |
Read source documents such as canceled checks, sales reports, or bills, and enter data in specific data fields or onto tapes or disks for subsequent entry, using keyboards or scanners.
92CI 87–97 · exposure 95 · augmentation 50 · importance 4.5/5 · click for rater detail
Read source documents such as canceled checks, sales reports, or bills, and enter data in specific data fields or onto tapes or disks for subsequent entry, using keyboards or scanners.
92| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Financial services, insurance, and large retailers have already widely deployed automated document processing and RPA for data entry tasks. Adoption is deep and accelerating in digitized sectors, though lagging in small firms and cash-based operations. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Finance, banking, and back-office administrative sectors have aggressively adopted automated data capture and OCR pipelines for years, representing one of the most mature automation use cases. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists data entry workers by flagging uncertain fields for review, suggesting corrections, and auto-populating common fields, but the core task lends itself more to replacement than augmentation as automation improves. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Where full automation isn't yet reliable (poor scans, unusual formats), AI-assisted entry (auto-fill suggestions, validation) still meaningfully speeds up remaining human keyers. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Data entry from source documents is highly automatable through OCR, document parsing, and structured data extraction. Current AI systems can read and digitize most standard documents (checks, reports, bills) and populate fields with >50% time savings compared to manual keying, meeting the Eloundou threshold. |
| Task automatability | claude-sonnet-5 | 5/5 | OCR/ICR and document AI systems can extract structured data from checks, invoices, and reports with high accuracy, meeting the ≥50% time-saving bar for most standardized documents. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers exist; organizations can freely substitute AI for manual data entry. Minor friction includes integration with legacy systems and the need for occasional human verification on ambiguous documents, but nothing prevents substitution. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or human-contact requirements attach to routine data entry from documents; it's a purely clerical function with minimal regulatory protection. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based document scanning and data extraction is orders of magnitude cheaper than human keyers when accounting for infrastructure, labor, and error correction. A single OCR/RPA system can process thousands of documents daily at near-zero marginal cost. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated OCR/data-capture pipelines cost fractions of a cent per document versus a human keyer's per-hour wage, delivering order-of-magnitude cost savings at volume. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Multiple mature products (automated invoice processing, check scanning platforms, RPA tools) reliably perform this task in production across banking, accounting, and retail sectors today at scale with acceptable error rates. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Mature production systems (e.g., bank check processing, invoice capture platforms like ABBYY, Google Document AI) already handle this at scale, though messy handwriting or nonstandard forms still require exception handling. |
Compile, sort, and verify the accuracy of data before it is entered.
85CI 84–86 · exposure 75 · augmentation 75 · importance 4.7/5 · click for rater detail
Compile, sort, and verify the accuracy of data before it is entered.
85| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Data-heavy industries (finance, healthcare, logistics, e-commerce) have rapidly deployed RPA and ML-based validation tools at scale; this is a high-digitization task with strong economic incentives and demonstrated production deployment across multiple sectors. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Back-office and administrative data processing functions across finance, healthcare admin, and business services have seen fast, deep adoption of automation and AI-driven data capture tools. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI tools significantly assist human data entry keyers by flagging suspect records, auto-sorting data, and highlighting mismatches before manual verification, substantially increasing the volume and accuracy a human can handle per hour while they remain responsible for final judgment on ambiguous cases. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up sorting, flagging inconsistencies, and pre-validating data even where a human remains in the loop for exceptions and final sign-off. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can automatically sort, classify, and verify data accuracy using pattern recognition and rule-based checks with high speed; however, the compilation phase and handling of ambiguous/malformed entries may still require human judgment, achieving meaningful time savings well above 50% on routine data verification tasks. |
| Task automatability | claude-sonnet-5 | 4/5 | OCR, intelligent document processing, and validation rules can compile, sort, and check data accuracy for most structured/semi-structured inputs, meeting the 50% time-saving bar in many contexts, though edge cases still need human review. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | Data entry and verification have minimal regulatory or licensing barriers; there is no legal requirement for a licensed human to perform these tasks, and organizational adoption is purely economic. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or human-contact requirements attach to data compilation and verification tasks. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based data verification and sorting is substantially cheaper per task than human data entry keyers when measured across inference, integration, and light oversight; automated systems process orders of magnitude more records per dollar than human operators. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated data capture and validation software costs a fraction of a cent per record compared to loaded wages for manual keyers, especially at volume. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products exist in data validation, quality assurance, and automated verification workflows (e.g., RPA tools, ML-based anomaly detection, ETL platforms) that perform these tasks reliably in production; performance is strong for structured data, though edge cases and novel formats may require human oversight. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Mature products (IDP platforms, RPA with validation, ETL tools with anomaly detection) are deployed at scale in production for data compilation and verification, though error rates persist on messy or unstructured sources. |
Locate and correct data entry errors, or report them to supervisors.
78CI 75–81 · exposure 75 · augmentation 88 · importance 4.8/5 · click for rater detail
Locate and correct data entry errors, or report them to supervisors.
78| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Data-driven organizations in finance, healthcare, e-commerce, and logistics have aggressively adopted automated validation and data quality tools; this is a high-digitization domain with rapid, broad AI/RPA uptake. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Data quality automation is heavily adopted in finance, IT, and back-office administrative functions where digitization is already high, though smaller organizations lag. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered error detection and highlighting greatly assist human data entry staff by surfacing candidates for review and correction, raising productivity even in workflows where humans retain final judgment. The technology is widely used in this augmentative mode today. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI significantly enhances human error-detection by flagging anomalies and suggesting corrections, letting keyers focus on verification and edge cases. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can reliably detect many types of data entry errors (typos, format violations, logic inconsistencies) and flag or correct them automatically, achieving significant time savings. However, some context-dependent or ambiguous errors may still require human judgment, preventing a perfect 5. |
| Task automatability | claude-sonnet-5 | 4/5 | AI-based validation, OCR reconciliation, and anomaly detection can identify and often auto-correct common data entry errors with substantial time savings, though some ambiguous cases still need human judgment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers prevent automation; the task is largely procedural. Some organizations may require human sign-off on high-stakes corrections, but this oversight layer does not meaningfully block deployment of the underlying detection automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates human error correction; main friction is organizational trust in automated systems and need for oversight on ambiguous cases. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated error detection via APIs or embedded validation logic costs pennies per record compared to human keyers earning $15–25/hour to perform the same scanning task, representing an order-of-magnitude savings. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated validation rules and AI-based anomaly detection run at a fraction of the cost of manual error-checking per record, though initial setup and occasional human review add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (data validation tools, RPA platforms, LLM-based checkers) demonstrably perform error detection and correction in production environments at scale. Minor limitations exist around novel error types and edge cases, but the core capability is mature and reliable. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Data validation and error-flagging tools are widely deployed in production ETL and data-quality pipelines, though full autonomous correction without human review is less common for ambiguous or context-dependent errors. |
Select materials needed to complete work assignments.
59CI 39–79 · exposure 50 · augmentation 63 · importance 4.4/5 · click for rater detail
Select materials needed to complete work assignments.
59| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Retail, manufacturing, logistics, and healthcare supply chain operations are actively deploying automated material selection as part of ERP and inventory management systems; adoption is rapid in digitized sectors. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Data entry as an occupation is undergoing gradual automation but this specific preparatory task lags behind more visible AI adoption in the broader clerical/administrative sector. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assists keyers by suggesting or auto-populating materials based on work type and history, significantly reducing manual lookup and decision time while the keyer retains oversight and can override selections. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-driven search, document classification, and file retrieval tools can meaningfully speed up locating and selecting the right materials, assisting but not replacing human judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can reliably identify and select required materials by analyzing work assignment specifications, templates, and inventory systems, achieving substantial time savings. A small residual human check for unusual cases keeps this below 5, but the core task is highly automatable. |
| Task automatability | claude-sonnet-5 | 2/5 | Selecting the right materials/files/forms for a data entry job often requires contextual judgment about ambiguous or non-standardized source documents that current AI handles inconsistently without significant workflow-specific setup.rules. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or authorization barriers exist for AI-driven material selection. Some organizations may require human sign-off for cost or compliance reasons, but no legal requirement mandates human involvement in this selection task itself. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human select materials, though organizational workflows and file/system access permissions create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated material selection via database queries and simple logic is negligible in cost compared to a data entry keyer's loaded wage; integration and oversight are minimal once configured. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Where automation is built, per-task cost is low, but the integration effort to reliably identify correct materials for varied assignments raises effective cost closer to parity in many settings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Production systems already use rule-based and ML-driven inventory selection in order management and warehousing; ERPs and specialized procurement tools perform this at scale. Minor edge cases and exceptions prevent a perfect 5, but deployment is widespread. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some document-routing and file-retrieval tools exist, but no mature off-the-shelf product reliably performs the material-selection step across varied data entry workflows in production. |
Resolve garbled or indecipherable messages, using cryptographic procedures and equipment.
20CI 18–23 · exposure 20 · augmentation 38 · importance 3.4/5 · click for rater detail
Resolve garbled or indecipherable messages, using cryptographic procedures and equipment.
20| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Cryptographic services and message resolution occur primarily in highly regulated sectors (government, defense, intelligence) with strict security protocols and low digitization pressure for automation, resulting in laggard adoption of AI-driven replacement. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | This is a narrow, legacy task within data entry occupations tied to specific communications systems, with little evidence of AI adoption in this specific niche. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist human cryptographers by surfacing pattern anomalies, suggesting decryption strategies, or flagging likely valid outputs, thereby raising human productivity on individual cases without removing the human from the loop. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could assist with pattern recognition or partial decryption support, but the specialized cryptographic equipment and procedures limit general-purpose AI usefulness here. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with pattern recognition in garbled text, the task requires domain-specific cryptographic knowledge, context understanding, and human judgment to determine when decryption is successful and valid. Current AI cannot reliably resolve indecipherable messages end-to-end without significant human oversight and manual intervention. |
| Task automatability | claude-sonnet-5 | 2/5 | Cryptographic decoding of garbled/indecipherable messages requires specialized procedures and equipment tied to specific coding systems, which is not a generic text task current LLMs handle end-to-end reliably.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Cryptographic work is heavily regulated and often requires clearance, licensing, or authorization in government and defense contexts. Legal liability for incorrect decryption, data integrity requirements, and organizational controls create strong barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Cryptographic work often involves security clearances, chain-of-custody, and regulatory/organizational controls that restrict who and what can perform decoding tasks. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference for pattern analysis is cheap, but the integration with cryptographic equipment, validation workflows, and required human expertise make the total cost comparable to or higher than a trained keyer handling the task with traditional tools. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Specialized cryptographic tools and oversight requirements likely keep AI costs comparable to or higher than human specialists trained in these narrow procedures. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably handles cryptographic message recovery as a standalone task. AI tools can aid in frequency analysis or pattern matching, but operational systems still require human cryptographers to validate attempts and apply context-specific decryption procedures. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No mainstream deployed AI product performs cryptographic message resolution as a production service for data entry roles; this remains a niche, often classified or specialized function. |
Load machines with required input or output media, such as paper, cards, disks, tape, or Braille media.
19CI 15–24 · exposure 8 · augmentation 0 · importance 3.5/5 · click for rater detail
Load machines with required input or output media, such as paper, cards, disks, tape, or Braille media.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Data entry is a declining occupation in low-digitization environments; adoption of automation for physical media loading remains minimal because organizations are already reducing these roles or consolidating onto digital workflows, not investing in robotic solutions. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Data entry is a low-digitization physical task and there is no evidence of robotic automation being deployed at scale for media loading in this occupation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 1/5 | The task is purely mechanical media handling with no decision-making or judgment component; AI systems offer no meaningful assistance capability to a human performing the loading operation itself. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI provides no meaningful assistance for the physical act of loading media into machines. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Loading physical media into machines requires dexterous manipulation and spatial reasoning in unstructured environments. Current AI systems (including robotics) lack reliable generalization across diverse media types, machine configurations, and real-world variability, making the 50% time-saving threshold unattainable for the full task without significant setup. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical manipulation task requiring hands to load physical media into machines, which current AI systems (software-based) cannot perform without robotic embodiment.: |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | The task involves minimal regulatory licensing and human-contact requirements. The primary friction is technical and economic (automation difficulty and cost) rather than legal or organizational barriers, which are relatively low. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or regulatory barriers exist, but the physical nature of the task creates a practical barrier since it requires embodied manipulation rather than software automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Specialized robotic systems capable of media loading remain capital-intensive and require significant integration; the total cost per task-equivalent (hardware, maintenance, oversight) substantially exceeds the modest loaded wage of a data entry keyer. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Robotic solutions for this narrow physical task would require expensive specialized hardware, far exceeding the cost of a human performing this simple manual action. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | While research-stage robotic systems can perform narrow loading tasks in controlled settings, no deployed commercial products reliably perform this task end-to-end at production scale across the variety of media and machines described. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product performs physical loading of paper, cards, disks, or Braille media into machines; this remains a manual physical task. |
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.