The Next Stage in Computing
Compared with the way we view other major technologies, our attitudes about
computers are an anomaly. By almost every critical measure - performance,
reliability, price, functionality - computers continue to improve at an
astonishing rate. If airplanes were on the same trend line, we would be
flying around the world at supersonic speeds, on teacups of fuel, for
pennies a flight. Yet, we continue to expect even more from our computers.
Clearly, we judge them by a different standard. Accustomed as we have
become to a wealth of new applications and lower costs made possible by
higher levels of integration, we tend to focus on a quite different set of
issues. What excites us is not the machines themselves, but how we interact
with them.
Fictional computers, like HAL in 2001, helped set high expectations.
Today, many people wonder why, if computers are so smart, we are still not
able to talk to them and be understood as naturally and completely as we
would by another human being. Why not, indeed?
It is a worthy goal. In fact, speech is just one of the modalities by which
we communicate, and our interactions with our computers should be just as
broad. Our researchers agree, and they are working on the technology to
make that natural interaction a reality.
Speech recognition technology has already yielded products like IBM's
ViaVoice Gold that prove we can talk naturally to our computers and have
our words captured and displayed on-screen. Great as that accomplishment
has been, it is only a beginning. In our labs, we are seeking to make our
computers understand, or at least appear to
understand, the meaning of what we say - and do.
Whether by watching our hands as we gesture at on-screen objects, following
our gaze to anticipate what we want to do, or interpreting our feelings by
our looks or the inflections in our voices, computers stand ready to become
far more accommodating. These future capabilities, some not so far distant,
will make computing easier, more productive and, ultimately, far more
satisfying. Eventually, the human-computer interface will become
effectively transparent. No longer will we worry about how to interact; we
will simply act normally and leave the rest up to our computers.
Research's major efforts in natural computing are the subject
of the cover story of this issue. I
expect great things from this work. The simplifications it promises will
lower the perceived and real barriers to using computers and extend the
power of computing to almost everyone, a result from which we will all
benefit.
Paul M. Horn
Senior Vice President,
Research
