AI Surrender issues

For topics that do not fit in another sub-forum.
Message
Author
User avatar
Dilvish
AI Lead, Programmer
Posts: 4713
Joined: Sat Sep 22, 2012 6:25 pm

Re: AI Surrender issues

#16 Post by Dilvish » Fri Mar 16, 2018 7:09 pm

Vezzra wrote: Like I said above, I think Geoff is concerned with ideas that would make an AI do things a human player would not. E.g. when surrendering and deciding which opponent to gift my ships to, or if I should scrap all my ships, as a human I couldn't care less about how much influence my opponents invested in swaying me. AIs should work the same.
A human's decision is always going to be at least somewhat mysteriously unpredictable, whereas the AI's is always going to be either random, deterministic, or some known mixture of the two (most likely the latter, which we will try to make complex enough to be an enjoyable simulacra of diplomacy). Anything not fully random will to some extent have "gameable" elements potentially subject to possible manipulation by another player (whether either player be human or AI-- and yes, I will assert that to some limited degree human players could be subject to diplomatic manipulation by an AI player).

My understanding of our plans for Influence does not make it seem like something that would be so extremely "gameable" that it would be a problem for the AI to take Influence into account; to me, for my understanding, it actually seems like it would fit fairly well for it to be at least a consideration. Perhaps I am just not understanding the plans for Influence properly. I suppose that any such planning for what consideration the AI might give to Influence can be put off for the actual Influence planning, and need not be resolved just now.
If I provided any code, scripts or other content here, it's released under GPL 2.0 and CC-BY-SA 3.0

Post Reply