Using Artificial Intelligence to Solve Accounting Problems: A Comparative Study of Microsoft Copilot and Grok

This study compares the performance of two artificial intelligence assistants—Microsoft Copilot and Grok—in generating the correct journal entry for a stock acquisition involving unrecorded goodwill. Using a homework exercise from an Intermediate Accounting course, the author prompted both AIs with identical details: Wolfson Corporation acquires all outstanding common stock of Barney Corporation for $17 million cash, where Barney’s book values are $18 million in assets and $10 million in liabilities, and fair values are $23 million and $9.5 million, respectively. Both AIs initially produced incorrect entries by treating the transaction as a direct asset acquisition under the acquisition method (ASC 805), recording individual assets and liabilities at fair value along with $3.5 million in goodwill on Wolfson’s standalone books. Upon correction that the acquisition involved stock (not assets), both AIs accurately revised the entry to debit Investment in Barney Corporation $17 million and credit Cash $17 million, noting that fair value adjustments and goodwill appear only in consolidated financial statements. The results highlight a common AI limitation in distinguishing stock from asset acquisitions without explicit prompting, underscoring the need for human oversight in technical accounting applications.