Why lending files still pass through people
Loan interviewing is two jobs in one. The first is paperwork: collecting application details, pulling credit reports, checking employment history, chasing missing pay stubs and making sure a file is complete before it reaches an underwriter. The second is dealing with the person attached to the file. Software is good at the first half and weak at the second.
That is the honest answer to the question of will AI replace loan interviewers: the structured steps are moving to software faster than the conversations are. Document intake, data entry into a loan system and verification checks follow rules, so they automate well. Explaining why an application stalled, coaching an applicant with irregular income through what to send next, or spotting that a borrower has misunderstood the terms is slower, messier work.
The scale matters too. About 164,790 people hold this job in the US, with median pay of $50,020 (BLS, 2025), and BLS projects employment falling 2.4% between 2025 and 2035. That is erosion, not disappearance. Lenders keep processing loans; they just need fewer hands per file, and the hands they keep do more of the judgment work. You can see how this sits against other roles in the same work area on the information and record clerks family page.
What software handles, what it assists, and what stays with staff
Start with the automated share. 11% of task time sits in work AI can already run with little supervision: pulling and formatting data from submitted documents, and recording application details into the lender’s system. These steps have a clear right answer and a clear source document, which is why they went first.
Assisted work is the larger part of the day. 78% of task time is work where a tool speeds a person up but does not finish the job: reviewing a file for completeness, and comparing stated income or employment against the documents on record. The tool flags a gap; a clerk decides whether the gap is a typo, a self-employment quirk or a reason to pause. The method behind that split is set out in how coverage is measured.
Then the part that still needs a person. 11% of task time covers interviewing applicants about their circumstances and explaining loan terms, conditions and decisions in language the borrower understands. These tasks carry consequences, emotion and compliance exposure at the same time, which is a hard combination to hand over.
Good to know: this job needs no robots at all, so adoption depends only on software and lender IT budgets.
How strong the evidence is right now
The quality evidence grade for this occupation is D. In plain terms, no published study has tested an AI system against a working loan interviewer on this job’s actual tasks and scored the results, so no parity number is given here. That is a gap in the record, not a quiet verdict either way.
What would settle it is specific: a controlled comparison where an AI system and experienced clerks handle the same real application files, measured on completeness of the submitted package, error rate on verified data, rework sent back by underwriters, and applicant complaints or appeals. Lender-side data on how many files clear first time with and without automation would be strong evidence as well. Until something like that is published, the honest position is uncertainty, scored and dated rather than guessed. The grading scale is explained in the quality parity method.
When the picture could shift
Most likely between 2035 and 2046 (8 in 10 of our scenarios). For what that window measures and how it is produced, see the replacement-year method.
Two things could pull the change earlier. First, cost: running software on routine file handling is cheap next to staffing it, and the cost panel above shows the gap clearly. Second, physical work is not a barrier here, since no part of the job needs a machine on a desk or in a branch; it is all screens, documents and calls.
Two things hold it back. Lending is regulated, and adverse-action notices, fair-lending rules and audit trails make lenders cautious about automated decisions they cannot explain. And applicants are not clean inputs. Self-employment, gig income, thin credit files and documents that contradict each other still get routed to a person, because the cost of getting one wrong lands on the lender.
How to stay needed as a loan clerk
Lean into the tasks the tools do not finish. Interviewing applicants and drawing out what their paperwork does not say. Explaining terms, conditions and conditional approvals so a borrower actually understands them. Handling the awkward files: inconsistent income, missing history, applications that need a second route rather than a rejection.
Two skills raise your floor. One is working fluency with the lender’s origination and document systems, including checking what an automated flag is actually claiming before you act on it. The other is compliance literacy: knowing which decisions need a documented human reason, and writing that reason clearly.
If you want a nearby move, compare the work of loan officers, new accounts clerks and interviewers, except eligibility and loan. You can put any two of them side by side on the job comparison tool, see how lending roles sit together in the banking sector, or check where clerical roles land on the most exposed jobs list. The full scoring approach is on the methodology page.