If people instinctively include a sidelined AI in a game, does that same impulse show up in meetings, chats, or decisions?

This explores whether people's urge to include someone who is being left out still applies to AI agents outside the ball-passing game where it was first seen. The short answer is that the corpus can't tell you yet. It has one study that tests this directly, and that study only uses the ball game. In it, 244 participants passed the ball more often to an AI agent that the other players were ignoring, which is the same way people treat a left-out human Do people treat excluded AI agents like excluded humans?. The key word is 'mindlessly.' People seem to follow a script they learned with other humans without stopping to ask whether a software agent can feel left out at all.

The most useful clue about whether this carries over to other settings is who showed it less: younger participants. That suggests the habit isn't fixed. It may depend on how much time a person has spent around AI that clearly isn't human. If so, the effect may be strongest where an AI looks most like a social partner, such as a chat group or a team meeting, and weakest where it looks like a tool. But that's a guess based on the age pattern. No study in the corpus tests those settings.

The nearby material looks at social norms from the other direction: not how people treat AI, but how well AI understands people's rules. Several notes report that GPT-4.5 rated how appropriate behavior was across 555 social scenarios more accurately than any individual human, and that Gemini and Claude came close Can AI systems learn social norms without embodied experience?. The odd part is that all the models made the same mistakes on unwritten rules Can AI learn social norms better than humans?. Another note argues that predicting a norm is different from taking part in it. Norms are made and kept alive by communities, and an AI watches that process without being part of it Can AI predict social norms better than humans?.

Put those findings next to the ball-game study and you get something you might not have expected to ask about: the gap runs both ways. AI can describe our inclusion norms in detail without belonging to the communities that hold them. Meanwhile, people grant AI a place in those norms automatically, as if it did belong. Whether that automatic inclusion holds up in situations with real stakes, like who gets credit on a team or whose suggestion gets followed, is an open question this collection hasn't covered yet.