Accounting and Technology: formula sheet

Key formulas

AI vs ML vs RPA

AI = broad capability; ML = AI that learns from data; RPA = rule-based automation, no learning

Most exam errors come from treating RPA as AI that learns. RPA follows programmed rules.

Suitability test for RPA

RPA suits tasks that are rule-based + repetitive + high-volume + structured data

If the task needs judgement or unstructured data, think ML or human review.

Answer frame

Task → Technology → Benefit → Risk → Control

Use this order for every case-based answer.

Fraud detection logic

Anomaly flagged ≠ fraud proven; flagged item → human investigation

ML gives indicators. Conclusions need evidence and professional judgement.

Blockchain in one line

Blockchain = distributed ledger + blocks linked by hashes + consensus

Use this to define it quickly. Add smart contracts if the question mentions automation.

Big data characteristics

Volume, Velocity, Variety, Veracity

Some sources add Value. Mention the four and say sources may differ.

Types of analytics

Descriptive → Diagnostic → Predictive → Prescriptive

Increasing sophistication. Match each type to a reporting use.

Benefit–limitation rule

Feature → reporting benefit → limitation

A balanced answer structure that earns marks in descriptive questions.

What XBRL is

XBRL = open, XML-based standard for tagging and exchanging business and financial data

It is a reporting format, not an accounting standard. It does not alter recognition or measurement.

Taxonomy

Taxonomy = dictionary of elements + definitions + relationships (labels, references, calculations, presentation)

Prepared centrally. The company uses it; it does not create it for the filing.

Instance document

Instance document = facts (values) + context (entity, period) + unit (e.g. INR, scale)

This is the file you generate for each filing. Every number must map to a taxonomy element.

Filing logic

Financial statements → map to taxonomy → generate instance document → validate with the MCA utility → attach to AOC-4 / AOC-4 CFS

Validate before filing and fix any errors the check reports.

MCA applicability test (Companies (Filing of Documents and Forms in Electronic Form) Rules)

XBRL required if the company meets a condition in the rules at the time of the question. Conditions usually tested: listed company or its Indian subsidiary OR paid-up capital of ₹5 crore or above OR turnover of ₹100 crore or above OR Ind AS company

Thresholds and categories are as per the rules at the time of the question. On the conditions as framed, meeting any one is enough. The rules also exempt certain categories (for example banking, insurance, power and non-banking financial companies), and the treatment of Indian subsidiaries and Ind AS companies has nuances. The rules change over time, so verify the text in force. In an exam, if the question gives conditions or thresholds, apply those and rely only on the facts given.

Risk to control chain

Threat → Vulnerability → Risk to financial data → Control → Residual risk

Use this chain to structure any answer. Controls reduce risk but never remove it fully.

Four ITGC areas

Access security + Change management + IT operations + System development/acquisition

Name all four when a question asks for IT general controls.

Control types by timing

Preventive + Detective + Corrective

Give at least one example of each for any risk.

Security objectives (CIA)

Confidentiality + Integrity + Availability

Breach hits confidentiality, tampering hits integrity, system failure hits availability.

Control levels

ITGC (entity-wide) vs Application controls (process-specific)

Do not mix them up. ITGCs support the reliable working of application controls.

Quick revision

• Technology changes how data is captured, processed, stored and reported, not the underlying accounting principles.
• AI mimics human judgement for tasks like classification and anomaly spotting; ML learns patterns from data without being explicitly programmed for each rule.
• ML quality depends on data quality; poor or biased data gives poor or biased results.
• Blockchain is a shared, append-only ledger where records are linked and hard to alter after validation.
• Big data is described by large volume, variety and velocity; analytics turns it into insight.
• Data analytics can be descriptive, diagnostic, predictive or prescriptive.
• XBRL tags financial data so it is machine readable and comparable across entities.
• XBRL uses taxonomies as the standard dictionary of tags; instance documents hold the tagged data.
• Digital reporting reduces re-keying, errors and time, and helps regulators and users analyse data.
• Cyber risks include unauthorised access, malware, phishing and data loss or breach.
• Controls can be preventive, detective or corrective, and cover access, backups, encryption and monitoring.
• Automation does not remove management responsibility for accuracy and control over reporting.

Common mistakes

• Writing that technology changes Ind AS recognition or measurement rules. Fix: State that Ind AS requirements stay the same. Technology changes speed, accuracy, consistency and control, not the standard.
• Listing only benefits and ignoring risks. Fix: Give at least one risk for every tool: security, access, data quality, configuration errors or vendor dependence. Then add the control.
• Saying RPA learns from data like ML. Fix: Remember RPA follows fixed programmed rules. ML learns patterns from data.
• Claiming AI replaces the accountant or auditor. Fix: Write that AI supports staff. Management and the auditor remain accountable and must review outputs.
• Saying blockchain makes data always correct. Fix: Write that blockchain protects recorded entries from alteration, but wrong data entered at the start stays wrong.
• Treating blockchain and a traditional ledger as the same thing with a new name. Fix: Contrast them: one entity keeps a traditional ledger centrally, while a blockchain is shared across participants with consensus and linked blocks.
• Treating XBRL as an accounting standard that changes how items are measured. Fix: Say clearly that XBRL is only a tagging and delivery format. Ind AS decides the numbers.
• Mixing up taxonomy and instance document. Fix: Use the form analogy. Taxonomy is the blank form with defined fields. Instance document is the filled form with your values, period and unit.
• Listing generic cyber threats without linking to financial reporting. Fix: End each risk with its effect on the accuracy, completeness or availability of financial data.
• Confusing ITGCs with application controls. Fix: ITGCs cover the whole environment (access, change, operations, development). Application controls sit inside one process, such as input checks.

Exam tips

• Answer in the pattern tool, effect, benefit, risk and control. It covers most questions on this topic.
• Use facts from the case in every point. Examiners reward application, not memorised notes.
• In MCQs, reject options that say technology removes the need for judgement, controls or management responsibility.
• Never suggest that technology alters Ind AS recognition or measurement. Keep the focus on process, close and control.
• Link with the related digital topics such as AI, blockchain, XBRL and cybersecurity, as a case may combine them.
• Expect case-scenario MCQs that ask which technology fits a task. Decide on rule-based versus learning-based first.
• In descriptive answers, always pair a benefit with a risk and a control. This earns balanced marks.
• Define AI, ML and RPA in one line each at the start. Clear definitions score easily.