Limits of Moore's law

These thoughts emerged after speaking with Clément Vidal, from an ongoing discussion at the FRIAM mailing list, and are also related to the so called Singularity. Many people say that in a few decades computers will have the same computational power as the human brain, and then AI machines will take over. This argument is based on Moore's law, which states that the transistor density of semiconductor chips doubles roughly every 18 months, giving an exponential growth in computing power. Ignoring the fact that brains do different types of computations than PCs, these people argue that when we'll have the computational power to model all the neurons of a human brain (100 billion, with about 100 thousand connections each), computers will be smarter than humans. And since their capacity will keep on growing, soon we'll be left in the shadow of obsolence.
Now, just imagine that you have all that computational power right now. Even more, let's say with the capacity of 10^100 CPU's running in parallel, or even more. How the hell do you program a human mind in there??? It takes us several years just to learn to talk! This is related to the so called software crisis: even when computing power increases, the ability to produce more complex software is not increasing as fast as Moore's law, so in practice we can make faster computations, but not more complex. I don't know if there is a measurement of software complexity, to see how it has increased over the years, but my guess would be that it would be linear, or even less... just look how long does it take to release a new operating system. Software is progressing, but not as fast as hardware, thus limiting the practicality of Moore's law.

Related ideas can be found on my essay: Why Computers Will Not Take Over the World, where I argue that rather than machines taking over humans, we will converge in our evolution, since we have different niches, which are actually complementary.

Comments

Carlos said…
Hi Clément,

I saw that you also mention the issue of programming the "brain"...

Well, that is supposing that the human brain is Turing-computable... see
"On the possible Computational Power of the Human Mind", Hector Zenil, Francisco Hernandez-Quiroz
http://arxiv.org/abs/cs.NE/0605065
(it's a chapter in the Philosophy and Complexity book)

About Humphrys, I met him at SAB'02 in Edinburgh, he also presented the World Wide Mind stuff. In principle, like many other clever ideas, it's interesting. But in practice, you need people's participation. This is a big danger in collaborative systems and portals, where there's the infrastructure to create a great knowledge base (as with Nagarjuna's www.gnowsys.org ), but then you need people to fill the knowledge in, which in Nagarjuna's case are his students, in PCP were Francis, Cliff, and Valentin... but for a larger system it is more complicated... however Wikipedia has been a success in this.
3:16 PM
Post a Comment

Popular posts from this blog

The Past, Present and Future of Cybernetics and Systems Research

Summary of "The Past, Present and Future of Cybernetics and Systems Research" Symposium M at  European Meetings on Cybernetics and Systems Research (EMCSR), Vienna, Austria, April 12, 2012. Organizer: Carlos Gershenson This guided reflection on the challenges and opportunities of cybernetics and systems research (CSR) included initial interventions by panelists Peter Erdi, Helena Knyazeva, Stefan Thurner, Peter Csermely, and Alexander Lazlo. Afterwards, the floor was opened to interventions from the general public and further interventions by panelists. Commentaries were made from a broad variety of perspectives, but several general ideas can be distilled from the discussion. CSR have strongly influenced all scientific disciplines. As an example, the term "system" is used commonly in daily language. One of the breakthroughs of CSR lies in the search of commonalities across disciplines. Even when this was achieved to a certain degree, there is still a lack of
Read more

flocking memes...

Some ideas just came up, after a talk I had with Klaas , who is making a PhD on memetics... In cultural transmission of fads and so on, I've heard that there are models in which the majority imitates an elite, and the elite tries to go away from the crowd, and that gives you the dynamics... The idea is that maybe another way of looking at this phenomenon is to take the analogy of flocking: birds try to keep together (safety, direction, whatever reason), but also keep average distance to their neighbours (not to crash). If we take this mechanism into the memetic space, people will imitate many aspects surrounding them, just because of mirron neurons and the evolutionary advantage of imitation. But the separation is also important, for people assert their individuality with it... This could be seen as taking the inverse of a meme... starting to sounds like dialectics... (who would have thought that Hegel and starlings had lots in common...) Of course in the social context there are m
Read more

PhD thesis: "final" version online

After making several minor corrections, I have a "final" version of my thesis, which you can download here . I should defend it privately on April 20th, and if everything goes well, the public defense would be on May 2nd. Everybody's welcome! Design and Control of Self-organizing Systems PhD Dissertation presented by Carlos Gershenson Promoters: Prof. Dr. Bart D’Hooghe and Prof. Dr. Francis Heylighen Advisors: Prof. Dr. Bruce Edmonds Prof. Dr. Bernard Manderick Prof. Dr. Peter McBurney Prof. Dr. Ann Nowé Abstract Complex systems are usually difficult to design and control. There are several particular methods for coping with complexity, but there is no general approach to build complex systems. In this thesis I propose a methodology to aid engineers in the design and control of complex systems. This is based on the description of systems as self-organizing . Starting from the agent metaphor, the methodology proposes a conceptual framework and a se
Read more