A friendly first lesson · No technical background needed
AI, in words
you already know.
ChatGPT. Claude. Gemini. So many names. Let’s slow down and understand what they are, who makes them, and how you can start.
Start from zero ↓Explore the 20 namesRead one section at a time. You do not need to memorise the names.
01 / First, the basics
What is AI, really?
AI means Artificial Intelligence. It is computer software trained using examples, so it can do jobs such as answering questions, writing a message or recognising things in a photo.
Think of it as a helper you can give instructions to. It can be useful, and it can also make mistakes. It does not know everything.
Three different things. One simple example.
The company
The people and business that make the AI.
Like the car maker.
Example: OpenAI.
The model
The AI engine that does the work behind the scenes.
Like the car’s engine.
Example: a GPT model.
The app or tool
The website or app you open and use.
Like the car you drive.
Example: ChatGPT.
A useful detail: An app can use several AI models and tools. The engine may change while the app keeps the same name. GPT-3 is an older model; today’s ChatGPT does not mean “GPT-3”.
And what is an AI “method”?
A method is a way to build or teach AI. Think of a recipe. A recipe is not a restaurant you can visit.
GAN and Convolutional Autoencoder are methods. They are not chat apps like ChatGPT.
02 / Something you can actually open
Start with one app.
Choose one. Type a question in normal language. There is no need to learn all 20 models first.
ChatGPT
An assistant you can ask to explain something, suggest ideas or help with writing.
Open ChatGPT ↗Claude
An assistant you can ask questions, give writing to, or ask for help improving a draft.
Open Claude ↗Gemini
An assistant for questions, explanations and creative tasks. Available features can vary.
Open Gemini ↗These links open other websites. Accounts, plans and available features differ. Using these apps does not require coding.
A little picture of the possibilities
Words, pictures and a little imagination.
AI can do different kinds of jobs. These two examples make the difference easier to see.
Help with words and ideas
“Help me write a friendly invitation.” A language model can suggest a draft for you to read and change.
Understand language models →
Help with pictures
“Where is the bicycle?” An object detector can find it and show where it appears in the picture.
Understand object detection →Illustrative examples to help you learn. These pictures are not results from the named models.
03 / Make sense of the names
Your 20 names, explained gently.
Some are engines. Some are designs or methods for building AI. Some names are also used for apps. Open any name to see who made it, what it is, an example, and where to go.
You are learning to use AI. You can leave the developer details for later. A research link explains a model; it is not always a tool you can open and chat with.
01BERTopicOrganising text
Read the full BERTopic beginner lesson →
Who made it? Maarten Grootendorst.
What is it? Software that puts writing about similar subjects into groups.
How would I use it? A developer usually sets it up. It is not a chat app.
Official project ↗02GPT-3Language engine
Read the full GPT-3 beginner lesson →
Who made it? OpenAI.
What is it? An older AI engine that can understand and write text.
How would I use it? For an easy app, try ChatGPT. ChatGPT uses newer models, not the original GPT-3.
Try ChatGPT ↗03GANAI building method
Read the full GAN beginner lesson →
Who made it? Ian Goodfellow and fellow researchers.
What is it? A way of teaching AI to create things. One part makes something; another checks it.
How would I use it? GAN is a method, not one company or one app. Developers build tools with it.
Original research ↗04Convolutional AutoencoderPicture method
Read the full Convolutional Autoencoder beginner lesson →
Who made it? A general research method; no single company.
What is it? A way to teach AI to rebuild a picture after turning it into a smaller description.
How would I use it? There is no single official chat app with this name. Developers build it into software.
Learning example for developers ↗05ResNetPicture engine design
Read the full ResNet beginner lesson →
Who made it? Microsoft researchers.
What is it? A design for AI that can learn to recognise what is in a picture.
How would I use it? Usually works inside another app. You do not need to learn its design to use AI.
Microsoft research ↗06VGGPicture engine design
Read the full VGG beginner lesson →
Who made it? University of Oxford researchers.
What is it? Another design for AI that learns to recognise pictures.
How would I use it? A building block for developers, rather than an everyday chat website.
Oxford research ↗07Gemini NanoSmall device engine
Read the full Gemini Nano beginner lesson →
Who made it? Google.
What is it? A small AI engine designed to work on devices such as supported phones.
How would I use it? It works inside supported features and apps. The Gemini website does not mean you are using Nano.
Google’s explanation ↗08XLM-RoBERTaLanguage understanding engine
Read the full XLM-RoBERTa beginner lesson →
Who made it? Facebook AI researchers, now Meta.
What is it? An AI engine that helps software understand writing in many languages.
How would I use it? Mostly for developers. It is not a ready-made assistant you chat with.
Official model page ↗09PaLMOlder language engine
Read the full PaLM beginner lesson →
Who made it? Google.
What is it? An older family of AI engines that understands and writes language.
How would I use it? Google’s everyday assistant is now Gemini, a different model family.
Try Gemini ↗10YOLOObject finding engine
Read the full YOLO beginner lesson →
Who made it? Originally Joseph Redmon and fellow researchers; later versions by different teams.
What is it? AI that spots objects and shows where they are in a picture or video.
How would I use it? Developers use it in camera and image software. Ultralytics makes several later versions.
Ultralytics YOLO ↗11DeepSeekLanguage engine family + app
Read the full DeepSeek beginner lesson →
Who made it? DeepSeek.
What is it? The company, its AI engines and its chat assistant share the DeepSeek name.
How would I use it? You can use DeepSeek Chat directly. The model is the engine behind the conversation.
Try DeepSeek Chat ↗12Microsoft PhiSmall language engine family
Read the full Microsoft Phi beginner lesson →
Who made it? Microsoft.
What is it? Smaller AI engines developers can put inside their own software.
How would I use it? Mainly a building block. Phi and Microsoft Copilot are not the same product.
Microsoft Phi ↗13LlamaLanguage engine family
Read the full Llama beginner lesson →
Who made it? Meta.
What is it? AI engines that developers can use to build assistants and other apps.
How would I use it? Different apps can use Llama. There is no single required Llama app.
Meta’s Llama page ↗14MistralCompany + engine family
Read the full Mistral beginner lesson →
Who made it? Mistral AI.
What is it? A company that makes AI engines and an assistant people can use.
How would I use it? Its assistant is called Vibe, previously Le Chat. Product names can change.
Try Mistral’s assistant ↗15Claude Instant 1.2Older language engine
Read the full Claude Instant 1.2 beginner lesson →
Who made it? Anthropic.
What is it? An older engine in the Claude family. Claude is also the name of the assistant app.
How would I use it? You can try today’s Claude app. That does not mean you are using Instant 1.2.
Try Claude ↗16Gemini OmniCreative video model
Read the full Gemini Omni beginner lesson →
Who made it? Google DeepMind.
What is it? AI for creating and editing video from instructions and other material.
How would I use it? Google lists Gemini and Flow as places to try it. Access depends on the service and plan.
Official page and tools ↗17NVIDIA NemotronAI engine family
Read the full NVIDIA Nemotron beginner lesson →
Who made it? NVIDIA.
What is it? AI engines developers can use to build assistants and software that carries out tasks.
How would I use it? Mainly models and demonstrations for developers, rather than one simple consumer app.
NVIDIA Nemotron ↗18Sarvam 30B / 105BLanguage engine family
Read the full Sarvam 30B / 105B beginner lesson →
Who made it? Sarvam AI, an Indian company.
What is it? AI engines with a focus on Indian language use. 30B and 105B describe model size.
How would I use it? Sarvam introduced Indus as an assistant for experiencing its models. Check its website for access.
Meet Indus ↗19ERNIEAI engine family
Read the full ERNIE beginner lesson →
Who made it? Baidu.
What is it? AI engines from Baidu, a Chinese technology company.
How would I use it? Baidu offers an ERNIE / Wenxin assistant. Language and account requirements may vary.
Baidu’s assistant ↗20EfficientNetPicture engine family
Read the full EfficientNet beginner lesson →
Who made it? Google researchers.
What is it? AI designs that recognise pictures while using computing power efficiently.
How would I use it? Usually a building block inside an app. It is not a chat website.
Google research ↗“Any version” means you may choose one member of that family, such as ResNet-50. “B” in 30B means billion parameters: roughly, the adjustable numbers inside a model. Bigger does not automatically mean better for your task.
04 / Learn by doing
Talk to it like a helpful colleague.
Your instruction is called a prompt. That simply means the message you type. A useful prompt says what you want, gives a little context and asks for a clear format.
“Write an email.”
The AI has to guess who it is for and what you want to say.
Make the request clearer
“Write a polite email to my colleague asking for the project update by Friday. Keep it under 80 words.”
Give it a small, specific job.
“Explain project risk to a beginner. Use the example of planning a birthday party. Keep it to five simple points.”
Job: explain risk. Example: a birthday party. Format: five simple points.
Three examples you can copy and try
📚 Help me learn something
Then say: “Make it simpler” or “Give me another example.” Learning can be a conversation.
💡 Help me think of ideas
Then say: “I like idea 2. Give me three different ways to do it.”
📝 Help me with a project task
Then review the checklist yourself. Add the details only you know about your event.
These are sample prompts, not live AI responses. Copy one into an assistant to try it.
05 / Your first week
Ten minutes a day is a good start.
You do not need a technical course to begin using AI. Start with one tool, try small tasks and notice when the answer needs improvement.
Understand the three names
Learn company, model and app using the car example above.
Your small task: Say out loud: “OpenAI is the company. GPT is an engine family. ChatGPT is the app.”
Open one assistant
Choose ChatGPT, Claude or Gemini. Ask it to explain a familiar topic.
Your small task: Try: “Explain AI using the example of a kitchen.”
Ask a clearer question
Give the assistant a job, a little context and the format you want.
Your small task: Try: “Write a friendly three-line invitation to a team lunch on Friday.”
Have a conversation
Ask it to change the answer instead of starting over.
Your small task: Try: “Make that shorter.” Then: “Use simpler words.”
Check the answer
Choose a topic you know. Look for errors, missing details or confident guesses.
Your small task: Ask: “Which parts should I verify?” Check important facts using a trusted source.
Try a real small task
Use AI to help with a checklist, a message or a study explanation.
Your small task: For PMP study: ask for a simple example, then compare it with your study material.
Explore one more name
Read two or three entries in the directory. If you want, try a second assistant.
Your small task: Give both assistants the same request. Decide which answer helps you more.
06 / A little confidence check
Do the basics make sense?
Choose an answer. You can try again as often as you like.
Three quick questions
Show the answers and explanations
2. Start with one assistant. You do not need all 20 models.
3. Check important facts. An AI answer can be wrong.
Good habits from day one
A few helpful habits.
- Check facts that matter. AI can invent names, dates, sources and answers.
- Keep private details private. Use made-up examples instead of passwords or confidential project information.
- Review before using. Read a message, plan or explanation before relying on it.
- Use it to learn. Ask “Why?” and “Give me an example”, then explain the idea in your own words.
For PMP learning, compare AI explanations with official PMI resources and your study materials. This AI introduction is general learning, not an official list of exam requirements.
A tiny word guide
Five words. That’s enough.
AI: Software trained to do tasks such as writing or recognising pictures.
Model: The engine doing the AI work.
App / tool: The website or software you use.
Prompt: The instruction or question you type.
Training: The process of teaching a model using examples.
Sources, model names and updates
Reviewed on 3 November 2026. Each directory entry links to an official project, maker’s website or original research. App names, access and models can change.
- Official OpenAI documentation: asking and refining questions
- Anthropic: Claude Instant 1.2 is a specific older model
- Mistral: Le Chat is now Vibe
- Google DeepMind: Gemini Omni
- Sarvam: 30B and 105B models
- Google: PaLM 2 research
This is an independent educational guide. Named companies and products belong to their respective owners. Examples are illustrative; individual abilities depend on the version, training and app.