Mercor Study: AI Outperforms Certified Accountants on Accounting Tasks
Understand the story
A study by Mercor used the APEX Accounting Benchmark, having 12 licensed accountants with an average of 5.5 years of experience complete simplified tasks from the benchmark, and compared their performance with AI models. Results show that today's AI models outperform these licensed accountants in speed, accuracy, and cost. However, AI still cannot independently complete book closings without supervision. The study also notes that this gap is recent: 18 months ago, the best AI model still scored about 37% below the accountants' average on the benchmark. This indicates that AI's capability in accounting tasks has improved dramatically during this period, going from lagging behind humans to surpassing them in speed, accuracy, and cost, though it still falls short of fully completing complex accounting work unsupervised.
AI overview of reporting · Updated 16 days ago
Reporting timeline
Explore different aspects of the story through its reporting.
- The DecoderAI beats licensed accountants on speed and accuracy, but still can't close the books without supervision
Mercor's research had 12 licensed accountants with an average of five and a half years of experience complete simplified tasks from the APEX Accounting Benchmark; results showed today's AI models outperform accountants on speed, accuracy, and cost, whereas 18 months ago the best models scored about 37% below the accountants' average.