Learn AI
What is machine learning? Plain English, South African examples
What is machine learning? It is the part of artificial intelligence that learns from examples. Instead of a programmer writing every rule, you give a system lots of data, it finds the patterns, and it uses them to predict or decide something about new cases. It sits quietly behind things South Africans use every day, from card fraud alerts to spam filters.
This guide gives a plain definition, real examples, the three types of machine learning, how it relates to AI, deep learning and chatbots, and what the jobs pay.
What is machine learning? A plain definition
In their 2025 report on AI in South Africa’s financial sector, the FSCA and the Prudential Authority describe machine learning as a branch of AI in which systems learn from data, spot patterns and make decisions with little human involvement.
The easiest way to see the difference is a spam filter. The old way was to write rules by hand, such as blocking any email that mentions a lottery win. The machine learning way is to show a system a large number of emails already marked spam or not spam, and let it work out the patterns itself, including ones no person thought to write down.
Machine learning meaning in everyday South African life
These are documented examples, not guesses:
- Card and bank fraud. The FSCA and Prudential Authority report found that fraud detection is the leading use of machine learning in the financial institutions they surveyed. Banks also use it to profile or cluster clients and transactions, and to identify and verify transactions to combat money laundering.
- Insurance. The same report says insurers mainly use machine learning for underwriting and claims management.
- Credit decisions. Creditworthiness assessment and credit scoring appear among the machine learning use cases the report lists across banks, insurers and investment firms.
- Tax and customs. In August 2024 SARS said it would make greater use of data, artificial intelligence, machine learning and algorithms to facilitate trade while managing risk.
- Your inbox. Google said in 2019 that its machine learning models, with its other protections, help block more than 99.9% of spam, phishing and malware from reaching Gmail inboxes.
The same report found that 52% of banking institutions surveyed were actively using AI in 2024, the highest of any part of the financial sector.
The three types of machine learning
IBM groups machine learning methods into three kinds:
| Type | How it learns | A simple example |
|---|---|---|
| Supervised learning | From examples with the right answer attached, called labelled data | Emails labelled spam or not spam |
| Unsupervised learning | Finds patterns and groups in data with no labels | Grouping customers or transactions that behave alike |
| Reinforcement learning | By trial and error, getting a reward for good actions | Fine-tuning a chatbot using feedback from people |
Real systems often mix them. IBM notes that large language models are usually pre-trained first and then fine-tuned, including with reinforcement learning from human feedback.
Artificial intelligence vs machine learning vs deep learning vs LLMs
These terms sit inside one another, like nesting boxes:
| Term | What it is | Example |
|---|---|---|
| Artificial intelligence | The broad field of making machines infer, predict, recommend or decide | Any smart system, including rule-based ones |
| Machine learning | AI that learns patterns from data | A fraud model trained on past transactions |
| Deep learning | Machine learning that uses neural networks with many layers | Digital assistants and fraud detection |
| Large language models | Deep learning models trained on huge amounts of text to understand and generate language | The models behind ChatGPT and Claude |
Every machine learning system is a kind of AI, but plenty of AI, such as a simple rule-based system, involves no learning at all. For a plain explanation of the last row, Zaiq’s guide to what an LLM is goes further.
Machine learning engineer salary in South Africa
Solid public salary data for machine learning roles in South Africa is thin, and most of it is self-reported. Here is what the credible sources show:
| Source | Role | Figure |
|---|---|---|
| PayScale, updated 22 May 2026, 57 salary profiles | Machine learning engineer | Average about R428,920 a year |
| PayScale, updated 7 July 2026, 341 salary profiles | Data scientist | Average about R460,138 a year |
| OfferZen 2026 report, reported by MyBroadband | Developer with 10+ years in Cape Town | About R105,092 a month |
The machine learning engineer figure rests on only 57 people, and PayScale’s own split by experience uses just six salaries at entry level, so treat it as a rough guide. Data scientist figures come from a larger sample. The OfferZen figures reported by MyBroadband cover developers in general rather than machine learning engineers; the same report found that 97% of tech teams use AI in their daily work.
For roles, pay and remote AI training work, see our guide to AI jobs in South Africa.
How to start learning machine learning
A common starting point is Python, basic statistics and working with data in spreadsheets or code, followed by small projects on real data, such as predicting prices or sorting messages. Free online courses are a good first step before you pay for a qualification. Our guide to AI courses in South Africa compares free and paid options in rand.
Related guides
Questions people ask
What is machine learning in simple words?
Machine learning is a way of building software that learns from examples instead of following rules a programmer wrote by hand. Show it thousands of past card transactions marked fraud or not fraud, and it learns patterns it can use to flag new suspicious ones. The more good data it sees, the better it usually gets.
What is the difference between artificial intelligence and machine learning?
Artificial intelligence is the broad goal of making computers do things that seem intelligent. Machine learning is one way of getting there, where the system learns from data. As IBM puts it, "all machine learning is AI, but not all AI is machine learning." Deep learning and large language models are, in turn, types of machine learning.
What are the three types of machine learning?
Supervised learning learns from labelled examples, such as emails marked spam or not spam. Unsupervised learning finds patterns in data without labels, such as grouping similar customers or transactions. Reinforcement learning learns by trial and error, earning rewards for good actions, and is also used to fine-tune chatbots with human feedback.
How much does a machine learning engineer earn in South Africa?
Public data is thin. PayScale's South African figure for machine learning engineers is an average of about R428,920 a year, but it rests on only 57 salary profiles, last updated in May 2026. Its larger sample of 341 data scientists averages about R460,138. Treat both as rough guides, not offers.
Where is machine learning used in South Africa?
In finance, banks use it mainly for fraud detection and to profile or cluster clients and transactions, and insurers use it for underwriting and claims, according to a 2025 FSCA and Prudential Authority report. SARS has said its Customs division will make greater use of data, AI, machine learning and algorithms. Your email spam filter is another everyday example.
Is ChatGPT machine learning?
Yes. ChatGPT runs on large language models, which are a type of deep learning, which is itself a type of machine learning. According to IBM, large language models are first pre-trained on huge amounts of text and then fine-tuned, including with reinforcement learning from human feedback.
Sources
- FSCA and Prudential Authority: Artificial Intelligence in the South African Financial Sector (November 2025)
- FSCA and Prudential Authority release report on AI in South Africa's financial sector (24 November 2025)
- SARS: Changes to Customs import system (8 August 2024)
- Google Workspace blog: Ridding Gmail of 100 million more spam messages with TensorFlow (7 February 2019)
- IBM: What is machine learning?
- IBM: AI vs. machine learning vs. deep learning vs. neural networks
- PayScale: Machine learning engineer salary in South Africa (updated 22 May 2026)
- PayScale: Data scientist salary in South Africa (updated 7 July 2026)
- MyBroadband: Top paying developer jobs in South Africa in 2026 (OfferZen data)
Checked September 2026. Prices and features change, so confirm on the official site before you buy.