Article
files/machine

Are You Satisfied with Improvements in AI Technology?

Alex J. Champandard on October 16, 2007

This week’s developer discussion is about technological improvements in the game development industry, following up on last week’s discussion about visible progress of in-game AI.

Remember, you can win an AiGameDev.com T-Shirt at the end of the month for interesting comments, so join in by posting below!

  • Do you think professional developers have made significant technical progress over the decades in game AI?

  • How do you rate the practical contribution of the academic community in the same time period?

Don’t hesitate, join the discussion.

Game Characters Chatting

Next week’s discussion: pushing back the uncanny valley.

Discussion 3 Comments

Ian Morrison on October 17th, 2007

I'm not sure about that. After all, take the first person shooter. Sure, there have been important innovations in the last few years, and each game at least makes an attempt to be different, but within the various sub genres (tactical shooters, Unreal Tournament style run and gun, etc) you end up looking at *basically* the same gameplay. Smaller parameters change, yes, but the overaching themes, skillsets and challenges remain.

diegix on October 18th, 2007

I don't think the goals of Academic AI and Game AI are different. Of course there are some research in the academic that are not at all applicable to games or not at the moment for performance reasons. I think the goals are basically the same, building believable and realistic AI. Probably the methodology to reach that goal is different, academic work requires theoretic support which AI in the game industry doesn't require at all. I have read tons of AI papers that have a similar goal as I could have for a game, but I couldn't extract anything really practical from it, because it focused on the theory, or on proof that this high level method is optimal in some situations or things like that. I have been trying to find research groups working in AI trying to apply it to a game scenario and I'd be more than happy to know more groups working on such things. Anyways, to the point of the discussion. I think that AI technology has improved a lot in these years, but the problem in my opinion is that is not established as a technology yet. There are some middleware solutions for AI already and you can certainly build an AI engine for you game and reuse it in similar and not so similar games in the same way you reuse the animation system or the entity system in your game. However, I think this is rarely done since there is still the impression that AI is too dependent on the game itself. We often start the AI from scratch for a new game, maybe with the exception of sequels, and even though we are advancing and improving we could improve more if we stop doing that.

alexjc on November 19th, 2007

Welcome [B]Jim[/B]. Thanks for your questions... I'll post more about it on [URL=http://aigamedev.com/questions]Questions[/URL] Friday, but here's the redux: 1) Yes, definitely. 2) It's subject to discussion: [URL=http://aigamedev.com/discussion/most-difficult]What's Most Difficult about Game AI?[/URL] 3) That's a great question; one for another friday I think... but keep in mind that it's not only about typing in code. With the experience, you can rebuild certain system in a fraction of the time, much better than the previous iteration. Of course, it still takes time :-) Alex

If you'd like to add a comment or question on this page, simply log-in to the site. You can create an account from the sign-up page if necessary... It takes less than a minute!