Exposed The task profile has little on the defensive side of the ledger.

Bookkeeping, accounting and auditing clerks

Rule-based verification of numbers, at high volume, in a digitised system.

Horizon: to 2030
Protection from automation: Weakly protected · our reading

The uncomfortable lesson from the ILO’s 2025 re-scoring is that exposure is not fixed. Between the 2023 and 2025 versions, financial analysts, web and multimedia developers and applications programmers moved up the exposure gradients, because models acquired multimodal and agentic capability in the interim.ILO 2025

Which means: a role’s safety is a claim about today’s model capabilities, and that claim has a shelf life.

What holds it up

Reconciliation is pattern matching against rules the machine already holds
High volume and routine, so savings per hour are large

The four tests

What protects this job

Four things that keep a job out of a machine's reach. Score them 0–3 and add them up: the higher the total, the harder the job is to automate.

Needs hands and a body — not at all (not at all)
A person is answerable — a little (a little)
It needs a relationship — a little (a little)
The role can't be split — a little (a little)

Almost none of these hold, so nothing here keeps the work out of a machine’s reach.

Our reading of the job, not a measurement — the evidence is below.

Verified claims

Employers expect a sharp fall in accounting and auditing clerk roles to 2030, alongside cashiers, ticket clerks and administrative assistants. WEF 2025
Accounting and bookkeeping clerks sit in the ILO's highest generative-AI exposure gradient. ILO 2025
The ILO finds strongly digitised occupations have increased in exposure as models gained multimodal and agentic capability. ILO 2025