Labelling humans as intelligent is a by-product of times gone by when, even if we were secular, we still thought we were the centre of the universe.
It's easy to get the impression that we actually are intelligent when one spends time on Freethought forums, but a quick survey of the global community puts that idea to rest. People probably have a small amount of cognitive capacity that other animals don't have, which makes us able to follow small logic chains, use math, and store a lot of data in our memory. This looks exceptional because the ability to do basic math and internalise causation multiplied over hundreds of thousands of years produces some exceptional technologies, but as individual people we're not too far removed from some of our closer relatives.
In reality, most of our lives are spent using cached responses when we talk to other people so we don't look like idiots and get outcast from our tribes, bouncing around trying to convince people to give us money, and trying to get laid. There are likely some very real neural components that ensure we do those things successfully. Outside of that, all of our 'thinking', 'philosophizing' and so on are just ways to pass the time. So in a sense this can be thrown back to the OP, because giving the impression that we're competent is probably more important than actually being competent. The only way to really gauge that is social performance.
Anyway, I'm not convinced that a simple AI bot in a game is in anyway different than humans themselves. A character in a game works on a set of conditionals, and so do humans. The only difference between a human and a bot, is that humans can perceive many more conditions, and they can respond in many more ways. So maybe the implication of the OP is that a human only needs to perceive a finite number of conditionals, and use a finite number of responses to be perceived as 'intelligent'.