Rule-based bot and Machine-learning based bots

Sometimes people like to boast they have "Artificial Intelligence" when in reality they have a bunch of "If-Else-conditions". When it comes to bots, they might be Rule-based or AI-based.

In a Rule-based approach, a bot answers questions based on some rules on which it is trained on. One of such languages is AIML (Artificial Intelligence Markup Language)

Machine-learning based bots are of two types:

Retrieval based models are trained on a set of questions and their possible outcomes. The bot cannot generate new answers.

Generative models can generate the answers and not always replies with one of the answers from a set of answers.

Machine learning bots have many pitfalls and when in doubt, a rule-based approach should be preferred.

 Now, when to use rule-based bots and when to use AI-based bots? A rule based bot only understands a pre-defined set of options, while an AI-bot can understand the intent - obviously.

Rule-based bots can be developed quickly and are cheaper to build. Therefore most bots in the market are actually rule-based.

In case of rule-based bots it has to be decided in what format to keep the rules. AIML ( Artificial Intelligence Markup Language) is a possibility, here is a tutorial.