Sander Olson Interviews
Francis Heylighen
CONDUCTED NOVEMBER 2001
Question 1: Tell us about yourself. What is your background, and what
projects are you currently working on?
My background is in mathematical physics, and I got my PhD in 1987 from the Free
University of Brussels (VUB). I am presently a Research Professor and a
co-director of the transdisciplinary research Center "Leo Apostel" at the VUB.
I have been working at the VUB since 1982 first on the foundations of physics
(quantum mechanics and relativity theory). The focus of my research then turned
to the evolution of complexity, which I study from a cybernetic viewpoint. I
have worked in particular on the evolution of knowledge (including memes), and
the creation of new concepts and models. More recently, I have extended the
underlying principles to understand the evolution of society, and its
implications for the future of humanity. The theoretical framework I am
developing intends to integrate knowledge from different disciplines into an
encompassing “world view.”
Together with my collaborator Johan Bollen, I have applied this framework by
implementing a self-organizing knowledge web, that "learns" new concepts and
associations from the way it is used, and "thinks" ahead of its users. As such,
it forms a simple model for a future intelligent computer network, the “global
brain.”
To study the technological and social implications of this vision, in 1996 I
co-founded the “Global Brain Group,” an international discussion forum that
groups most of the scientists who have worked on this issue. Since 1990, I am
also an editor of the Principia Cybernetica
Project, an international organization which attempts to consensually
develop a cybernetic philosophical system, with the help of computer
technologies for the communication and integration of knowledge.
At the moment, I am focusing on developing what I call “evolutionary
cybernetics,” an encompassing theory of how intelligent, purposeful organization
can originate and develop through the mechanism of blind variation and natural
selection. One of the applications of this theory is the emergence and
development of intelligence in the web. I am therefore further researching
algorithms that would allow the web to self-organize so as to become more
intelligent.
Question 2: Describe your concept of a Global Brain.
The "Global Brain" is a metaphor for the emerging collectively intelligent
network formed by the people of this planet together with the computers,
knowledge bases, and communication links that connect them together. This
network is an immensely complex, self-organizing system that not only processes
information, but increasingly can be seen to play the role of a brain: making
decisions, solving problems, learning new connections and discovering new ideas.
No individual, organization, or computer is in control of this system: its
knowledge and intelligence are distributed over all its components. They emerge
from the collective interactions between all the human and machine subsystems.
Such a system may be able to tackle current and emerging global problems that
have eluded more traditional approaches, but at the same time it will create new
technological and social challenges which are still difficult to imagine.
Without doubt, the most important technological, economic, and social
development of the past decade is the emergence of a global computer-based
communication network. This network has been growing at an explosive rate,
affecting -- directly or indirectly -- ever more aspects of the daily lives of
the people on this planet. Amidst this growing complexity, we need to look
ahead, and try to understand where all these changes are leading to.
A general trend is that the information network becomes ever more global, more
encompassing, more tightly linked to the individuals and groups that use it, and
more intelligent in the way it supports them. The web doesn't just passively
provide information, it now also actively alerts and guides people to the best
options for them personally. To support this, the web increasingly builds on the
knowledge and intelligence of all its users and information providers
collectively, thanks to technologies such as collaborative filtering, agents,
and online markets. It appears as though the net is turning into a collective
nervous system for humanity: a global brain.
Question 3: How does the concept of a Global Brain differ from
conventional theories of Intelligence Amplification? How related are the two
concepts?
I didn't know there were "conventional" theories of Intelligence
Amplification! I just know that several people have proposed that concept to
emphasize that computer technology should be used not so much to build
independently intelligent programs (Artificial Intelligence, AI), but to develop
support systems that would enhance our own human intelligence (Intelligence
Amplification, IA), but these people never became part of the mainstream. Two
pioneers that come to mind are Ross Ashby, one of the founders of cybernetics,
whose contribution was mainly theoretical, and Doug Englebart, the computing
pioneer who was the first to experiment with such basic interface elements as
the mouse, windows and hypertext.
I believe both of these pioneers would basically agree with the way I envisage
IA as supported by an intelligent web. The difference is rather one of emphasis:
while "conventional" IA might imagine the amplification of individual
intelligence by an individual computer system (e.g. a PC), I emphasize the
amplification of individual and collective intelligence by means of a shared
information network (the web). The power of the web is something that early
pioneers would have found hard to imagine, although Englebart in his later work
seems very much aware of it.
Question 4: How much longer do you believe that the Internet will continue
growing? Can one truly claim that beyond a certain point the Internet will
become sentient?
"Growth" is for me not the main issue. More and more people will use the net for
longer and longer times, using ever-faster processors and communication links.
Up to the point where every person and every appliance will be connected to the
net full-time, I don't see anything that will stop this growth.
More important than quantitative growth is qualitative development: will the net
be organized in a more intelligent way, so that it can e.g. autonomously learn,
reorganize, make decisions, solve problems... If this deep qualitative
reorganization takes off, then perhaps something like "sentience" will emerge,
but this will be a very difficult process, fraught with technical, scientific,
political, and social problems.
Question 5: Vernor Vinge argues that a group of PhDs with an Internet
connected workstation could ace any intelligence test ever devised. Ray Kurzweil
argues that as soon as computers reach parity with human intelligence they will
necessarily soar past it. Which opinion do you think is more accurate?
I'd rather side with Vinge here. Kurzweil's view neglects the important lessons
that have been learned from AI: to build real intelligence into a computer, you
don't just need a powerful processor, you need a huge mass of common-sense
knowledge and intuition, which you can only accumulate through a life-time of
experience interacting with a truly complex environment (this requirement is
sometimes called "situatedness" or "embodiment").
Such interaction requires very sophisticated sensors, effectors, and neural-type
circuits connecting the two. These are extremely difficult to build into any
artificial, robot-like creature, but are inexpensively available in any human
being. It is much easier to tap into that human experience and augment it with
computer memory and processing, than to build a computer intelligence from
scrap. Even if such a computer with human-level intelligence would be built,
there is no reason why its intelligence would grow faster than the intelligence
of a synergetic system consisting of intelligent humans and intelligent
computers intimately working together.
Question 6: What is your opinion of molecular nanotechnology? Do you
believe that molecular assemblers will ever be feasible?
I don't know enough about nanotechnology to have firm opinions about it. In
principle, I don't see any physical obstacles to building molecular assemblers,
but the issue that seems to be neglected is control: how do you make an army of
microscopic machines do precisely what you want? For simple machine-like
functions, such as cogs and wheels, that may not seem too difficult. But then
you don't gain such a great deal by building a microscopic lever. You'd rather
have nanosystems that can tackle complex problems, like building living cells
from scratch. But that will require either an unmanageably complex problem of
programming the "software" to execute these tasks, or give the system a large
measure of autonomy and self-organization. The latter seems most realistic to
me, but the danger is that you lose control, and your nanodevice will not do
exactly what you want.
Yet, I am not afraid of “grey goo” scenarios in which nanorobots run amok and
destroy everything in their wake. I think we can get the best inspiration for
what may happen from existing molecular devices, namely those developed by
biological systems, such as enzymes and DNA. Biological self-organization is
obviously quite efficient, but it has taken billions of years for evolution to
get there, and organisms are still rather unreliable as machine-like
“assemblers.” Now and then, something runs amok and a new killer virus appears
(e.g., AIDS), but until now, this has never happened on a scale even remotely
similar to the “grey goo scenario.” The best way forward to me seems that we
should better understand biological self-organization, and support or augment it
in a way similar to the way computers may augment human intelligence.
Question 7: What is your opinion of a technological singularity? If you
think it is likely, when do you think it will happen?
The more I think about the singularity, the less I believe it is a realistic
description of what will happen. It is true that most parameters of
technological progress have been showing a spectacular acceleration over the
past century, but this doesn't mean that the speed of progress will ever become
infinite, as the mathematical definition of a singularity would imply. I have
rather the feeling that we can already see the first signs of a deceleration.
The spectacular wave of innovation unleashed by the first user-friendly PCs in
the 1980's and of the Web in the 1990's seems to have gotten drowned in
complexity and confusion, as software developers are scrambling to keep their
systems up-to-date with all the new standards, plugins and extensions, while
merely adding esthetic improvements to the existing GUI-Web interface. While we
constantly hear announcements of the most spectacular innovations, in practice
most of these never reach maturity, because the developers underestimated the
complexity of the task environment.
I believe we are confronted with a complexity bottleneck, which will
significantly dampen the speed of further progress. The human mind simply is no
longer able to cope with the information overload. This also means that all the
big software projects that require a lot of coordination between different
people and sources of information (e.g., the present "Semantic Web" efforts)
either will get seriously delayed or end up with buggy products.
The only way to overcome this will be a shift to a radically different way of
tackling problems, where the main burden is no longer on individuals or teams,
but on the distributed, self-organizing, synergetic system that I call the
global brain. This shift will require a lot of time and effort, and won't just
happen instantaneously.
A better model of this transition is not the singularity (hyperbolic function
into infinity) but a logistic curve (exponential growth which slows down until
it is practically linear, and then slows down further, stabilizing at a new
plateau). We are now probably somewhere in the middle, linear part of the curve.
Seen from a distance (say with a million-year scale), a logistic curve may look
like a step function, which implies a singularity or discontinuous jump between
plateaus. In that sense, the singularity is not such a bad model, but in our
present, year by year, time scale, the singularity view doesn't make much sense.
If you would ask me when the singularity would take place in the million-year
view, then I would answer that we are right in the middle of it. But it may take
another 50 years or so to come to an end, unlike a real singularity, which is by
definition instantaneous.
Question 8: Speaking of the Singularity, how much longer do you believe
that Moore's Law will continue? Do you think that we will ever have molecular
electronics?
As you may have guessed by now, I'm not much preoccupied by Moore's Law. The
real bottleneck will be organizational: how will we cope with the complexity
involved in programming the powerful processors promised by Moore's Law to do
more than number-crunching? I believe Moore's Law, or advances in processing
speed more generally, will continue long enough to give us more than sufficient
computing power for the tasks we would like to achieve.
Question 9: Do you believe that the barriers to machine intelligence are
more hardware related or software related? Can we truly have either AI or IA
without a software breakthrough?
As I already indicated, the real challenge will be software rather than
hardware, and breakthroughs are necessary to achieve both AI and IA. I have no
doubts that these are possible, and a lot of good theoretical ideas are floating
around. The biggest problem is to integrate all of these into an elegant and
encompassing system that would have the power to self-organize and adapt to the
problems that are posed to it.
Question 10: What are your plans for the future?
As I said, my main focus now is the development of evolutionary cybernetics, a
theoretical framework that would hopefully give us a solid foundation for the
integration of all these promising ideas about self-organization, autonomy,
distributed knowledge systems, etc. I plan to give lectures on this subject,
write a textbook, and a number of papers. At the same time, I plan to test my
algorithms for a learning and "thinking" web in a more realistic environment, to
demonstrate their practical usefulness. I further want to continue developing
and spreading the global brain vision together with my colleagues in the Global
Brain Group, through lectures, conferences, publications and websites.
This interview was conducted by Sander
Olson. The opinions expressed do not necessarily represent those of CRN.
RETURN TO LIST OF
INTERVIEWS