Well I'm a combatibilist, So I certainly believe in free will. I also will accept any entity that consistently acts as if it has a will that it does actually have that. That's has always been my point, treating things as what they appear to be is the only rational approach when you cannot prove or disprove the existence of the property in question.

It follows from that that you cannot exclude something that appears to have a property if you cannot prove it doesn't or even prove it if it does.

Fair enough. I wouldn’t say a program is acting with a will of its own just because it’s trained to respond to questions in a human like way. That doesn't even say anything about its capacity to have a will. Language is a tool that can convey internal state, not the thing itself.