Unit 5: Artificial Intelligence and Analytics - Practice Quiz

1 What does AI stand for in the context of finance and accounting transformation?

2 Which of the following is a common repetitive accounting task that AI can automate?

3 What is RPA commonly used for in accounting?

4 How does AI primarily help in fraud detection?

5 Which AI capability allows software to learn from data and improve over time without being explicitly programmed?

6 What is a key benefit of using AI for financial forecasting?

7 Which term describes AI that can understand and process human language in documents and queries?

8 In accounting, chatbots powered by AI are mostly used to:

9 Which of the following best describes automation in accounting?

10 How does AI impact the role of accountants?

11 What is a common AI application in the audit process?

12 Which of these is a benefit of AI in finance and accounting?

13 Predictive analytics in finance mainly helps organizations to:

14 Which task is AI least suited to perform on its own?

15 What does data-driven decision making rely on?

16 AI-powered reconciliation tools help accountants by:

17 Which of the following is a risk or challenge of adopting AI in accounting?

18 What is real-time reporting enabled by AI?

19 Which department within an organization benefits from AI-driven expense management by automatically categorizing and flagging spending, reviewing employee claims against policy, and detecting out-of-policy transactions before payment?

20 Overall, AI transformation in finance and accounting is best described as:

21 A mid-sized firm processes thousands of vendor invoices monthly. Its finance team wants to automatically extract invoice fields (amount, date, vendor) from scanned PDFs without manual keying. Which AI technology is most directly suited to this task?

22 An auditor replaces sample-based testing with an AI model that reviews 100% of journal entries to flag anomalies. What is the primary advantage of this shift?

23 A company deploys a rule-based bot to log into a banking portal, download statements, and paste balances into a reconciliation sheet daily. This is best classified as:

24 A credit team builds a model that assigns each loan applicant a default probability using historical repayment data. Which AI application does this represent?

25 A CFO is concerned that an AI credit model may reject applicants unfairly based on patterns tied to protected attributes in the training data. This risk is best described as:

26 An analytics dashboard flags that Q3 marketing spend rose while revenue fell , and asks the user to investigate why. This dashboard is performing which type of analytics?

27 A finance system recommends the optimal mix of short-term investments to maximize yield within liquidity constraints and executes the reallocation automatically. This capability represents:

28 A bank's monitoring system learns each customer's normal transaction pattern and alerts when a sudden large overseas transfer deviates sharply from that pattern. Which technique underlies this alert?

29 An accounting team uses a generative AI tool to draft the narrative section of the management commentary from structured financial figures. What is the main residual responsibility of the accountant here?

30 A firm claims AI has fully automated its month-end close. Which task is least likely to be fully automated and still needs human judgment?

31 A treasury team uses machine learning to forecast daily cash positions. The model performs well on training data but poorly on new data. This symptom most likely indicates:

32 A regulator asks a bank to explain exactly why its AI model denied a specific loan. The bank struggles because the model is a complex neural network. This challenge is known as the:

33 An organization introduces AI chatbots to handle routine expense-policy queries from employees. The most likely immediate benefit for the finance function is:

34 A company wants AI to classify thousands of general-ledger transactions into the correct expense categories using labeled historical examples. This is an example of:

35 A fund manager uses AI to scan thousands of news articles and social posts to gauge market mood toward a stock before trading. Which technique is being applied?

36 As AI automates transactional accounting work, the role of finance professionals is most likely to shift toward:

37 An AI-based continuous-auditing system runs controls tests in near real time throughout the year rather than only at year-end. The primary benefit is:

38 A finance team feeds an AI model incomplete and inconsistently formatted historical data, and the forecasts prove unreliable. This outcome best illustrates the principle of:

39 A firm combines RPA to move data and machine learning to make judgment-based classifications within one workflow. This blended approach is often called:

40 When adopting AI for financial reporting, which governance practice most directly addresses accountability for AI-driven outputs?

41 A finance team deploys an ML model to predict late-paying customers. The model achieves 95% accuracy, but only 4% of invoices are actually late. The controller rejects it as unhelpful. What is the most defensible technical reason?

42 An RPA bot automates a three-way match (PO, receipt, invoice) but the underlying process has unstandardized supplier invoice formats. What is the likely outcome and best remedy?

43 A CFO wants an AI system that both explains why a journal entry was flagged as anomalous and can be audited by regulators. Which approach best balances performance and transparency?

44 An AI cash-flow forecasting model performs well in backtesting but degrades sharply after a major economic regime change. What phenomenon is this, and what is the correct response?

45 A company uses generative AI to draft MD&A narrative from financial data. Which risk most directly threatens the reliability of the reported disclosures?

46 Two audit-analytics models flag expense fraud. Model A: precision 0.9, recall 0.4. Model B: precision 0.6, recall 0.85. An audit firm must minimize missed fraud given limited but adequate review capacity. Which is the better choice and why?

47 An AI-driven continuous auditing system reviews 100% of transactions in real time rather than sampling. What is the most significant methodological shift for the auditor?

48 A bank's AI credit model shows strong overall accuracy but denies loans to a protected group at a disproportionately high rate, even after removing the protected attribute from inputs. What is the most likely cause?

49 A finance transformation program automates reconciliations with AI. After deployment, staff stop reviewing outputs and error rates in edge cases rise. This best illustrates which risk?

50 An NLP model classifies contracts to determine lease vs. service arrangements for ASC 842 / IFRS 16. Which limitation most threatens the accounting conclusion?

51 A team trains a revenue-forecasting model and reports near-perfect validation results. Later they discover a feature that was only available after period-end was included. What error occurred and its consequence?

52 An organization considers replacing rule-based anomaly detection with unsupervised ML for expense monitoring. What is the key tradeoff to communicate to auditors?

53 A CFO wants to quantify ROI of an AI accounts-payable automation. Annual manual cost is , automation reduces it by 70% but adds recurring platform cost and a one-time build. What is the first-year net benefit?

54 A generative AI assistant is integrated into the general ledger to answer ad-hoc queries from managers. Which control is most critical before granting it live database access?

55 An AI model predicts goodwill impairment risk. Management wants to use its output to support the impairment test under IAS 36. What is the core governance concern?

56 A firm's fraud-detection classifier is retrained monthly on newly confirmed fraud cases identified only from prior alerts. Over time recall on unflagged fraud types worsens. What feedback problem is this?

57 During finance transformation, a company migrates to AI-assisted forecasting but keeps legacy spreadsheets as a parallel check. When outputs diverge materially, what is the most sound response?

58 A finance leader claims AI will fully eliminate the need for professional skepticism in audits. Which counterargument is strongest?

59 A model scores 10{,}000 invoices for fraud. It flags 500; of these 100 are true fraud, and 40 true frauds were missed. What are precision and recall respectively?

60 An AI vendor offers a black-box model that improves expense-anomaly detection AUC from 0.82 to 0.91 but provides no feature-level explanations. For a SOX-regulated control, what is the most appropriate stance?