Gains:
- Ability to recognize and verify that AI may produce fictitious figures, sources and legislation
- Ability to manage the use of AI within the limits of financial data privacy, KVKK and corporate policy
- Understanding that the ultimate responsibility for auditing, regulatory compliance and ethics remains with the finance professional
Module Exam
1. When working with an AI tool as a financial analyst, the ultimate responsibility for which of the following should always remain with the human?
- A) Final approval and signature of a loan allocation decision ✔
- B) Creating a first draft of an income statement summary
- C) Editing the title and axis label of a chart
- D) Preparing a glossary of financial terms for a report
Description: AI; It can speed up tasks such as table summarization, formula generation, and report drafting. However, the accuracy of an investment, loan or financing decision and its reflection in an official document as a signature are the responsibility of the finance professional; this decision cannot be delegated to AI without independent verification.
2. You had the AI interpret a company's cash flow statement and it cited an 'investment income' item that was not in the table. What is the first correct action to take?
- A) Add the comment to the report as is because AI seems confident
- B) Compare each item on which the comment is based with the original table and reject the item that is not in the source ✔
- C) Adding that item to the table later
- D) Just change the name of the item
Explanation: Language models can convincingly fabricate items and numbers that do not exist in the source (hallucination). It is imperative to compare each item on which the interpretation is based exactly with the original table and not to accept any number that is not in the source.
3. AI calculated current ratio for a company as 'current assets/total assets'. What is the main problem with this output?
- A) It is unnecessary because the rate will be too high
- B) The denominator of the current ratio should be short-term foreign resources, not total assets. ✔
- C) Current ratio is calculated only for banks
- D) The result is incorrect because it is a decimal instead of a percentage
Explanation: Current ratio measures short-term solvency and its definition is 'current assets / short-term liabilities'. AI got the denominator wrong. In ratio analysis, verifying the definition of each ratio is the first and mandatory step; The wrong denominator produces a completely misleading result.
4. What is the most valuable output of an AI-powered early warning logic when setting up a weekly cash flow projection?
- A) Summarizing the average expenses of previous months in a single line
- B) Marking in advance the week in which the cash balance will fall below the critical threshold ✔
- C) Formatting the table more colorfully
- D) Converting all numbers to percentages
Explanation: The main purpose of cash management is to foresee the week when the cash balance will fall below the critical threshold. Thus, actions such as financing, collection acceleration or payment postponement are taken in a timely manner. It is valuable not only to summarize the past but also to warn about the future.
5. A circular reference warning appeared in the formula created by AI in a financial model. Which is the most correct approach?
- A) Turn off the warning and continue the model
- B) Review the formula chain, find the source of the loop and re-establish the logic ✔
- C) Delete the relevant cell and write a constant number instead
- D) Opening the model completely from scratch in another file
Explanation: Circular reference is the direct or indirect connection of a cell to itself and usually occurs in items that feed each other, such as interest and debt. Ignoring the problem produces the wrong result; It is necessary to review the logic and use the iteration setting consciously or re-establish the formula chain.
6. When valuing DCF (discounted cash flow) with AI you want to see which assumption the result is most sensitive to. What do you do?
- A) Changing the discount rate and terminal growth in the range and establishing a sensitivity table ✔
- B) Produce and finish a single result with a single discount rate
- C) Only averaging the profits of previous years
- D) Reporting the result with the highest assumption
Explanation: The DCF result is particularly sensitive to the discount rate and terminal growth rate. Changing these assumptions within a reasonable range and establishing a sensitivity table showing how the valuation result changes reveals how fragile the result is.
7. When you ask the AI about a portfolio, it produces a definitive recommendation that says 'you should buy this stock'. Which is the correct assessment?
- A) Trusting the recommendation because AI has access to up-to-date data
- B) Considering definitive buy/sell advice unreliable and treating it as analysis based solely on data and assumptions ✔
- C) Delivering the recommendation directly to the customer
- D) Reconstructing the entire portfolio based on this single output
Disclosure: AI is not an authorized advisor providing personalized investment advice; It cannot promise definite returns for the future, and the directional statements it gives are outputs that are limited to data and assumptions and need to be audited. Definitive buy/sell advice is an unreliable and responsible area in terms of risk profile and legislation.
8. You audit AI-generated credit risk scoring in a risk report. Which control is most critical?
- A) Checking the font and color compatibility of the report
- B) Validate the inputs, data quality, and assumptions/biases the score uses ✔
- C) Adjusting the score just to make it go higher
- D) Approve the report without reading it at all
Explanation: Risk scoring is completely dependent on the inputs it uses and the assumptions/biases it contains. Verifying which variables are used, whether there is missing/erroneous data, and whether the logic of the score is explainable is essential to a fair and defensible risk decision.
9. You predicted sales revenue with AI. What is the most appropriate method to test the reasonableness of the estimate?
- A) Applying the forecast to a past period and comparing it with the actual result (backtest) ✔
- B) Accept the estimate as it is and write it into the annual budget
- C) Taking the highest growth scenario as the only real number
- D) Just beautify the graph of the prediction
Explanation: Backtesting is applying the forecasting method to a past period and comparing it with the actual result. Seeing how accurate the method has been in the past is the most concrete way to understand how reliable the future prediction can be.
10. A draft management report produced by AI contains figures that are fluid but of uncertain origin. What is the correct behavior?
- A) Presenting the report as it is because the text is fluent
- B) Match each figure with the original data source and remove the unconfirmed ✔
- C) Rounding and glossing over numbers
- D) Just read the executive summary and skip the rest
Description: Each figure in the report must be traceable to the original data source (table, system, account). Even if it is a number in a fluent sentence, no number whose source has not been verified should be presented to management; Otherwise, a fictitious number may enter the decision basis.
11. What is the best step before using an AI-generated Excel formula (e.g. nested IF and VLOOKUP)?
- A) Applying the formula directly to the entire live dataset
- B) Run the formula on a small test data with known results and try edge cases ✔
- C) Deleting part of the formula to shorten it
- D) Just looking at the length of the formula
Explanation: Complex formulas may appear correct but produce incorrect results. Running the formula on a small test data whose result you know manually, comparing it to the expected output, and trying edge cases such as null/zero/unmatched will catch the error before it carries over to the live data.
12. You want to paste your company's unaudited quarter results into a public AI tool and request analysis. Which is the most correct approach?
- A) Checking the institutional policy and anonymizing the data or using a secure/institutional tool ✔
- B) Pasting the data as is for speed
- C) It is enough to just delete the company name and share all the remaining numbers.
- D) Sharing the data and then asking the AI to delete it
Description: Sensitive financial data that is not publicly disclosed; It is risky in terms of privacy, KVKK and insider trading. Before sharing data, it is necessary to check corporate policy, anonymize it if possible, or use a corporate/non-data sharing tool.
13. The AI gave an exact rate for a tax imposition with reference to a particular article of law. What is the correct engineering/finance behavior?
- A) Applying the rate directly because the AI is up to date
- B) Verifying the rate and article number from official, current legislation source ✔
- C) Raising the rate randomly to stay on the safe side
- D) Adding the item number to the report without checking it at all
Explanation: Language models may misremember or make up legislation article numbers and proportions. In binding areas such as tax and legislation, every value must be verified verbatim from official and current sources (legislative text, administrative notification, financial advisor); The AI output only shows where to look.
14. Which input approach is right to get the most benefit from AI in budget-to-variance analysis?
- A) Giving only realization figures and skipping the budget
- B) Only give the total turnover as a single line
- C) Presenting the budget and realization on an item basis, with period information ✔
- D) Asking for deviation interpretation only without giving the numbers
Explanation: For a meaningful deviation analysis, it is necessary to provide AI with both budget and realization figures on an item basis and with period information. AI then captures deviations and can calculate them as percentages and break them down into categories such as price, volume, and timing. Incomplete or contextless data produces superficial interpretation.