Years ago, I had the pleasure of working with a very talented user experience engineer who was very worried about long-term affects of using computers on his body. From his head to his toes. So, he used every possible input device & medium to avoid over-using any one part of his body. Needless to say, he used the most ergonomic chair and desk available and he had them configured at the right hight and angle.

He also used voice commands to start applications, type, start a new email, etc. He had mounted a retrofitted mouse to the leg of his desk, so he could use the side of his left thigh to give his right hand a break.

You get the picture.

Anyway, I sat in the cube next to this colleague, and I couldn’t help but be distracted by his voice commands, especially in situations where he had to repeat the same command several times or would have to back-out and use a different command. This went on all day, I mind you.

Why am I writing about this? Because I just spent some time digging through Kozoru’s site to learn more about their technology, and found out that one of the main features is that the user can use voice to submit his/her question to be answered by Kozoru! As shown in one of the demo videos, you could even do this from your mobile phone! Sounds great, eh?

Not so quick.

IVR is a very very old telephony technology that in the past couple of decades was outfitted with natural language recognition capabilities. In my opinion, just recently the technology has finally become good enough to rely on. It’s still annoying to use and you look like a lunatic when interacting with one on a busy street (note that the only driver for wide use of IVR systems is to save on customer service costs). The experience using Kozoru from a mobile phone cannot be much different; may be even worse because you would have to speak-very-clearly-and—-slooowly.

How would you react if they guy sitting next to you on the train, while starring at his mobile phone, started asking it a question? And yes, you would definitely be able to tell the difference between him talking to someone or a search engine because he’ll be looking at the screen for visual feedback, in the later case.

It would be a bit strange, wouldn’t it? ‘Talking’ to a computer, no matter how smart the app that’s ‘listening’ to you, still seems strange, right? I think so. I am not sure if HAL 9000 really is the right way to go here, folks. May be we can have a very smart search engine that can answer our questions. Period.

Skip the voice thing, no? I mean, it was impressive that HAL 9000 understood natural language, but that lost its novelty pretty quickly, didn’t it? It was HAL’s brain that kept haunting you! By the way, while we are on the subject of HAL, let’s do this right and have Kozoru eventually interpret emotions, express emotions and reasoning. Oh, and it should also read the answers back to the user as that would be quite useful if using the service on a mobile phone while driving. Seriously, these features would put this search engine way ahead of everything else in the market.

I also wonder how well the engine would handle users with an accent? Well, we will find out in April of 2006 as that’s the promised launch date for Kozoru.

Don’t get me wrong, I am not trying to dis Kozoru or mock their vision. I am sure we are all going to be impressed by the accuracy and timeliness of the answers to our questions, but why not put the emphasis on that instead of the voice interface in demos?