← Back to full report

What Is Accounting and Finance Research For in the AI Era?

One-Page Summary

Dr Yuqian Zhang · October 2026

What This Report Is About

This report uses journal, citation, review and policy evidence to examine what accounting and finance research should do when AI makes writing, coding and synthesis cheaper. It shows where value shifts from output volume to problem choice, difficult data, verification and accountable judgement.

Key Findings

Output grew, but production is not the main constraint. Crossref-indexed research article output across the selected journals rose 27.9% from 2015 to 2025. The two journals with public submission records did not show one continuous AI-era surge, and a 2026 study of accounting papers found AI-associated vocabulary in abstracts up about 75% after 2023.
Attention and evaluation remain selective. After cleaning and matching the article cohort, the pooled share of 2020-2022 articles with no citations in their first three years was 1.0%, and journal differences were too small to defend. Observed median times to an editorial decision ranged from 30 to 52 days.
Value shifts toward evidence and judgement. Problem selection, difficult data and context, verification, accountability and decision use become more valuable as drafting and coding become easier.

Key Statistics

918 Crossref-indexed research articles in the selected journals in 2025
27.9% increase in indexed research article output from 2015 to 2025
1,227 total submissions to The Journal of Finance in 2015
1,066 total submissions to The Journal of Finance in 2024
1,664 new and dual submissions to The Review of Financial Studies in its latest period
1.0% pooled zero-citation share after cleaning (26 of 2,578 matched articles)
75% rise in AI-associated words in accounting abstracts after 2023
6.4% of 1,226 editorial decisions at The Journal of Finance in 2023
43% desk-rejection rate at The Journal of Finance in 2023
8 of 11 latest available official journal pages showed explicit AI guidance

How to Prepare

Develop a small number of defensible problem agendas. Build access to difficult data or settings. Learn research design, measurement, causal inference, model evaluation and computational verification. Keep an audit trail from data to claim.

Use AI to test, challenge and implement ideas. Retain human ownership of problem choice, assumptions, interpretation and accountability. Workflow optimisation matters when it reduces error and time to a defensible answer, not when it only increases output.

Why It Matters

“Publish or perish” is likely to persist as a threshold and weaken as a measure of capability. A stronger evaluation model would show contribution across problem importance, evidence strength, provenance and reproducibility, decision use, maintained value and scholarly stewardship. This brief presents that portfolio as a design proposal, with its own gaming and administrative risks.