Exactly. They could try to use a synthesized voice, but that would probably sound fake enough to ruin the illusion of reality.Vek said:It's Seaman with an Eyetoy.
THey'd have to have tens of thousands of words, phrases, terms, and fullsentences recorded, in multiple languages, to make this more than one of those AIs you can chat with on AIM or something similar.
THis is going to be a difficult project, and if they get it right, it'll be great. But no one, not even dedicated AI researchers, have been able to accomplish what this game says it will. Make an interesting, unique, learning AI with audio-visual user recognition, and make it react engagingly and realisticly.
At some point, storage and hardware limitations are going to meet up with the Uncanny Valley, and really drag this game down.
And that's really the problem - any small hiccup, no matter how small, would completely ruin the illusion. It's infeasible to try to account for every situation in a perfectly realistic way.