Gains:
- Ability to use artificial intelligence as a risk idea generator and drafter in the process of identifying the risk of material misstatement by knowing the business and its environment
- Ability to script materiality and insignificance threshold calculations with artificial intelligence support and independently verify each number
- Ability to maintain that the final risk assessment and materiality judgment belongs to the auditor's professional judgment and that artificial intelligence only provides input
Good auditing is won or lost at the desk, long before fieldwork. If an auditor starts without knowing the business, without guessing where there is a high probability of going wrong, and without directing his resources (time, team, tests) towards these risks, he may look in the wrong places, no matter how hard he works. Therefore, BDS 315 (Standard for determining and assessing the risk of material misstatement by knowing the business and its environment) and BDS 320 (Materiality standard in auditing) establish the framework of the audit.
In this unit, we will cover how you can apply AI to the two main tasks of the planning phase - materiality calculation and risk mapping - as an idea partner and draft generator, but why you will always be the final judge. Let's clarify two concepts first. A material misstatement is an error or fraud in the financial statements that, individually or collectively, is large enough to affect the economic decisions of users of the statements. Risk assessment is an effort to determine in which account, class of transactions or disclosure, and for what reason, such a misstatement is most likely to occur.
Getting to know the business: Where AI becomes a thought partner
ISA 315 asks the auditor to know the entity and the environment in which it operates (industry, legislation, business model, internal control). During this recognition process, the auditor asks “what typically goes wrong in this line of business?” he asks. This is where AI is a valuable brainstorming partner: it can quickly list an industry's typical risk areas, common patterns of accounting fraud, seasonality and revenue recognition intricacies. But beware: this list given by AI is general and generic; does not recognize your business. So the output is a "checklist draft", not a "risk assessment".
Let's set up the process step by step:
- Give the context. Anonymously describe the business' industry, size, business model and notable features. (No name, TIN.)
- Request a generic risk idea. “Where is the risk of material misstatement typically high in this type of business?” ask. AI gives you a starting map.
- Customize to business. Filter the AI's list with facts you know (this business's history, internal controls, prior year findings). Some risks do not apply to this business; Some are unique risks that the AI will never see.
- Tie it with evidence. Link each risk to which test you will address in the field study.
Tip: Think of AI as “where might there be a risk of material misstatement?” Use it to make you think; Not to make you say "this is where the risk is, for sure". Risk assessment is professional judgment; risk idea brainstorming.
Materiality: number, but number based on judgment
ISA 320 requires the auditor to determine a materiality level when beginning the audit. In practice, this starts with choosing an indicator (e.g. tax before profit, revenue, total assets, or equity) and applying a percentage to it. For example, 5% of pre-tax profit from continuing operations, or 0.5-1% of revenue. This is not a mechanical formula: the auditor decides which indicator is most meaningful to the business, what the percentage will be and the final amount. Performance materiality (the lower threshold set below materiality to tighten testing during planning) and clearly unimportant threshold (the limit of differences too small to be reported) are also determined.
AI is an excellent calculator and scenario engine here: it instantly calculates the amounts for different indicators and percentages, puts them in front of you in a table. But the choice of indicator and the final judgment is yours. The following table shows a typical scenario output (numbers are examples):
indicator
Amount (TL)
% Applied
Materiality (TL)
profit before tax
12,000,000
5%
600,000
revenue
240,000,000
0.5%
1,200,000
Total assets
180,000,000
1%
1,800,000
equity
90,000,000
2%
1,800,000
Looking at this table, the auditor thinks: "If profits are unstable, a revenue-based materiality may be more reasonable; but if the main interest of users in this business is profitability, pre-tax profit is more appropriate." This assessment is professional judgment that AI cannot replace. AI calculates four scenarios in 5 seconds; You decide which one is correct.
Caution: Verify the materiality scenario given by the AI. Redo the percentage calculation yourself (is 600,000 really 5% of 12,000,000?) and make sure the indicators are the correct amounts. The AI may perform the multiplication incorrectly or use the wrong amount.
Risk map: AI drafts, auditor gives priority
A risk map is a worksheet that collects identified risks in one place and assigns evaluations (likelihood, impact, relevant claim, planned testing) to each. Assertions in accounting are truths that management implicitly asserts about an account or transaction: reality/existence, completeness, accuracy, valuation, periodicity (cutoff), rights-obligations, presentation. For example, "income is complete" is a completeness claim; "Stocks are valued at their net realizable value" is a valuation claim.
AI works well at turning a raw list of risks into an organized map: mapping each risk to the corresponding claim, listing possible tests, formatting the table. But never leave two things to him: (1) which risk really applies to this business, (2) prioritization of risks. These determine where you spend your audit resources and directly impact audit quality.
three mini cases
Case 1 — Correct use. An audit executive brainstormed risks for a newly acquired retail client. Ask AI “where is the risk of material misstatement typically high in a multi-branch retail business?” he asked; He received a draft of 11 items, such as cash management, stock count differences, revenue cutting, return/discount manipulation. He filtered this through his own knowledge: the client had low cash sales (card-heavy), reducing that risk; But the risk of stock cutting, which AI did not miss, was critical in this business because there was an intense campaign at the end of the year. Result: AI delivered the seed, the auditor customized the map to the business.
Case 2 — Significance multiplication error. A team member had the AI calculate a cardinality scenario. YZ mistakenly took the revenue as 24 million instead of 240 million and increased the materiality to 120,000 TL. The team member put this into planning without verification; Because performance materiality was underdetermined, the team began testing hundreds of additional items unnecessarily. The amount was corrected when the responsible person noticed. Lesson: Recalculate every multiplication of AI.
Case 3 — Mistaking the public list for a business. An intern pasted the list of generic "construction industry risks" given by YZ onto the risk assessment worksheet. However, this customer did not have the risk of "percentage of completion manipulation in multi-year construction contracts" listed (the customer was a material supplier, not a subcontractor); On the other hand, the customer-specific "transfer pricing in intra-group sales" risk was not on the list. Mistaking the general list as a business-specific assessment both inflated and left the risks incomplete.
Weak prompt / Strong prompt
Weak prompt:
Tell us the risks and materiality of this company.
Problem: the business is not introduced, the indicator and amount are not given, and the AI is made to produce a "certain risk" and a fictitious materiality amount.
Powerful prompt:
Your role: you are the planning assistant to an independent auditor. The final judgment of risk and materiality is mine.Context (anonymous): A medium-sized, multi-branch food retail business. There is an intense campaign at the end of the year. Card-based sales, low cash. Prior year finding: inventory count variances.Task A - Risk idea:List the areas where the risk of material misstatement would typically be high in this type of business, mapping each to the relevant accounting assertion (completeness, truthfulness, periodicity/cutoff, valuation, etc.). Please note that these are GENERAL and the business-specific evaluation belongs to me. Task B - Materiality scenario:Calculate the materiality scenario table with %s for the following indicators; Show each multiplication clearly so I can verify it. I will decide which indicator to choose.- Profit before tax: 12.000.000 TL- Revenue: 240.000.000 TL- Total assets: 180.000.000 TLRule: Do not make up any numbers for which I have not given data.
This prompt is strong because it introduces the business, requires claim matching, makes the account verifiable, and leaves the decision up to the auditor.
Common mistakes
- Mistaking the general risk list for a business assessment. Putting the generic output of the AI into the worksheet without filtering it.
- Not verifying cardinality multiplication. Not noticing the AI's error in amount or percentage.
- Leaving indicator selection to AI. Which indicator is appropriate is a matter of professional judgement.
- Not connecting risks to claims. Risk without claim matching cannot be linked to the test.
- Neglecting previous year findings and business specific information. AI doesn't know these things; If you don't tell him, the risk map will be incomplete.
Tip: After preparing the risk map, ask the AI "What are the risks that may have been overlooked in this map?" ask backwards. This helps you see blind spots; But filter the ideas yourself.
In summary
In the planning phase, AI brings two powerful uses: sectoral risk brainstorming and materiality scenario calculation. But assessing the risk of material misstatement in accordance with ASA 315 and determining materiality in accordance with ASA 320 is professional judgment. AI provides generic risk insight and verifiable accounting; Which risk is valid, which indicator to choose and the final materiality amount belong to the auditor. Recalculate every materiality multiplication, filter every risk idea business-specific.
Application task
Choose a hypothetical business (determine industry, size, previous year finding). With the powerful prompt pattern above, first get the generic risk idea, then the materiality scenario. Transform the AI's risk list into a three-column table: "risk / relevant claim / does it apply to this business (Y/N, justification)". Manually verify each multiplication in the materiality scenario and write in one sentence the justification for your chosen indicator.
checklist
- [ ] I described the business anonymously but with sufficient context (industry, size, historical findings).
- [ ] I filtered AI's risk ideas through business-specific knowledge; I removed the invalid ones and added the missing ones.
- [ ] I mapped each risk to the relevant accounting assertion and linked it to a test.
- [ ] I have independently verified each product and amount in the materiality scenario.
- [ ] I justified the choice of indicator; I made the decision.
- [ ] I set the threshold for performance materiality and obviously non-materiality.
- [ ] “Are there any risks being overlooked?” I asked backwards and filtered the ideas.