Data entry clerks
An AI agent can at least partly carry out 93.0% of the task time of data entry clerks, and finish 38.6% alone.
42nd of 433 · interval 75.5–100.0
7th of 433 · interval 0.0–68.5
7th of 433 · interval 56.5–82.3
task statements rated by two models and weighted by importance and frequency. World employment: 2.4 million.
How the tasks rate
What stops the agent
Share of task weight flagged by each barrier; a task can carry several.
| Task statement (O*NET occupation) | Sonnet · Haiku | Barriers | ||||
|---|---|---|---|---|---|---|
| Locate and correct data entry errors, or report them to supervisors.Data Entry Keyers | 13.0% | L2 | L3L2 | 78% | 25 min | Access to systems |
| Compile, sort, and verify the accuracy of data before it is entered.Data Entry Keyers | 12.9% | L3 | L3L3 | 82% | 42 min | Access to systems |
| Compare data with source documents, or re-enter data in verification format to detect errors.Data Entry Keyers | 12.8% | L3 | L3L3 | 88% | 20 min | Access to systems |
| 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.Data Entry Keyers · supplemental task | 12.8% | L3 | L3L3 | 90% | 19 min | Access to systems |
| Store completed documents in appropriate locations.Data Entry Keyers | 12.1% | L2 | L2L3 | 80% | 8 min | Access to systems |
| Select materials needed to complete work assignments.Data Entry Keyers | 11.7% | L2 | L2L2 | 60% | 10 min | Physical presence |
| Maintain logs of activities and completed work.Data Entry Keyers · supplemental task | 11.3% | L2 | L2L3 | 75% | 25 min | Access to systems |
| Load machines with required input or output media, such as paper, cards, disks, tape, or Braille media.Data Entry Keyers · supplemental task | 7.0% | L0 | L0L0 | 3% | 4 min | Physical presence |
| Resolve garbled or indecipherable messages, using cryptographic procedures and equipment.Data Entry Keyers · supplemental task | 6.4% | L2 | L2L2 | 52% | 38 min | Physical presence, Access to systems |
Level: consensus of the two models (the lower level when they disagree). Share: part of the task’s working time an agent could take over. Human time: minutes a competent person needs for one instance, as estimated by the models.
Exposure is not delegation
| Measure | Value | Percentile | Pct |
|---|---|---|---|
| ADI (this index) | 93.0% | 91 | |
| Eloundou et al. (2024), GPT exposure β | 0.866 | 99 | |
| Felten et al. (2021), AI occupational exposure | 0.311 | 60 | |
| Webb (2020), AI patent exposure | 94 | 94 | |
| Eisfeldt et al. (2026), generative AI exposure | 0.722 | 98 | |
| Anthropic Economic Index, observed use | 0.671 | 100 |
Earlier indices measure whether AI can touch a task; ADI measures how much of it an agent can carry out from a goal. – : the source does not cover this occupation.
Neighbouring occupations
| Occupation | ADI | % |
|---|---|---|
| Typists and word processing operators 4131 | 96.0 | |
| Secretaries (general) 4120 | 89.6 | |
| General office clerks 4110 | 84.9 |
Dashed line: data entry clerks (93.0%).
Espinal Maya, Cristian, and Santiago Jiménez Londoño. 2026. ADI v2026.1: Agent Delegability Index data, code and MCP server. Zenodo. https://doi.org/10.5281/zenodo.23552918. Record: Data entry clerks (ISCO-08 4132), https://balamlabs.org/adi/occupations/4132
@misc{espinal2026adidata,
title = {{ADI} v2026.1: Agent Delegability Index data, code and {MCP} server},
author = {Espinal Maya, Cristian and Jim{\'e}nez Londo{\~n}o, Santiago},
year = {2026},
publisher = {Zenodo},
version = {v2026.1},
doi = {10.5281/zenodo.23552918},
url = {https://doi.org/10.5281/zenodo.23552918},
note = {Record: Data entry clerks (ISCO-08 4132)}
}The same record as static JSON, with its citation: /adi/v2026.1/isco/4132.json.