suraj@jhunix (Suraj C. Surendrakumar) (10/04/89)
IS AI HEADING THE RIGHT WAY TODAY ?
WHAT DO WE NEED TO DO TODAY FOR THE FUTURE OF AI TOMORROW ?
Today Artificial Intelligence has become a glamorous field in Computer
Science. AI is great and revolutionized not only computer science but
also many other related fields. Applications and interest of this field
has given a boost to AI. Today there is a lot of research being done in AI.
A number of expert systems have been developed and other applications
have been found for Artificial Intelligence. But still there are many aspects
that are totally lacking in AI.
When the first conference on AI was held in the 50's it was said
that it would be possible to unravel the mysteries of the human mind and
program a computer to do the same in a few years. As time has gone by
it has been clear that this dream is far away even today. AI scientists
have developed a large number of inference systems, shells, representations
etc.. These systems are very specialized to their applications and have been
built for a specific job. They are inflexible to other applications in other
fields. The dream of every AI researcher is to have a system which is very
human like. A system which not only tells where to drill for oil or how to fix
an engine but ONE BASE SYSTEM LIKE THE HUMAN BRAIN THAT WITH THE SAME PROGRAM
CAN BE TRAINED TO DO MANY THINGS ON THE SAME MACHINE, AND CAN ALSO RELATE ITS
EXPERIENCE FROM ONE TASK TO THE OTHER WITHOUT BEING TOLD. WE NEED A SYSTEM
WHICH CAN LEARN FORM MANY FIELDS WITHOUT BEING TOLD TOO MANY THINGS AS WE DO
TODAY. The system today is limited to learning just a little more than it is
told and its rules allow.
This is not a easy or a small project. It will require a lot
of work. The answer to a lot of question for this project can be found by
studying cognitive science. After all the human brain - body is the only
perfect intelligent systems around us. Over the years AI scientists who have
been ignoring cognitive science are realizing that cognitive science has
a lot to offer. REMEMBER THE HUMAN BRAIN IS THE MOST INTELLIGENT SYSTEM
EXISTING TODAY. We will also need some revolutionary hardware to get to
this point. The computing needs of such a project will be enormous and at
a time when current technology is reaching limits in terms of speed. We may
need to develop specialized machines, look for new faster technology or who
knows our study of the brain and cognitive science may bring break throughs
that can bring faster implementations. The software of course will be
gigantic and will require enormous manpower and resources. But the results
of such a project will be worth the effort.
In conclusion I want to say that the shells and systems we have are
barely doing their work today. We need to build a system in the near future
that will be a TRUE ( not artificial ) intelligent machine. To do this
sort of a thing we have to get our strategies together today and start thinking
about it. A lot of want we want know can be got be understood by unrevealing
the brain and getting input from cognitive science and dropping the notion
the cognitive science and AI have little in common. If you want to climb over
the saturation level AI is reaching today and do this sort of a thing tomorrow
start thinking about it today.
SURAJ C. SURENDARKUMAR.
HOME: suraj@jhunix.hcf.jhu.edu ( Best for correspondence )
BITNET: SURAJ@JHUNIX
ARPA: suraj%jhunix@hopkins.ARPA
UUCP: allegra!hopkins!jhunix!suraj ( uucp not very reliable )
Also on Bitnet:
SURAJ@JHUVMS SURAJ@JHUVM SURAJ@COGSCI DAT_50@JHUNIX
suraj@cogsci.cog.jhu.edu
*****************************************************************************
* SURAJ C. SURENDRAKUMAR - The Johns Hopkins University *
* COGNITIVE SCIENCE CENTER and COMPUTER SCIENCE DEPT. *
*****************************************************************************rose@odin.ucsd.edu (Dan Rose) (10/06/89)
In article <2776@jhunix.HCF.JHU.EDU> suraj@jhunix (Suraj C. Surendrakumar) writes: > >IS AI HEADING THE RIGHT WAY TODAY ? >WHAT DO WE NEED TO DO TODAY FOR THE FUTURE OF AI TOMORROW ? >. . . >Over the years AI scientists who have >been ignoring cognitive science are realizing that cognitive science has >a lot to offer. . . . >. . . A lot of want we want know can be got be understood by unrevealing >the brain and getting input from cognitive science and dropping the notion >the cognitive science and AI have little in common. The notion that AI has ignored cognitive science is not true. In fact, many AI pioneers were founding members of the Cognitive Science Society. Its journal has published some of the most influential work in AI. Just to illustrate, here are a few of the well-known "AI scientists" who have published in the journal Cognitive Science: D. Bobrow A. Newell E. Charniak S. Papert F. Hayes-Roth R. Schank W. Lehnert H. Simon D. McDermott T. Winograd M. Minsky Y. Wilks Now, if you want to argue that AI & Cognitive Science have neglected studying the brain instead of the mind, that's an entirely separate issue. Dan Rose