perm filename FREDKI[E83,JMC] blob
sn#724582 filedate 1983-08-28 generic text, type C, neo UTF8
COMMENT ā VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 Here is Ed Fredkin's (EF@MIT-MC) proposal for a meeting and white
C00020 ENDMK
Cā;
Here is Ed Fredkin's (EF@MIT-MC) proposal for a meeting and white
paper on AI. I suppose the issue for the Council is whether we
consider a push for high level attention to AI at this time
appropriate and, if so, whether this is the right way to go about
it. Though Fredkin is not currently active in AI, he
is energetic and knows lots of people. I suppose
the Council should be represented more specifically on the Steering
Committee if we decide to proceed.
BACKGROUND AND HISTORY
We believe that the United State may be at a turning point,
facing either a great opportunity or a potential risk, depending
on the course of action taken now in the field of Artificial
Intelligence. The reason is not because of any startling new
research developments or breakthroughs, but rather because, for
the first time, there is a dramatically increased awareness of
the possibility of practical applications evolving from AI
research, especially in Japan. The Fifth Generation book is
actually the number one best seller (non-fiction) in Japan this
week, the last week in August. This comes at a time when the
rapid progress of computer technology finally allows AI programs
to run on relatively inexpensive systems. The confluence of
events may make it possible to do things now, that may not be
possible later. The reason that we are at a turning point is
because we face a window of opportunity that we must not miss.
For the past twenty years, the US government, through DARPA,
has supported research in the field of Artificial Intelligence.
While the work, which started at MIT and Stanford, was an obscure
area of basic research in the late fifties, it is now realized
that the practical applications of that research may have a very
large and increasingly important effect on the future strength of
this country, vis a vis its economic competitors. Japan, is
today the country with the most ambitious national program in the
field of AI. England has also embarked on a national effort.
While the US has been the leader in this field, we may not be
able to maintain that lead without embarking upon efforts of
greatly increased scope. The payoff, of maintaining our
preeminent position, appears to be so large as to justify very
great investments in this area. On the other hand, should we
fall behind in this technology, the risks we might face are not
merely economic ones, since, someday, advanced AI technology may
be an important factor in other aspects of national power. Most
importantly, the opportunity to precipitate large scale
governmental support appears to be present today, it was never
present in the past and, if this opportunity passes by, it may
not be present again for some time.
THE MEETING AND WHITE PAPER
The goals envisioned for the meeting would be to consider
certain aspects of AI research, development and applications
possibilities, consider what other nations are doing in that
context, and finally consider possible recommendations for
national action in the US. While it would be healthy for the
conferees to consider broadly the realm of future possibilities,
quite conservative extrapolations of what we already know should
be sufficient to allow one to conclude that the scale of prudent
activity for the US is greater than presently being considered.
If one looks at the range of practical applications, using
industry standard criteria such as projected return on
investment, a possible conclusion will be that the most
reasonable level of investment in all aspects of AI, from basic
research to practical applications of proven technology, may be
far greater than what is now being done. If the right group of
conferees did come to such a conclusion, and if it was able to
support its conclusions with reasoned arguments, then it is
likely that its recommendations might result in appropriate
governmental action.
WHO SHOULD BE INVITED
Attendance at this meeting should be limited to a small
number, perhaps 20 participants. The participants should consist
of two groups, those very familiar with AI, who are good at
conveying the concepts, possibilities and possible consequences
of various courses of action to a group of intelligent laymen
(laymen in the field of AI). And then the others, who should be
persons of great prestige who would be drawn from government,
science and industry.
ATTENDEES DRAWN FROM THE AI COMMUNITY
The members of the AI community, who attend the meeting,
will have the responsibility of educating the others, and
communicating the possible scenarios consequent to various
proposed actions. There is a great tendency for the experts in
an obscure field to talk over the heads of the most intelligent
of laymen, resulting in very little communication. In this case,
it will be very important to work hard at conveying the
information from expert to layman in ways that result in
understanding and appreciation for the most important issues.
There is, in this plan, the implication that it would be
sensible for the members of the AI community who attend the
meeting, to be of one mind on the important issues, and if that
is not possible, then there may be no point in holding the
meeting. The Japanese have made a great art of coming to a
consensus, where the natural tendency to display one's ego by
vociferous disagreement is suppressed. Some of the superstars in
the field AI are infamous for the opposite, never participating
in a meeting without showing strong disagreement with some point,
as a matter of principal. While the AI community can know and
love its heros for their weaknesses as well as for their
strengths, if we want to have an effect on the world, we will
have to be pragmatic, and hold back a little of our
individualism. The AI persons need not agree on all points, but
they must understand in advance what they disagree about, they
must be able to delimit and make clear the consequent
alternatives, and they must not allow such differences to get in
the way of communicating the broad areas of agreement. As a
consequence, the AI participants must get together before the
meeting to explore the issues and to come to a consensus.
THE NON-AI PARTICIPANTS
The main objective of the meeting will be to communicate the
rationale for greatly increasing government funding of various
aspects of AI research and technology. The audience should
consists of a group of prestigious movers and shaker who are
able to appreciate the possibilities. It will be important to
have representatives from three major groups: Government, because
they will best understand the process of turning such concepts
into actions; Industry, because much of the economic consequences
will be a result of what American Industry chooses to do, and
because of the practical experience of such persons; Science,
because the government is used to, for good reason, consulting
with certain trusted scientific advisors. Again, since the
objective is to give the best opportunity to the realization of
the goals of the meeting, the non-AI participants need to be very
carefully selected.
PERTINENT MATERIALS AND THE WHITE PAPER
A very important aspect of the conference would be the
distribution to the conferees, prior to the conference, of
materials that detail what is happening in the field of AI.
Examples might include the Fifth Generation by Feigenbaum and
McCorduck and papers such as "Artificial Intelligence: An
Assessment of the State-of-the-Art and Recommendation for Future
Directions" by David Waltz et al, in the Fall 83 issue of AI
Magazine. In addition, the organizers of the meeting should
write and distribute condensed background information that can be
easily and quickly digested. Subsequent to the conference, a
white paper should be prepared and distributed. That white paper
will be the focal point of consensus and a document that makes
specific recomendations for action.
THE SPECIFIC RECOMMENDATION TO THE AAAI COUNCIL
The council should agree that the AAAI should sponsor a
meeting, along the lines outlined above. While the AAAI would
sponsor the meeting, the opinions and conclusions expressed by
the participants would be their own, and not official positions
of the AAAI. A steering committee consisting of Raj Reddy,
Gordon Bell, Ed Fredkin, Bob Kahn (if he is willing) and perhaps
one or two others should be constituted. Ed Fredkin would be the
Chairman of the committee and meeting. The meeting will be at no
cost to the AAAI, it will pay its fair share of all expenses
incurred by the AAAI and return a modest profit. Claudia Mazzetti
has indicated that she would be able and pleased to handle
various arrangements for the meeting, for which the AAAI would be
reimbursed. The most likely site would be in the Washington
area, and the meeting would take place sometime before the end of
the year.