Can Smart Machines Bail Us Out?
Markram estimates that in order to accurately simulate the trillion synapses in the human brain, you'd need to be able to process about 500 petabytes of data (peta being a million billion, or 10 to the fifteenth power). That's about 200 times more information than is stored on all of Google's servers. (Given current technology, a machine capable of such power would be the size of several football fields.) Energy consumption is another huge problem. The human brain requires about 25 watts of electricity to operate. Markram estimates that simulating the brain on a supercomputer with existing microchips would generate an annual electrical bill of about $3 billion . _Henry Markram
Singularity devotees such as Ray Kurzweil place great hopes upon the advent of a super human level thinking machine that can show us the way out of the labyrinth. According to that meme, the appearance of superhuman intelligence and the onset of the singularity will coincide.Smart machines do not have to work like the human brain, but since the human brain is the only "smart machine" we currently know of, the first ones probably will. Because we already have a "proof of concept" of the biological thinking machine, the smart money for smart machines is on the people who are trying to reverse-engineer the brain.
A recent Wired article looked at the IBM Almaden project to reverse-engineer the human brain. Several such projects are ongoing around the world, including the Blue Brain project in Lausanne, which was launched in 2005 as a joint venture with IBM. Henry Markram predicts that if progress in new computing machines continues at the current rate, he should be able to simulate a human brain in silico within 10 years. A shrewd observer of technological predictions might observe that a ten year prediction is cheap, allowing plenty of time for one to explain away the prediction.
Still, Markram is definitely one of the people to watch, if one wants to see the state of the art. The video below is two years old, from the 2006 IBM Almaden conference on cognitive computing. For anyone truly interested in the topic, the videos from that conference are worth watching.
Can smart machines bail us out? It depends on where they go to school, on what they learn and from whom. Smart machines, like smart brains, can develop along twisted, dysfunctional lines. Wisdom, intelligence, character, experience, perspective, savvy, competence, vision, are all important for developing truly smart brains and machines. Is it fair to expect from machines what we are unwilling to develop in ourselves? If we were serious about all this, would our educational systems, entertainment methods, and news/informational means take their present forms? Not likely.
Intelligent machines are possible -- we are proof of concept. Whether we are foresighted enough to evolve machines that are both intelligent and wise (from a human perspective) is another question.
Labels: artificial intelligence, Intelligence, silicon brain, Wisdom
posted by al fin at 7:46 am
2 Comments:
I suspect that one of the key ingredients to wisdom is experience which is why many of the people who exhibit it are older (though of course not all old people are wise and not all wise people are old). An artificial system which can survive indefinitely and can share experience directly instead of teaching it to others imperfectly might see wisdom develop far faster than humanity using books, parenting and other forms of transmission.
Low-intelligence, unwise systems with subhuman intelligence, however, would be incredibly revolutionary on their own. In fact, human level and greater intelligent systems would probably still need such "assistant" level systems as much if not more than we do. The most educated and highly trained people are the ones who are generally given assistants so as to maximize the "professional's" productivity. The next massive jump in productivity might happen when lower and middle management and younger researchers gain access to the kind of assistance that is currently only available to the established bigwigs due to the funding limitations of hiring humans for this role even though most of a human brain is not needed to look up journal articles and conduct simple experiments or collect and process data needed to evaluate alternative bureaucratic procedures and such.
Yes, Baron. And just imagine the sense of power should one have control of an entire herd of super-human intelligent machines?
At a time when Americans are faced with the prospect of their entire country being re-made in the image of South Side Chicago, the hope for a wise superhuman intelligent machine can only grow stronger.
Post a Comment
“During times of universal deceit, telling the truth becomes a revolutionary act” _George Orwell
<< Home