It looks that "intelligence" is currently a too vaguely defined term.
Let me give an example: a system successfully learning to drive a car.
Do we consider this to be "intelligence", "true intelligence",
"artificial intelligence", "computational intelligence"
or a "system that successfully mimics human-like skills"?
Anyhow, for this type of tasks the relevant question is not
within how many years intelligence will be realized, but how
many years ago it has already been realized. Early successes in neural
networks were indeed e.g.
Pomerleau D., Efficient Training of Artificial Neural Networks for
Autonomous Navigation, Neural Computation, Vol. 3, No. 1, 1991, pp. 88-97.
and more recently
http://www.youtube.com/watch?v=2gQ3P5BpLjQ
All of this is reality today and there exist numerous examples
of intelligence at this level.
Therefore by "true artificial intelligence" one probably means
systems that are smarter than this, often referring to understanding
or trying to mimic the human brain.
For this ambitious goal it is expected that an interdisciplinary
understanding of the problem will be needed, see e.g.
- In mathematics
Steve Smale,
Mathematical Problems for the Next Century, Mathematical Intelligencer,
Vol 20, (1998) No. 2, 7-15.
in problem 18 limits of intelligence are explicitly posed as open
problems for the century
and connections between learning theory and neuroscience in
Mathematics of the Neural Response,
S. Smale, L. Rosasco, J. Bouvrie, A. Caponnetto, T. Poggio
Foundations of Computational Mathematics
February 2010, Volume 10, Issue 1, pp 67-91
- In physics e.g. comments by Sir Roger Penrose in
A theory of everything?
Nature 433, 257-259 (20 January 2005)
http://www.nature.com/nature/journal/v433/n7023/full/433257a.html
- the blue brain project http://bluebrain.epfl.ch/
- many others
An integrative interdisciplinary understanding of science is
of course extremely difficult or maybe even impossible to realize,
but will be needed in my opinion to finally arrive at the point
where one can understand (true) intelligence in all its facets,
artificial and human.
Possibly simple and unifying approaches could facilitate this
process. At the level of computational intelligence and machine learning
this has been a motivation in our work:
http://www.youtube.com/watch?v=0QOjatjtBWw
and recently new results at the interface between
quantum mechanics, machine learning, neural networks,
optimization, learning theory, mathematics and statistics
have been reported in
Johan A.K. Suykens,
Generating quantum-measurement probabilities from an optimality
principle, Phys. Rev. A 87, 052134 (2013)
http://pra.aps.org/abstract/PRA/v87/i5/e052134
http://homes.esat.kuleuven.be/~sistawww/cgi-bin/newsearch.pl?Name=Suykens+J
which may hopefully contribute to further bridging of different fields
related to intelligence.
Best regards,
Johan
----------------------
Prof. Dr.ir. Johan Suykens
Katholieke Universiteit Leuven
Departement Elektrotechniek - ESAT-STADIUS
Kasteelpark Arenberg 10
B-3001 Leuven (Heverlee)
Belgium
Tel: 32/16/32 18 02
Fax: 32/16/32 19 70
Email: johan.suyk...@esat.kuleuven.be
http://www.esat.kuleuven.be/stadius/members/suykens.html
On 08/23/2013 09:58 AM, Rainer von Ammon wrote:
Ray Kurzweil believes that, of course, like all transhumanists, e.g. Ben
Goertzel
Read his book
http://www.amazon.ca/How-Create-Mind-Thought-Revealed/dp/0670025291
and I commented a bit
http://forum.complexevents.com/viewtopic.php?f=13&t=257&p=1497#p1497
-Rainer
Am 21.08.2013 21:23, schrieb Richard E. Neapolitan:
This is the best response I obtained arguing that an artificial brain
is on the horizon:
-------- Original Message --------
Subject: RE: [UAI] Research indicating that we will have true
artificial intelligence in the next 10 to 100 years
Date: Wed, 21 Aug 2013 13:01:14 -0400
From: Fry, Robert L. <robert....@jhuapl.edu>
To: Richard E. Neapolitan <richard.neapoli...@northwestern.edu>
Firstly, we have collectively failed at this endeavor so far. Unlike
the past,
however, there are clearly ongoing research threads that imply that
the 10 to 100 year is reasonable. This is of course an opinion based
on my own research and views. These threads include:
o The DARPA Physical Intelligence program
<http://www.darpa.mil/Our_Work/DSO/Programs/Physical_Intelligence.aspx>,
its goals, and its progress to
establish the physical basis for intelligence, e.g., the Carnot
cycle whereby
an open system extracts information from its environment reducing its
entropy and allowing it to make decisions to control its future.
o The recent April 2013 /Phys Rev. Lett./ paper by Wissner-Gross
Causal Entropic Forces
<http://www.alexwg.org/publications/PhysRevLett_110-168702.pdf>
and their take on the thermodynamic basis of intelligence. See their
videos,
examples, and visit their start-up from MIT.
o The work of Karl Friston and his Free-energy model of the brain –
see his article
in Nature <http://www.nature.com/nrn/journal/v11/n2/abs/nrn2787.html>.
o My own research on a fundamental theory of computation
<http://www.nicta.com.au/__data/assets/pdf_file/0008/25991/Another_Information_Revolution_3.0.pdf>
which I am
transitioning into a formal engineering framework. I can
ostensibly completely
reverse-engineer the cortical neuron viewing it as an autonomous
intelligent
system. Might check out the short video I made for 5^th grades as
part of the
Flame Challenge that addresses the question “What is Time?”
<http://vimeo.com/60858372>
o Relationship of ALL the above to Maxwell’s Demon
<http://en.wikipedia.org/wiki/Maxwell%27s_demon>,Rolf Landauer’s
<http://en.wikipedia.org/wiki/Landauer%27s_principle> work
and the like.
I believe it will be around 10-15 years max. Anyway, these are my views
Bob Fry
Prof. Robert Fry
Johns Hopkins University Applied Physics Laboratory
*From:*uai [mailto:uai-boun...@engr.orst.edu] *On Behalf Of *Richard
E. Neapolitan
*Sent:* Tuesday, August 20, 2013 9:23 AM
*To:* uai@engr.orst.edu
*Subject:* [UAI] Research indicating that we will have true artificial
intelligence in the next 10 to 100 years
Dear Colleagues,
One of my publisher's asked me to review a proposal for a book. The
theme of the book is predicated on the statement that "it is widely
believed that in the next 10 to 100 years scientists will succeed in
creating human level artificial general intelligence." There is no
research that gives me any reason to believe this. In my recent AI
textbook I took the stance that we have essentially failed at this
endeavor. Does anyone know of any research that would make someone
make such a statement?
Thanks,
Rich
--
Richard E. Neapolitan, Ph.D., Professor
Division of Health and Biomedical Informatics
Department of Preventive Medicine
Northwestern University Feinberg School of Medicine
750 N. Lake Shore Drive, 11th floor
Chicago IL 60611
_______________________________________________
uai mailing list
uai@ENGR.ORST.EDU
https://secure.engr.oregonstate.edu/mailman/listinfo/uai
--
Dr. Rainer von Ammon
Centrum für Informations-Technologie Transfer GmbH
Agnesstraße 29
D-93049 Regensburg
Handelsregistereintragung: HR B 9928
Handelsregister Regensburg
Umsatzsteueridentifikationsnummer: DE245567586
Mobile: +49 (0)171 756 4557
Phone: +49 (0)941 8 97 04 72
Fax: +49 (0)941 8 97 04 73
http://www.citt-online.com
Postanschrift:
Eichelberg 8
D-93138 Lappersdorf
This e-mail message, including any attachments, is confidential and may contain
legally privileged information. This message is intended solely for the use of
the adressee. Use by others is prohibited. Disclosure, copying and distribution
of this information to third parties is not permitted. If you have received
this message in error, please notify us immediately and delete it from your
system. The integrity and security of this message cannot be guaranteed and it
may be subject to data corruption, interception and unauthorised amendment, for
which we accept no liability.
_______________________________________________
uai mailing list
uai@ENGR.ORST.EDU
https://secure.engr.oregonstate.edu/mailman/listinfo/uai
_______________________________________________
uai mailing list
uai@ENGR.ORST.EDU
https://secure.engr.oregonstate.edu/mailman/listinfo/uai
