
rumanchu |
The important thing to remember with the AI is that none of the AI options available to players (with the possible exception of a level 20 mechanic's drone) have any sort of self-awareness or actual consciousness, they're all basically just souped-up versions of Siri, Alexa, Bixby, etc. If your player wants to know what sort of questions that their AI would ask, a perfectly valid response would be "What kinds of questions do you want it to ask in this situation?" (Perhaps an even better response would be to point to your phone and ask what sort of questions Siri would ask in that situation, but that might be a bit snarky).
Another thing to remind your player is that the AI has a very limited set of skills that it can attempt tests on: Bluff, Diplomacy, Intimidate, and Sense Motive. That's it, and even those skills you have to consider through the lens of the AI initiating those test *at your request*, rather than on its own. It might be easier to think of it as effectively giving the character who has an AI-equipped computer training in those skills rather than thinking of the AI as being an NPC with those skills. Though not specifically called out in the rules, I'd also imagine that it'd be fairly obvious to someone that you're interacting with in person that your computer AI is communicating with them unless the player frames the conversation in such a way that it seems like someone is communicating remotely *through* the computer. While I wouldn't put a penalty on the test for that (unless the target has a specific reason to dislike AI or computers for some reason), I would certainly have it called out if the AI failed the test ("You insult me by having your *machine* try to convince me? Speak your own words!")
If this was in a game that I was running, I'd have the player come up with a personality for their AI and play it accordingly, as well as come up with an explanation for *how* the AI actually communicates things to him. Example: the character might have told his AI to observe any conversations that he has and advise him on whether or not the other party is being truthful (which is a pretty good justification to allow the AI to roll Sense Motive tests without the player having to specifically tell it to), but how is the result of that test relayed to the character? Is it a small light on a display (red for lie, green for truth), or does the AI just say, "They are lying" out loud for all to hear? (That would be my personal choice for the default behavior of an AI due to it being funnier). If he has a blanket "tell me if the person that I'm talking to is lying" rule (rather than calling on the AI to do so in specific situations), then have the player roll for the AI on *everything* that might apply. "Thanks for holding the door, I really appreciate it." "That's a lie, he appears to be sullen and angry that you assumed he couldn't open his own door."