kars on 11 March 2007
stewacide said:
T but with massively more processing power (which should be available by the next gen) there's a lot new you could do in terms of physics and AI that's not feisible today. E.g. apply real physics to *everything* in the game world (not just a select small number of things), apply much smarter AI techniques (e.g. recursion, etc.)
Well, in the case of physics and AI you wil probaly even reach limits on the current NG-Consoles, not in the way of what can i get, but what do i want. You are not really interested in what does happen under this circumstances, but more in how does it look, feel, when i do this.
And in an AI: Yo don't have real problems in how to give an AI perfect capabilities, but you don't really want these perfect capabilities. You want limited capabilities that look realistic. But there you will find real problems with AIs. There is no single program, that would really pass the turing test. It does not matter how smart a single AI can be, you will have to run multiple AIs at the same time. It doesn't really matter, what you do: players will always find the limitations of your behaviour models, or be doomed and frustrated by their superhuman capabilities.
stewacide said:
I don't think will become central (e.g. it adds an interesting element to things like Manhunt and Brain Age, but I don't know how you'd design a whole game around voice recognition, or if you'd want to)
Well, as an example if you woiuld have a space flight simulator you could give commands to your virtual colleagues on board of the ship. It would ba more realistic if they would report to you and you don't need to press a button in a multiple choice menu. You could have a bigger number of options than you would have buttons on your controller.
