Map the terms.
The broad field: systems designed for tasks associated with intelligence. Some use explicitly programmed rules.
Methods that fit models using data.
Machine learning using multilayer neural networks.
Where do chatbots fit?
Many modern chatbots use deep-learning language models. A chatbot is an interface or application; it may also combine rules, search and other software. Not every AI system is a chatbot.
AI in everyday language
Think of AI as computer software that can perform particular kinds of intelligent-looking tasks. For example, a phone may recognise spoken words, or an email service may help identify unwanted messages. That does not mean the software has a human mind.
Does AI actually learn?
Some AI systems use rules written by programmers. Many modern systems instead learn statistical patterns from training examples. Learning here means adjusting a mathematical model, not studying or understanding the world in exactly the way a person does.
Is AI always right?
No. AI can make mistakes, reproduce biases in its training data and generate convincing but incorrect answers. Important claims should be checked using reliable evidence.
CHECK THE ORIGINAL MATERIAL
Sources & further reading
Starting points for the explanation and the illustrative interface above. Sources are external; examples are not CLU capabilities.