Fundamentals of AI and ML · Topic 1 of 8
AI, ML and Deep Learning
Learn how artificial intelligence, machine learning, deep learning and generative AI nest inside each other, and the core terms the AIF-C01 exam uses.
- 4 min
- AI is the widest term, ML sits inside AI, deep learning sits inside ML, and GenAI sits inside deep learning.
- ML learns patterns from data instead of following hand written rules.
- Training builds the model, and inference uses the model on new data.
What it is
Artificial intelligence (AI) is the broad goal of making software do tasks that normally need human judgement, such as understanding language, recognising images or making decisions.
Machine learning (ML) is one way to reach that goal. Instead of writing every rule by hand, you give an algorithm example data and let it find the patterns. The result of that learning is called a model.
Deep learning is a part of ML that uses neural networks with many layers. Generative AI (GenAI) is a part of deep learning that creates new content, such as text, images or code.
How it works
A classic program follows rules that a developer wrote. An ML system works the other way round: you supply data and the expected answers, the algorithm adjusts itself, and it produces a model that can make predictions on new data.
The core vocabulary
| Term | Plain meaning |
|---|---|
| Algorithm | The learning procedure that is applied to data |
| Model | The learned result that you use to make predictions |
| Training | The phase where the algorithm learns from data |
| Inference | Using the trained model on new data to get an answer |
| Feature | One input property, such as age or house size |
| Label | The correct answer attached to a training example |
| Parameter | A value inside the model that training adjusts |
Training is usually costly and happens before the application goes live. Inference is what the application does every time a user asks for a result.
Neural networks
A neural network is a set of connected layers of simple units. Data enters the input layer, passes through hidden layers, and leaves the output layer as a prediction. During training, the connections are adjusted until the output gets close to the right answer.
A network with many hidden layers is a deep neural network, which is where the name deep learning comes from. Deep learning needs a lot of data and computing power, but it handles unstructured data very well. Unstructured data means things like images, audio and free text.
Related terms the exam uses
- Computer vision is AI that understands images and video.
- Natural language processing (NLP) is AI that understands and produces human language.
- Large language model (LLM) is a very large neural network trained on huge amounts of text.
- Foundation model (FM) is a large model trained on broad data that can be adapted to many tasks.
- Speech recognition turns spoken audio into text.
Where each level fits
| Level | Typical example |
|---|---|
| Rule based software | A fixed script that approves orders under a set value |
| Classic ML | A model that predicts whether a customer will leave |
| Deep learning | A network that identifies objects in photos |
| GenAI | A model that writes a draft email from a short instruction |
An ML model does not understand a topic the way a person does. It finds statistical patterns in the data it was given, so its answers are only as good as that data.
What to remember for the exam
- AI is the widest term, ML sits inside AI, deep learning sits inside ML, and GenAI sits inside deep learning.
- ML learns patterns from data instead of following hand written rules.
- Training builds the model, and inference uses the model on new data.
- A feature is an input, and a label is the known correct answer.
- Deep learning uses neural networks with many layers and works well on images, audio and text.
- An LLM is a language focused foundation model, and a foundation model is trained on broad data for many tasks.
- Computer vision handles images, and NLP handles human language.
- A model is the output of training, while the algorithm is the method used to train it.
When two answer options look alike, check the nesting order. If a question asks for the broadest term it is AI, and if it asks for content creation it is GenAI.
Common mistakes
Treating AI, ML and deep learning as the same thing. They are nested, not interchangeable. Every deep learning system is ML, but not every AI system uses ML, since a rule based system can also count as AI.
Mixing up training and inference. Training teaches the model and is done up front. Inference is the live use of the model, and it happens many more times than training.
Confusing algorithm and model. The algorithm is the recipe for learning. The model is what you get after the recipe has been run on data.
Assuming more data always means a better model. Data must also be relevant and of good quality. Poor or biased data produces a poor or biased model.
Thinking GenAI replaces all other ML. Predicting a number or sorting items into classes is often better done by classic ML, which is cheaper and easier to explain.
Check yourself
Which term is the broadest, AI, ML or deep learning?
AI is the broadest. ML sits inside it, and deep learning sits inside ML.
What is the difference between training and inference?
Training is when the algorithm learns from data to build a model. Inference is when the finished model is used on new data to give an answer.
What is a label in a training set?
A label is the correct answer attached to an example, such as the category of an email.
Topic 1 of 8 in Fundamentals of AI and ML
Test yourself on Fundamentals of AI and ML
The full AWS AIF-C01 question bank has 849 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by AWS. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.