Will AI replace data entry jobs? What the evidence shows

Reviewed by Colin Balfe on Oct 27, 2026 · Updated Oct 27, 2026 · Graded, dated evidence, not predictions
The short answer

This is one of the occupations BLS expects to shrink fastest, by about a quarter by 2035. Modern AI models pull fields from clean invoices with about 96% accuracy in tests, and scanning has replaced some manual keying. But big paper workflows are slow to convert: the IRS digitized only a small fraction of its planned paper returns in 2025, and errors on tricky fields still need a human.

The job today

131,800 (2025)people employed in the USBLS Employment Projections
Outlook-25.5% (-33,600 jobs), 2025-35, the fourth-fastest projected decline of any occupation. BLS has no Occupational Outlook Handbook profile for this job, and the projections table we could read does not state a reason.BLS Employment Projections

Task by task: what AI can actually do

In real use means deployed at scale; shown in tests or pilots means demonstrated but not yet routine; not yet means we found no credible evidence it does this reliably.

TaskStatusEvidence
Keying data from paper formsIn real useThe IRS scanned over 410,000 paper Forms 940 in its 2023 Digital Intake effort, reducing the digit-by-digit manual keying it relied on during the pandemic backlog. (Thomson Reuters Tax, Apr 2023)
Pulling fields from invoicesShown in tests or pilotsFraunhofer researchers found Gemini 2.5 Pro extracted invoice fields with 96.5% accuracy on clean invoices and 92.7% on scanned ones, zero-shot; the open datasets may have been in training data. (arXiv (Fraunhofer IAIS), Aug 2025)
Reading receipts and error-prone fields without checkingNot yetIn the same benchmark the best model reached only 87.5% on scanned receipts, and bank account numbers (IBANs) were consistently the hardest field, with errors like confusing zero and 'O'. (arXiv (Fraunhofer IAIS), Aug 2025)
Converting a whole paper workflow to automated captureNot yetTIGTA found IRS contractors had scanned 517,000 forms by May 2025, about 5% of the plan to digitize most paper-filed returns that season. (Accounting Today, Feb 2026)

Jobs evidence so far

Labelled by who says AI was the reason: the company itself, reporters, or a study.

  • IBM CEO Arvind Krishna says IBM will slow or pause hiring for back-office roles, saying 'I could easily see 30% of that getting replaced by AI and automation over a five-year period' (about 7,800 roles reported) (company says AI) (Neowin, May 2023)
  • BLS projects data entry keyers to fall 25.5%, behind only word processors and typists, telephone operators and switchboard operators among the fastest-declining occupations (study) (BLS Employment Projections)
  • Pantheon Macroeconomics names office administration among sectors slowly shrinking as AI costs the US 10,000+ jobs a month, net (study) (Fortune, Oct 2026)

What to do

  1. Move from keying to checkingAI extraction is good but not perfect, especially on scans and account numbers. Exception handling and quality control are the parts that remain.
  2. Learn the capture toolsDocument-processing and spreadsheet automation skills make you the person who runs the system rather than the one it replaces.
  3. Build toward an adjacent roleBookkeeping, records management, payroll or operations roles use the same accuracy skills with more judgment.
  4. Don't wait for your employerBLS expects about a quarter of these jobs to go by 2035. Start retraining while you still have a paycheck.

Questions people ask

Will AI replace data entry jobs?

Many of them. BLS projects a 25.5% decline (-33,600 jobs) for 2025-35, one of the steepest of any occupation, though it does not say how much of that is AI versus older automation.

How accurate is AI at data entry?

In a 2025 benchmark the best model got about 96% of invoice fields right on clean documents and 87% on scanned receipts, so humans still check the output.

Why hasn't automation happened faster?

Large paper workflows are hard to convert. The IRS reached only about 5% of its 2025 digitization goal, according to its inspector general.

Which companies have tied back-office cuts to AI?

IBM said in 2023 it would slow hiring for back-office roles AI could do. Most other data entry losses are not announced as such.

How we grade

Each task is graded 'deployed' only when a named organisation reports AI doing it in real work at scale, 'demonstrated' when it works in tests, benchmarks or pilots, and 'not yet' when we found no credible evidence AI does it reliably. For job losses we separate what a company said in its own words ('company says AI') from what reporters or analysts attributed to AI ('reported as AI') and from research ('study'); employment and outlook figures come from the BLS 2025-35 projections.

Sources

Spot an error or something out of date? Tell us; corrections are made and logged. Written with AI assistance from public sources; every fact checked and approved by Colin Balfe. General information, not legal, medical or financial advice.