Ask HN: Why haven't we seen AI as opponents in video games? | Hacker News Reader

Ask HN: Why haven't we seen AI as opponents in video games?

5 pointsduxup6 comments

Civilization used to be my favorite video game series. I played it since Civ I and as much as any software it came along for the ride with my interest with computers. But around Civ VI the competition provided by the computer opponent seemed to fall apart completely. Often the computer just rage quits and doesn't build a military, declares war and does nothing, and so on.

This makes me wonder why we haven't seen AI as a gaming opponent much. I suspect even a flawed AI would at least do something. Gaming AI opponents seem like a natural place for something like AI where they could provide a lot of unexpected fun and "interesting decisions".

Is there something about the current AI landscape that makes it incompatible, difficult, or unprofitable?

View on HN

Assuming you're referring to a "smart computer opponent", it's probably a mix of many factors. Indie gamedevs usually won't have the resources to implement cutting-edge technology; they just go with whatever standard gamedev toolkit supports and focus on the gameplay, story, art, or other aspects. Even a studio with the resources to do so might struggle to find where it should fit into the gameplay.

Contrary to what gamers usually think, making an impossibly difficult computer opponent is trivial. Give a Counter-Strike/Valorant/FPS bot a wallhack & aimbot and everyone will compete on kicking each other in the match rather than playing the game. Making a computer opponent that's fun to play against, on the other hand, is hard and not even a guaranteed hit for the game.

I don't think current "AI" could do well in standard video game context. It's way too computationally hungry for most of local machine and it's probably hard to "program" it the way game designer envision it to be.

I'd be interested in seeing AI dialog in video games or game-adjacent experiences.

In particular, VR worlds could really use an NPC you share the space with and have a conversation with like you do in real life or in VRChat. One reason Horizon Worlds is so daft is that if a business wanted to create an experience where you could interact meaningfully business it would have to staff the space with a real employee.

I'm not sure if I would trust an LLM to play a major character ("you are dan!"), but could it be like one of the cast members at Disney or Colonial Williamsberg, I think yes.

Lots of games use AI for their opponents. One example I recall fondly is F.E.A.R. which used[1] Goal Oriented Action Planning for enemies.

It was the first first-person shooter I played where the enemies actually were tactically challenging. Sit and camp a spot and they'd find a way to flank you.

So, I suspect you mean something particular when you say AI. What are you thinking of?

[1]: https://www.gamedeveloper.com/design/building-the-ai-of-f-e-...

> This makes me wonder why we haven't seen AI as a gaming opponent much.

Wat? Chess?

Video game opponents are absolutely AI in the "good old fashioned AI" (GOFAI) sense - they have very clearly defined objectives and action spaces, and algorithms like A* pathfinding [1] and Goal Oriented Action Planning [2] which perform planning over a space of possible sequences of actions to achieve a specific goal. The Game AI Pro [3] articles online give a good picture of the kind of implementation decisions which go into game AI.

The problem is that a lot of the time, bruteforce search can beat the snot out of any human player, but the real game design objective isn't "beat the player", it's "give the player enough of a challenge to make beating the AI fun".

In Civ's case, it might be theoretically optimal play for the computer to capitalize on rushing players with warriors before they have a chance to establish defenses, but it is also a great recipe for players to angrily request refunds after the tenth consecutive round of being crushed by Gandhi in turn 5. A lot of game AI development time goes into tweaking action probabilities or giving the player advantages to counteract the AI advantage - the reluctance to build military units you saw could have been the result of such a tweak.

As for why LLMs typically aren't applied as game opponents:

* They are quite compute intense, which is tricky when players expect at most 16 ms latency per frame (for 60 FPS), and get ornery if they have to wait more than a few seconds, but also do not like having always-online requirements imposed by cloud compute (or subscription costs to fund running LLMs for every player)

* The bridge between tokens and actions also means it's hard to tweak probabilities directly - while A* can let you specify that a certain path should be taken approx. 20% of the time, implementing this in an agent-LLM approach means you have to actively select and weight token probabilities during the beam search, which is a bit of a hassle, to put it mildly

* The issues with long-term coherence in LLMs, famously demonstrated by Vending-Bench [4], makes reliability and debugging harder

[1] https://en.wikipedia.org/wiki/A*_search_algorithm

[2] https://www.gamedeveloper.com/design/building-the-ai-of-f-e-...

[3] https://www.gameaipro.com/

[4] https://arxiv.org/abs/2502.15840

a) too computationally intensive and b) unfun for the player

Reply on news.ycombinator.com