Strategic Business Leader · Machine learning, AI and robotics

Artificial Intelligence and Machine Learning Basics 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 branch of AI where systems learn patterns from data instead of following fixed rules. To answer SBL questions, define the term, name the learning type, link it to the scenario's data and state the business effect.

Understand Artificial Intelligence and Machine Learning Basics

Artificial intelligence (AI) is the broad idea of machines doing tasks that normally need human intelligence, such as recognising speech, judging risk or making decisions. It is the umbrella term. Not all AI learns. Some AI simply follows rules written by people.

A traditional rule-based system works on instructions a programmer writes in advance: if X, then do Y. It is predictable and easy to explain and audit. But it only handles situations its designers thought of, and someone must update the rules by hand when conditions change.

Machine learning (ML) is a subset of AI. Instead of being given rules, the system is given data and finds patterns itself. It then uses those patterns to predict or classify new cases. Its performance usually improves as it sees more good-quality data. Poor or biased data gives poor or biased results.

There are three main types of learning:

Supervised learning: the system learns from labelled data, where the right answer is known. Example: past loan applications marked 'repaid' or 'defaulted' are used to predict default on new applications.
Unsupervised learning: the system gets unlabelled data and looks for structure on its own, such as grouping customers into segments or spotting unusual transactions.
Reinforcement learning: the system learns by trial and error. It takes actions, receives rewards or penalties, and adjusts to maximise reward. Example: dynamic pricing or a robot learning to move items.

A neural network is an ML model loosely inspired by the brain. It has layers of connected nodes that pass and weight signals. Deep learning uses neural networks with many layers. It suits complex data such as images, speech and text. The trade-off is that it needs large data sets and computing power, and it is often a 'black box': hard to explain why it reached a decision. This matters for governance, ethics and accountability, which SBL often tests.

Key rules to remember

AI, ML and deep learning hierarchy: Deep learning ⊂ Machine learning ⊂ Artificial intelligence; Each is a subset of the one before. Use this to answer 'what is the difference' questions.
Rule-based vs machine learning: Rule-based: rules + data → answers. Machine learning: data + answers → rules; In supervised ML the system works out the rules from examples. In rule-based systems people write them.
Supervised learning: Labelled data → predict a known outcome; Used for classification (default or not) and prediction (a value, such as sales).
Unsupervised learning: Unlabelled data → find hidden groups or patterns; Used for segmentation and anomaly detection. No right answer is supplied.
Reinforcement learning: Action → reward or penalty → improved action; Learning by trial and error to maximise reward.

How to solve Artificial Intelligence and Machine Learning Basics questions

Use this method for any SBL task on AI and machine learning. Keep the answer tied to the scenario, not to textbook definitions alone.

1Read the requirement and note the verb: explain, evaluate, recommend or advise. It sets the depth and the structure of your answer.
2Pick out the scenario facts: what data the organisation holds, whether outcomes are labelled, what decision is being made and who is affected.
3Define the key term in one sentence, only as far as needed. Do not write a textbook essay.
4Match the technology to the problem: supervised for predicting known outcomes, unsupervised for finding patterns, reinforcement for repeated decisions with feedback, deep learning for images, speech or text.
5Apply it to the business: state the benefit in the scenario, such as faster decisions, lower cost or better customer insight.
6Balance with limitations: data quality, bias, explainability, cost, skills, security and loss of human judgement.
7Give a clear recommendation or conclusion, with conditions such as a pilot, human oversight or governance controls.
8Check you have earned professional skills marks: structure, scepticism about the vendor's claims, commercial awareness and the right tone for the reader.

Quickest way: Define, match, apply, caution

When to use it: Use when time is short, for example a 6 to 8 mark part of a task asking you to explain or advise on an AI tool.

Define: one line on the term asked.
Match: name the learning type that fits the scenario data.
Apply: two scenario-specific benefits.
Caution: two risks, with one control for each.
Conclude: one sentence recommending the next step.

Common mistakes in Artificial Intelligence and Machine Learning Basics

Treating AI, machine learning and deep learning as the same thing.
News and vendors use the terms loosely, so students copy that habit.
Fix: State the hierarchy: deep learning is part of machine learning, which is part of AI. Say it in one line when the question asks for the difference.
Mixing up supervised and unsupervised learning.
Students remember the names but not the key feature, which is whether the data is labelled.
Fix: Ask one question: does the data include the known answer? If yes, it is supervised. If the system must find groups itself, it is unsupervised.
Saying AI is always better than rule-based systems.
Students assume newer technology is superior.
Fix: Rule-based systems are transparent, cheap and suit stable, clear-cut decisions. Recommend ML only where patterns are complex or change often, and say why.
Ignoring data quality and bias.
Students focus on benefits and forget that ML learns from whatever data it is given.
Fix: Always mention that biased, incomplete or outdated data produces unreliable output, and suggest data checks and human review.
Writing generic technology answers that ignore the scenario.
Students recall notes and do not link them to the case facts.
Fix: Quote or refer to specific facts, such as the type of data held or the customers affected, in every paragraph. This also earns professional skills marks.
Ignoring the black box problem in deep learning.
Students describe neural networks only as powerful.
Fix: Add that decisions can be hard to explain. Link this to accountability, regulation and the need for human oversight.

Worked examples

Example 1

A retail bank holds ten years of loan records, each marked as repaid or defaulted. The board asks you to explain which type of machine learning could help predict defaults on new applications, and why this differs from its current credit-scoring rules. (8 marks)

Show the solution

Identify the data: the records carry known outcomes (repaid or defaulted), so they are labelled.
Match the type: this is supervised learning. The model learns from past applicants' details and outcomes, then predicts the outcome for new applicants.
Explain the difference: the current scoring uses fixed rules written by staff, such as income thresholds. A supervised model finds its own patterns and may combine many factors that staff did not think of.
Give a benefit: possibly more accurate risk assessment, faster decisions and lower bad debts, and the model can be retrained as customer behaviour changes.
Give limitations: the model reflects past data, so historic bias could be repeated. Complex models may be hard to explain to customers and regulators.
Recommend: pilot the model alongside the current rules, test for bias, keep human review of borderline cases, then roll out if results are reliable.

Answer: Supervised learning fits because the loan data is labelled. It learns patterns from outcomes rather than following fixed rules, so it can improve accuracy and speed. The bank should guard against bias and lack of explainability by piloting it, testing it and keeping human oversight.

Example 2

A online fashion retailer has large amounts of unlabelled customer browsing and purchase data. A director says, 'We should just buy a deep learning system, because AI will find everything.' Evaluate this view. (8 marks)

Show the solution

Correct the terms: deep learning is one form of machine learning, which is one part of AI. It is not a single product that finds everything.
Match the problem: the data is unlabelled, so the likely first use is unsupervised learning, for example grouping customers into segments or spotting unusual buying patterns.
Say when deep learning helps: it suits complex data such as images and text, for example recommending items from product photos. It may be unnecessary for simple segmentation.
Give benefits: better targeted marketing, personalised offers and higher customer retention.
Give limitations: deep learning needs large data sets and computing power, can be costly, and is hard to explain. Data protection rules apply to customer data, and poor data gives poor results.
Conclude: the view is overstated. Start with a clear business question, use the simplest method that works, and run a pilot with data governance in place.

Answer: The director's view is overstated. AI is not a magic fix, and deep learning is only one tool. Unsupervised learning is a sensible first step for unlabelled data, with deep learning considered later where complex data justifies the cost. A pilot with strong data governance is the right approach.

Exam tips

Examiners reward application. Tie every definition to the data, customers or decision in the scenario.
Give a balanced answer. If you describe benefits, add risks and controls, and finish with a recommendation.
Use the labelled or unlabelled data test to choose between supervised and unsupervised learning, and show your reasoning in one sentence.
Link black box models to governance, ethics and accountability, and refer to the ACCA ethical principles where the scenario involves decisions affecting people.
Keep definitions short. Spend the saved time on analysis and evaluation, where the professional skills marks are earned.

Practice questions from Machine learning, AI and robotics

Artificial Intelligence and Machine Learning Basics 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 and Machine Learning Basics: frequently asked questions

What is the difference between artificial intelligence and machine learning?

Artificial intelligence is the wide field of machines doing tasks that need human-like intelligence. Machine learning is a subset where systems learn patterns from data instead of following fixed rules. All machine learning is AI, but not all AI is machine learning.

What is the difference between supervised and unsupervised learning?

Supervised learning uses labelled data, where the correct outcome is known, to predict outcomes for new cases. Unsupervised learning uses unlabelled data and finds groups or patterns by itself. The key test is whether the data includes the answer.

What are neural networks and deep learning?

A neural network is a machine learning model made of layers of connected nodes that weight and pass on signals. Deep learning uses networks with many layers and suits complex data such as images, speech and text. It is powerful but often hard to explain.

Do I need to know the mathematics behind machine learning for SBL?

No. SBL tests whether you understand what the technologies do and can advise on their business use, risks and governance. Focus on concepts, application to the scenario and professional judgement.