Last reviewed:
What is a chatbot? Definition, examples and difference with an AI agent
A chatbot (or conversational agent) is software that converses in natural language with a human, in writing or by voice, to answer questions or trigger tasks. Since August 2, 2026, the EU AI Act requires informing users that they are interacting with an AI, and the company remains responsible for what its chatbot says.
The French Official Journal of December 9, 2018 recommends the term "dialogueur" (or "agent de dialogue") and defines it as "software specialised in natural language dialogue with a human", able to answer questions or trigger the execution of tasks. Everyday usage sticks with "chatbot" or "conversational agent". Three generations coexist. The rule-based chatbot follows a decision tree written in advance: buttons, keywords, prepared answers. It is predictable but stumbles on any unforeseen question. The LLM chatbot, popularised by ChatGPT from November 30, 2022, understands freely worded questions and writes its answers. It is often connected to the company's documentation (RAG). It can also invent a plausible but false answer. Finally, the AI agent does more than answer: it acts within systems (changing an order, opening a ticket). Two rules frame their use in Europe. Article 50 of the AI Act, applicable since August 2, 2026, requires informing people that they are interacting with an AI system, unless this is obvious. And case law is starting to settle liability: in February 2024, a Canadian tribunal ruled that an airline is responsible for information given by its chatbot just as for any page of its website.
Concrete example
Real case: Moffatt v. Air Canada, decision of the Civil Resolution Tribunal of British Columbia of February 14, 2024. In November 2022, after his grandmother's death, a customer asks Air Canada's chatbot about bereavement fares. The chatbot tells him he can apply for the reduction retroactively, within 90 days of the ticket being issued. The official page of the website says the opposite. Air Canada refuses the refund and argues before the tribunal that the chatbot is a separate legal entity, responsible for its own actions. The tribunal calls this a "remarkable submission" and rejects it: the company is responsible for all the information on its website, whether it comes from a static page or a chatbot. Air Canada is ordered to pay CAD 812.02 (damages, interest and fees).
Illustrative case: a French online retailer with 30 employees deploys an LLM chatbot connected to its terms of sale and order tracking. It displays a message "You are chatting with an automated assistant" and offers a button to transfer to a human adviser for any dispute.
Comparison
| Rule-based chatbot | LLM chatbot | AI agent | |
|---|---|---|---|
| How it works | Decision tree, keywords, answers written in advance | Language model that understands and writes, often connected to your documents (RAG) | Language model that plans and uses tools |
| Unforeseen questions | Fails or returns to the menu | Answers, sometimes wrongly | Answers and looks up information in your systems |
| Can act | Only predefined actions | Rarely | Yes: orders, tickets, emails |
| Main risk | User frustration | Invented answer (hallucination) | Wrong action with real consequences |
| AI Act transparency (Art. 50) | Covered if it is an AI system | Yes | Yes |
FAQ
What is a chatbot?
Software that converses with a human in natural language, in writing or by voice, to answer questions or trigger actions. It is found on websites, in apps, on messaging platforms and on the phone.
What is the difference between a chatbot and a conversational agent?
In everyday usage, none: they are two names for the same kind of software. The real distinction is between rule-based chatbots, LLM-based chatbots and AI agents.
What is the difference between a chatbot and an AI agent?
A chatbot answers within a conversation. An AI agent pursues a goal over several steps and acts within your systems: it queries a database, changes an order, sends an email. A chatbot can become an agent once it is given access to tools.
What are examples of chatbots?
ChatGPT, Claude, Mistral's Le Chat or Gemini for general-purpose assistants. In business: a customer service assistant on an online store, an HR chatbot answering leave questions, an appointment booking assistant, an internal IT support bot.
Is it mandatory to disclose that a chatbot is an AI?
Yes, in the European Union. Article 50 of the AI Act, applicable since August 2, 2026, requires that people be informed they are interacting with an AI system, unless this is obvious to a reasonably well-informed person.
Is a company liable for its chatbot's mistakes?
Yes, according to Moffatt v. Air Canada (British Columbia tribunal, February 2024): the chatbot is part of the company's website, and the company is responsible for all the information on it. No comparable French ruling is known to date, but consumer law also protects customers against misleading information.
See also
Further reading
Moffatt v. Air Canada, 2024 BCCRT 149, full decision, Civil Resolution Tribunal of British Columbia
Sources
- Vocabulaire de l'intelligence artificielle (list of adopted terms and definitions), French Official Journal no. 0285, December 9, 2018, Légifrance. https://www.legifrance.gouv.fr/jorf/id/JORFTEXT000037783813
- Moffatt v. Air Canada, 2024 BCCRT 149, Civil Resolution Tribunal of British Columbia, February 14, 2024. https://decisions.civilresolutionbc.ca/crt/crtd/en/item/525448/index.do
- Regulation (EU) 2024/1689 on artificial intelligence, Article 50 (transparency obligations), AI Act Service Desk, European Commission. https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50
- Introducing ChatGPT, OpenAI, November 30, 2022. https://openai.com/index/chatgpt/