AI can feel like a subject everyone else already understands. The good news is that you can start with a few ideas, without learning to code or memorising a dictionary of technical terms. This is the first article in our five-part beginner series.
Three terms worth knowing
Artificial intelligence is the broad field of building computer systems that perform tasks such as recognising patterns, interpreting language, and making predictions. AI is not one product, and it does not always take the form of a chatbot.
Machine learning is one approach within AI. Instead of specifying every rule by hand, developers train a model using data. A model is the system that has learned patterns it can apply to new inputs.
Generative AI produces content, including text, images, or audio. When you ask a text-based tool to draft an invitation, you are using a generative application. Other AI systems might classify an email as spam or predict a number rather than create a piece of writing.
Background: IBM’s explanation of AI and its main branches
Training and using a model are different
Training is the stage where a model’s internal settings are adjusted using data. Using the trained model to produce an output is called inference. In a conversation, the tool can use earlier messages as context; that is different from retraining the underlying model with every reply.
Background: IBM’s introduction to AI inference
Try a task you can judge
For your first experiment, use an imaginary situation. You can concentrate on the response without worrying about whether you have shared someone’s private details. Write down what success would look like before you ask.
Now read it like an editor. Is it welcoming? Is it close to the requested length? Did it add something you never supplied? Keep the original prompt beside the answer so the comparison is concrete.
An output still needs your judgement
Producing a response and establishing that it is correct are different jobs. Generative AI can produce convincing errors. That means an answer can be a useful starting point while still needing corrections, outside evidence, or a completely different approach.
Source: NIST’s Generative AI Profile, including the risk of confabulation
Your first learning goal
After one experiment, finish these sentences: ‘I asked for…’, ‘The tool produced…’, and ‘I changed…’. If you can explain the difference, you are already moving beyond simply accepting whatever appears on the screen.

