AI在會計核算任務中超越會計師的速度和準確性
ai模型在結構化會計核算任務中擊敗持牌會計師，取得近乎完美的結果，但仍需要監督來完成核算。
a study by mercor compared the performance of 12 licensed cpAs with an average of 5.5 years of experience to ai models on simplified tasks from the apex accounting benchmark. the study found that 18 months ago, the best ai models scored below the accountants' average of 37 percent, but today they solve the same tasks almost flawlessly. claude opus 5.5, fable 5.1, and gpt-6 astra achieved 61.8 percent, 61.0 percent, and 57.9 percent of grading criteria met, respectively. however, mercor notes that the study's tasks test exactly what ai does best and left out key parts of the job, such as client communication and context building.