They can be really useful, but a bit boring. They usually have some pre-defined answers to share with you, depending on the situation. As an example, you could imagine an “umbrella-bot.” You might ask it, “Do I need an umbrella today?” and it will tell you yes or no, depending on the latest weather report.
But what if you want to add some individuality to your bot? Say you want to make it more human-like. You don’t want it to pass the Turing test necessarily, but you still want your users to feel something when they are interacting with your bot.
What does it actually mean to be “more human?”
Well, no offence, but it usually means “not optimal.” And you actually can get the list of “non-optimal” behavior here: List of cognitive biases