Approximate
From what I understand, their understanding comes down to probability:
Machines and systems built to converse and carry-out actions based on the spoken word. The advance that is pulling them closer to reality? Probability.
They learn and remember, then try to guess the phrase you are most likely to use next. From that, a strange sort of understanding arises.
Supposedly, by 2010, there will be speech recognition systems that can transcribe the spoken word better than humans bent to the same task.
I wonder what understanding like that would (or will) be like?
Knowing by approximation.
Building a mystery, as Sarah sings.
I wonder if it is like music? Sometimes certain notes feel right following others. Sounds that fit together like a smooth pretty puzzle.
I wonder how far ahead a machine or program could guess correctly?
When processor speed goes from breeze to hurricane can prediction blur into precognition?
They say it works better with faces. Vision. If the system can watch while you speak, its accuracy increases considerably.
At some point, will you stand before a fast construct, open your mouth to speak, and have it gently tell you:
“I understand.”
Approximately.
July 9th, 2003 at 9:38 PM
That inspires and frightens me. I guess ‘awes’ is a better word.
I love Sarah. Where were you when I was single? *grin*
Thanks for the plug in your sidebar. You are my favorite online writer at the moment, because, John, you are so a poet in disguise.
Feith
July 9th, 2003 at 10:09 PM
Thank you, Feith!
(I don’t know about poethood but I do promise not to share any of my limericks with the world!)
As far as speech recognition goes, things really are getting amazing. There probably will be some pretty nifty displays of that in the coming years.
Or maybe they’ll just end up being like trying to talk to a cranky, half-deaf uncle:
“You want to do what with WHAT?”