TrainLearnAI Start learning free

Free AI resources that make the ideas click.

Visual guides, diagrams and charts, ready-to-use prompts, a plain-English AI glossary and checklists. Everything here is free, written for beginners and useful at work from day one.

The AI family diagram Four nested circles. Artificial intelligence is the largest. Machine learning sits inside it, deep learning inside machine learning, and generative AI inside deep learning. Artificial intelligence Machines doing smart tasks Machine learning Learns patterns from data Deep learning Neural networks Generative AI Creates content
The AI family: every generative AI tool is a kind of deep learning, which is a kind of machine learning, which is one part of AI.

Pick what you need today

Each resource is colour-coded so you can find it fast anywhere on the site.

Visual guides and diagrams How a chatbot answers you, the machine learning cycle and the anatomy of a great prompt, drawn out step by step. 3 diagrams Learning charts See how long each stage takes, what each path focuses on and how to split your weekly study time. 4 charts Prompt library Copy-ready prompts for work, study, data, marketing and code. 12 prompts AI glossary Searchable, plain-English definitions of the words you keep hearing. 24 terms Free guides and cheat sheets Short reads you can finish in a coffee break. 8 guides

How an AI chat assistant answers you

A large language model (LLM) does not look answers up. It predicts the most likely next piece of text, one small piece at a time, until the reply is finished.

  1. You write a prompt

    Your question, plus any context, examples and rules you add.

  2. Text becomes tokens

    The words are split into small chunks called tokens and turned into numbers.

  3. The model finds patterns

    A neural network trained on huge amounts of text weighs how every token relates to the others.

  4. It predicts the next token

    It scores every possible next token and picks a likely one. Then it repeats.

  5. You get a reply

    The tokens become words. Always check facts, numbers and sources before you use them.

Step 2 up close: tokens

A token is often a whole short word, part of a longer word, or a punctuation mark. Here is roughly how one sentence might be split:

Learning AI is surprisingly fun!

Illustration only. Each model splits text in its own way. Tokens matter because models have a limit on how many they can read at once, called the context window.

Step 4 up close: next-token prediction

Prompt: "The capital of France is". The model gives every possible next token a score:

  • Paris91%
  • a4%
  • the2%
  • known1.5%
  • other1.5%

Illustrative numbers. This is also why AI can sound confident and still be wrong: it picks likely text, not verified facts.

The machine learning cycle

Training a model is a loop, not a straight line. Real projects go round this cycle many times, and most of the time goes into the data.

Six-step machine learning cycle: collect data, clean and prepare, train, evaluate, deploy, monitor, then repeat. Collectdata Clean &prepare Train Evaluate Deploy Monitor Repeat and improve
  • Collect dataGather examples that match the real problem, such as past sales, labelled emails or customer reviews.
  • Clean and prepareFix gaps and errors, remove duplicates and split the data into training and test sets.
  • TrainThe algorithm adjusts itself until its predictions on the training data get better.
  • EvaluateTest on data the model has never seen. Check accuracy and fairness across different groups.
  • DeployPut the model into an app, a spreadsheet or a workflow where people can use it.
  • MonitorWatch for drift: when the real world changes, results get worse and the cycle starts again.

Want to go round this cycle with real data? The Machine Learning with Python course walks you through every step with a portfolio project.

Anatomy of a great prompt

Strong prompts have five parts. Select a part on the right to see where it sits in the example.

You are an experienced HR manager at a mid-sized software company. We are hiring a junior data analyst who will work with the sales team. Most applicants are recent graduates. Write a job description for this role. Keep it under 300 words, use friendly plain English, avoid jargon and do not mention salary. Use these headings: About the role, What you will do, What you bring, Why join us. Put duties in bullet points.

Weak prompt

"Write a job description for a data analyst."
Generic result, wrong tone, random length.

Strong prompt

The five-part prompt above.
Right audience, right length, ready to post after a quick check.

Your AI learning, in charts

Plan your time with a clear picture of how long each stage takes and what you will focus on along the way.

How long each learning stage takes

Beginner AI foundations takes 4 to 6 weeks, intermediate applied AI takes 6 to 10 weeks, and advanced build and train AI takes about 13 to 17 weeks. Beginner: AI foundations 4–6 weeks Intermediate: applied AI 6–10 weeks Advanced: build and train AI 3–4 months 02468 1012141618 Weeks of study
  • Fastest pace
  • Comfortable pace

Hands-on hours in each course

Machine Learning with Python 40 hours, Build AI Agents and Automations 30 hours, AI for Absolute Beginners 16 hours, Prompt Engineering Masterclass 12 hours, Generative AI for Work 8 hours. Machine Learning with Python 40 h Build AI Agents and Automations 30 h AI for Absolute Beginners 16 h Prompt Engineering Masterclass 12 h Generative AI for Work and Productivity 8 h

Total guided learning time, including video lessons, exercises and the final project. Compare all courses.

What each path focuses on

Radar chart. Beginner focuses on prompting and AI tools. Intermediate adds data skills. Advanced focuses on Python, machine learning models and automation. Prompting AI tools Data skills Python ML models Automation
  • Beginner
  • Intermediate
  • Advanced

How to split 5 study hours a week

Suggested split: learn 25 percent, practise 35 percent, projects 30 percent, review and certify 10 percent. 5 h per week
  • Learn, 25%, about 75 min of lessons
  • Practise, 35%, about 1 h 45 min in real tools
  • Projects, 30%, about 90 min on real work
  • Review, 10%, about 30 min recapping

A suggested split based on our Learn, Practise, Train, Get certified method. Hands-on time matters most.

Prompt library

Copy a prompt, replace the parts in [square brackets] and paste it into any AI chat assistant. Remove names and private data first.

Work

Summarise a long report

Summarise the report below for a busy manager. Give me: 1) a three-sentence summary, 2) the five most important findings as bullet points, 3) any risks or decisions needed. Use plain English.

[Paste report here]
Work

Turn rough notes into a clear email

You are a professional, friendly communicator. Turn my rough notes into a clear email to [recipient and role]. Keep it under 150 words, put the main request in the first line and end with one clear next step.

Notes: [paste notes]
Work

Plan a project in phases

Act as an experienced project manager. Break this project into phases with tasks, owners (by role), rough durations and dependencies. Show it as a table, then list the three biggest risks.

Project: [describe project, team size and deadline]
Study

Explain a hard topic simply

Explain [topic] to me as if I am a smart 15-year-old. Use one everyday analogy, then give a short real example, then ask me two questions to check I understood. Do not give the answers until I reply.
Study

Make a revision quiz

Create a 10-question multiple-choice quiz on [topic] for a [level] learner. Give four options per question. List the answers with a one-line explanation each at the very end, after a line that says ANSWERS.
Data

Find insights in a spreadsheet

You are a data analyst. Here is a table of [describe data]. Identify the three most important trends, any unusual values, and one question I should investigate next. Show any numbers you calculate and explain how you got them.

[Paste table or CSV]
Data

Write a spreadsheet formula

I use [Excel / Google Sheets]. Column A has [describe], column B has [describe]. Write a formula that [what you want]. Explain what each part of the formula does, and tell me what could cause an error.
Marketing

Social post variations

Write 5 LinkedIn post variations announcing [news] for [audience]. Make each a different style: story, question, list, bold statement and data point. Keep each under 120 words and end with a soft call to action. No hashtags in the body.
Marketing

Customer persona

Based on this description of our product and customers, create one realistic customer persona: goals, daily challenges, objections to buying, where they look for information and the words they would use to describe their problem.

Product and customers: [describe]
Code

Explain code line by line

Explain what this [language] code does, line by line, for a beginner. Then tell me one way it could fail and how to make it more robust.

[Paste code]
Code

Debug an error

I get this error: [paste error]. Here is the code that causes it: [paste code]. Explain the most likely cause in plain English, show the corrected code, and explain how to avoid this error in future.
Study

Check an AI answer

Review the answer below as a strict fact-checker. List every claim that could be wrong or out of date, rate your confidence in each, and tell me exactly what I should verify and where.

[Paste AI answer]

Want 100+ more? Browse the full prompt library or learn to write your own in the Prompt Engineering Masterclass.

AI glossary in plain English

The words you keep hearing, explained in one or two sentences each.

Agent (AI agent)
An AI system that can plan steps and use tools, such as search or email, to complete a goal with little human input.
Algorithm
A set of step-by-step rules a computer follows to solve a problem or learn from data.
Artificial intelligence (AI)
The broad field of making computers do tasks that normally need human intelligence, such as understanding language or recognising images.
Bias
Unfair or skewed results caused by unbalanced training data or design choices. It must be tested for and reduced.
Chatbot
A program you talk to in text or voice. Modern chatbots are usually powered by large language models.
Context window
The maximum amount of text, measured in tokens, a model can read and remember in one conversation.
Dataset
A collection of examples, such as rows in a spreadsheet or labelled images, used to train or test a model.
Deep learning
Machine learning that uses neural networks with many layers. It powers speech, image and language AI.
Embedding
A list of numbers that represents the meaning of a word, sentence or image so computers can compare similarity.
Fine-tuning
Training an existing model a little more on your own examples so it performs better on a specific task.
Generative AI
AI that creates new content, such as text, images, audio, video or code, from a prompt.
Hallucination
When an AI states something false or made up as if it were true. Always verify important facts.
Inference
Using a trained model to make a prediction or produce an answer, as opposed to training it.
Large language model (LLM)
A model trained on huge amounts of text to understand and generate language by predicting the next token.
Machine learning (ML)
A branch of AI where systems learn patterns from data instead of following fixed, hand-written rules.
Model
The trained result of machine learning: a file of learned patterns that turns inputs into outputs.
Multimodal
An AI that works with more than one type of input or output, such as text, images and audio together.
Neural network
A model made of layers of connected nodes, loosely inspired by the brain, that learns by adjusting connection strengths.
Overfitting
When a model memorises its training data so closely that it performs badly on new data.
Prompt
The instruction or question you give an AI model. Better prompts give better results.
Prompt engineering
The skill of writing clear prompts with role, context, task, rules and format to get reliable results.
RAG (retrieval-augmented generation)
A method where the AI first searches trusted documents, then uses what it finds to write a more accurate answer.
Token
A small chunk of text, often a short word or part of a word, that language models read and write.
Training data
The examples a model learns from. Its quality largely decides how good and fair the model is.

No match yet. Try a shorter word, or suggest a term for us to add.

See every term in the full AI glossary.

Free guides and cheat sheets

Short, practical reads. Start at the top if you are new to AI.

Responsible AI checklist

Run through this before you use any AI output at work. Tick each item to see your score.

Checklist progress 0/6

Tick the items you have done.

Questions about our free resources

Are TrainLearnAI resources free?

Yes. Every guide, diagram, prompt, glossary term and checklist on this page is free to read and use. You only pay if you choose a full course with projects and a certificate.

Where should a complete beginner start?

Start with the AI family diagram at the top of this page, then read how a chatbot answers you, then try three prompts from the prompt library. After that, the free lessons in AI for Absolute Beginners are the natural next step.

Can I use the prompts from the prompt library at work?

Yes. Copy and adapt them freely. Remove names, client details and any confidential data before pasting text into an AI tool, and check your company's AI policy first.

Can I share the diagrams and charts?

Yes, for personal learning, teaching and internal training, as long as you credit TrainLearnAI and link back to trainlearnai.com/resources.

How often are new resources added?

We add new guides, prompts and glossary terms regularly. Join the weekly newsletter below to get one practical AI lesson and the newest resources by email.

Get one practical AI lesson every week

A five-minute email with one AI technique, one ready-to-use prompt and one tool tip. Free, and written for beginners.

No spam. Unsubscribe in one click. Read our privacy policy.