Strategic Business Leader · Enabling success: disruptive technologies

Artificial Intelligence, Machine Learning and Automation for SBL

Updated 11 October 2026 · Fact-checked

Artificial intelligence (AI) is software that performs tasks needing human-like judgement. Machine learning (ML) is a type of AI that learns patterns from data. Robotic process automation (RPA) follows fixed rules to do repetitive tasks. In SBL, you apply them to the scenario, weigh benefits against risks, and address ethics and people.

Understand Artificial Intelligence, Machine Learning and Automation

Artificial intelligence (AI) is the broad idea of computer systems doing tasks that normally need human thinking. Examples are recognising speech, spotting fraud, answering customer queries and making predictions.

Machine learning (ML) is a part of AI. Instead of being given rules, the system is given large amounts of data and finds patterns itself. It improves as it sees more data. A bank can train a model on past loans to predict which new borrowers may default. The quality of the output depends on the quality of the data. Poor or biased data gives poor or biased results.

Robotic process automation (RPA) is different. It uses software 'bots' to copy what a person does on a screen: moving data between systems, matching invoices, producing routine reports. RPA follows fixed rules. It does not learn or use judgement on its own. It works best on high-volume, stable, rule-based tasks. It can be combined with AI so that bots handle judgement-based steps.

Business uses include faster and cheaper processing, fewer errors, 24-hour service, better forecasting, personalised marketing, fraud detection and better use of staff time. Risks and issues include high set-up cost, poor data, bias in decisions, lack of transparency (the 'black box' problem), cyber and data protection risk, over-reliance on systems, and job losses or role changes.

Ethical and workforce implications matter in SBL. Who is accountable when an algorithm makes a wrong decision? Is the decision fair to customers and staff? Is personal data used lawfully and with consent? How will redundancies be handled, and will staff be retrained? Professional accountants must apply the ACCA Code of Ethics, especially integrity, objectivity, professional competence and due care, and confidentiality. Governance matters too: the board should oversee AI use and its risks.

Key rules to remember

AI vs ML vs RPA: AI = broad field; ML = AI that learns from data; RPA = rule-based automation of repetitive tasks; Do not treat them as the same thing. RPA does not learn. Many real systems combine them.
Business case test: Net benefit = cost savings + revenue gains + risk reduction − (set-up + running + change costs); A reasoning frame, not a calculation. Include non-financial factors such as customer trust and staff morale.
Suitability test for RPA: High volume + rule-based + stable process + digital inputs = good RPA candidate; Tasks needing judgement, empathy or frequent change are poor candidates.
Ethics check for AI use: Fairness + transparency + accountability + privacy + human oversight; A memory aid for ethical and governance points. It is not an official ACCA list.

How to solve Artificial Intelligence, Machine Learning and Automation questions

Use this method for any SBL task on AI, ML or automation. Keep every point tied to the scenario.

1Read the requirement and note the verb: assess, evaluate, advise, recommend or discuss. This sets the depth and whether you need a recommendation.
2Identify which technology the scenario describes: AI, ML, RPA or a mix. Define it in one line only if it helps your argument.
3Pick out scenario facts: the industry, process, volume of work, data available, staff numbers, customers and any ethical clues.
4Explain benefits that fit those facts, such as speed, accuracy, cost, insight or customer service. Give a reason for each, not just a label.
5Explain risks and limits: cost, data quality, bias, transparency, cyber and data protection, staff resistance and over-reliance.
6Cover ethical, governance and workforce issues. Link to the ACCA Code principles and to board oversight where relevant.
7Give a reasoned conclusion or recommendation, such as a phased pilot, with safeguards and staff retraining.
8Check professional skills: a clear format for the audience, balanced analysis, scepticism about vendor claims and commercial sense.

Quickest way: Benefit, risk, people, recommend

When to use it: When time is short and you must plan an answer in about two minutes.

Write four headings on your plan: Benefits, Risks, People and ethics, Recommendation.
Under each, jot two or three scenario-linked points.
Put the strongest scenario facts first in each section.
Write the recommendation as one clear sentence with a condition, for example pilot first and review results.
Spend any spare minutes on professional skills: audience, tone and balance.

Common mistakes in Artificial Intelligence, Machine Learning and Automation

Writing a generic list of AI benefits and risks.
Students recall textbook lists and skip the scenario.
Fix: Use the company's industry, data, staff and customers in every point. Name the process affected.
Treating AI, machine learning and RPA as the same.
The terms are used loosely in the media.
Fix: State that RPA follows fixed rules and ML learns from data. Match the right tool to the task.
Ignoring ethics or mentioning it in one line.
Students focus on technical and cost points.
Fix: Give ethics its own section. Cover bias, transparency, privacy, accountability and the Code of Ethics principles.
Assuming automation always cuts costs and staff.
Students overlook set-up, maintenance and change costs.
Fix: Weigh costs against benefits. Discuss retraining, redeployment and morale, not just redundancy.
Failing to conclude or recommend.
Students run out of time after listing points.
Fix: Reserve time for a clear recommendation, such as a phased roll-out with human oversight.
Uncritically accepting vendor or management claims.
Students forget the scepticism skill being assessed.
Fix: Question the data quality, assumptions, evidence of results and who benefits from the claim.

Worked examples

Example 1

A mid-sized insurer in India processes about 5,000 motor claims a month. Staff re-key data between three systems and claims take ten days. The CEO wants to introduce RPA and a machine learning model to flag suspicious claims. Evaluate the proposal. (12 marks)

Show the solution

Identify the tools: RPA suits the re-keying between systems because it is high volume, rule-based and repetitive. ML suits fraud flagging because it finds patterns in past claims data.
Benefits of RPA: faster processing than ten days, fewer keying errors, lower cost per claim and staff freed for complex claims and customer contact.
Benefits of ML: better detection of fraud patterns humans may miss, lower fraud losses and more consistent screening.
Risks of RPA: bots break if systems or screens change, set-up and maintenance cost, and automating a poor process only makes errors faster.
Risks of ML: it depends on good historical data. If past decisions were biased, the model repeats the bias. False positives may delay or deny genuine claims and harm customers.
Ethical and people issues: customers must be treated fairly, and the decisions need to be explainable. Personal data must be protected. Staff may fear job losses, so consult them and plan retraining.
Governance: a human should review flagged claims before rejection. The board and risk committee should oversee the model and test it regularly.
Recommend: pilot RPA on one claim type first, then add ML with human review, measuring speed, accuracy and complaints before a full roll-out.

Answer: The proposal is sound in principle. RPA fits the repetitive re-keying and ML fits fraud detection. Approve a phased pilot with human review of flagged claims, data quality checks, bias testing, privacy safeguards and a staff retraining plan, then expand if measured results justify it.

Example 2

A finance director at a global retailer says: 'Our new AI tool will screen job applicants and we will cut the HR team.' As a consultant, advise on the ethical and workforce issues. (10 marks)

Show the solution

Fairness and bias: the tool learns from past hiring data. If that data favoured certain groups, the tool may discriminate. This risks harm to candidates, legal claims and reputational damage.
Transparency: candidates and managers should understand how decisions are made. A 'black box' tool makes it hard to explain or challenge rejections.
Accountability: the company stays responsible for hiring decisions, not the supplier. Assign a named owner and require human review of rejections.
Privacy: applicant data must be collected and used lawfully, kept secure and held only as long as needed.
Workforce: cutting the HR team removes people who can judge the tool's output. Consider redeployment and retraining into roles such as data oversight and candidate experience.
Professional ethics: a professional accountant should apply integrity, objectivity and professional competence and due care. They should not endorse a tool they do not understand or whose claims they have not tested.
Scepticism: ask the vendor for evidence of accuracy and bias testing, and review it independently.
Recommend: use the tool to support, not replace, human decisions. Test for bias, keep human oversight, tell candidates how it is used and manage any HR changes through consultation.

Answer: Advise against using the tool as the sole decision-maker or as a reason to cut the HR team. Use it as decision support with bias testing, transparency, a named accountable owner, data protection controls and human review, and handle workforce change through consultation and retraining.

Exam tips

Always tie points to the scenario. Generic AI answers score poorly, even if technically correct.
Give ethics and workforce impact their own section. Examiners often reward a balanced view of benefits and risks.
Match the tool to the task: RPA for rule-based repetition, ML for pattern finding. Showing this distinction signals real understanding.
Finish with a clear, conditional recommendation. Phased pilots with human oversight are usually a sensible position.
Show professional scepticism about claims from management or vendors. It supports your professional skills marks.

Practice questions from Enabling success: disruptive technologies

Artificial Intelligence, Machine Learning and Automation in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Artificial Intelligence, Machine Learning and Automation: frequently asked questions

What is the difference between AI, machine learning and RPA?

AI is the broad field of machines doing tasks that need human-like thinking. Machine learning is a type of AI that learns patterns from data. RPA uses bots to follow fixed rules on repetitive tasks and does not learn on its own.

What are the advantages and disadvantages of robotic process automation?

Advantages include speed, accuracy, round-the-clock work and lower processing cost. Disadvantages include set-up and maintenance cost, bots failing when systems change, and limited use for tasks needing judgement. It may also cause job concerns among staff.

How does machine learning help businesses?

It finds patterns in large data sets to predict outcomes, such as customer demand, credit risk or fraud. This improves decisions and personalisation. Results depend on good data and need human oversight to avoid bias and errors.

How are AI and ethics tested in SBL?

They usually appear inside the case study as a decision about adopting or using technology. You are expected to weigh benefits and risks, discuss fairness, transparency, privacy and accountability, and apply the ACCA Code of Ethics. A clear recommendation earns credit.