Posts Tagged ‘Kurzweil’

An AI Armageddon

July 27, 2017

This post is inspired by an article by Cleve R. Wootson, Jr. in the July 24, 2017 Washington Post article titled, “What is technology leader Musk’s great fear? An AI Armageddon”.

Before addressing an AI Armageddon Musk speaks of his company Neuralink, which would devise ways to connect the human brain to computers. He said that an internet-connected brain plug would allow someone to learn something as fast at it takes to download a book. Everytime HM downloads a book to his iPad he wonders, if only… However, HM knows some psychology and neuroscience, topics in which Musk and Kurzweil have little understanding. Kurzweil is taking steps to prolong his life until his brain can be uploaded to silicon. What these brilliant men do not understand is that silicon and protoplasm require different memory systems. They are fundamentally incompatible. Now there is promising research where recordings are made from the rat’s hippocampi while they are learning to perform specific tasks. Then they will try to play these recordings into the hippocampi of different rats and see how well they can perform the tasks performed by the previous rats. This type of research, which stays in the biological domain, can provide the basis for developing brain aids for people suffering from dementia, or who have had brain injuries. The key here is that they are staying in the biological domain.

This biological silicon interface needs to be addressed. And it would be determined that this transfer of information would not be instantaneous, it would be quite time consuming. And even if this is solved, both the brain and the human are quite complicated and there needs to be time for consolidation and other processes. Even then there is the brain mind distinction. Readers of this blog should know that the mind is not contained within the brain, but rather the brain is contained within the mind.

Now that that’s taken care off, let’s move on to Armageddon. Many wise men have warned us of this danger. Previous healthy memory posts, More on Revising Beliefs, being one of them reviewed the movie “Collosus: the Forbin Project.” The movie takes place during the height of the cold war when there was a realistic fear that a nuclear war would begin that would destroy all life on earth. Consequently, the United States created the Forbin Project to create Colossus. The purpose of Colossus was to prevent a nuclear war before it began or to conduct a war once it had begun. Shortly after they turn on Colossus, the find it acting strangely. They discover that it is interacting with the Soviet version of Colossus. The Soviets had found a similar need to develop such a system. The two systems communicate with each other and come to the conclusion that these humans are not capable of safely conducting their own affairs. In the movie the Soviets capitulate to the computers and the Americans try to resist but ultimately fail.

So here is an example of beneficent AI; one that prevents humanity from destroying itself. But this is a singular case of beneficent AI. The tendency is to fear AI and predict either the demise of humanity or a horrendous existence. But consider that perhaps this fear is based on our projecting our nature on to silicon. Consider that our nature may be a function of biology, and absent biology, these fears don’t exist.

One benefit of technology is that the risks of nuclear warfare seem to have been reduced. Modern warfare is conducted by technology. So the Russians do not threaten us with weapons; rather they had technology and tried to influence the election by hacking into our systems. This much is known by the intelligence community. The Russians conducted warfare on the United States and tried to have their candidate, Donald Trump, elected. Whether they succeeded in electing Donald Trump cannot be known in spite of claims that he still would have been elected. But regardless of whether their hacking campaign produced the result, they definitely have the candidate they wanted.

Remember the pictures of Trump in the Oval Office with his Russian buddies (Only Russians were allowed in the Oval Office). He’s grinning from ear to ear boasting about how he fired his FBI Director and providing them with classified intelligence that compromised an ally. Then he tries to establish a secure means of communication with the Russians using their own systems. He complains about the Russian investigation, especially those that involve his personal finances. Why is he fearful? If he is innocent, he will be cleared, and the best thing would be to facilitate the investigation rather than try to obstruct and invalidate it. Time will tell.

How could a country like the United States elect an uncouth, mercurial character who is a brazen liar and who could not pass an elementary exam on civics? Perhaps we are ready for an intervention of benign AI.

© Douglas Griffith and healthymemory.wordpress.com, 2017. Unauthorized use and/or duplication of this material without express and written permission from this blog’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Douglas Griffith and healthymemory.wordpress.com with appropriate and specific direction to the original content.

 

 

Advertisements

More on Revising Beliefs

August 10, 2015

This is the third post in a series of posts on Nilsson’s book, Understanding Beliefs.  Nils J. Nilsson, a true genius who is one of the founders of artificial intelligence, recommends the scientific method, as the scientific method is the primary reason underlying the progress humans have made in the past several centuries.  We know from previous healthy memory blog posts that beliefs are difficult to change.  Yet we inhabit an environment in which there is ongoing dynamic change.  Moreover, modern technology accelerates the amount of information that is being processed and the amount of change that occurs.

The immediately preceding healthy memory post, “Revising Beliefs,” expressed extreme skepticism that there was sufficient sophistication among the public to implement the scientific method on a large scale in the political arena. Suppose this is indeed the case.  Suppose the world will be characterized by increasing polarization so that little or no progress can be made.  What is a possible remedy?

Here I wish that Nils J. Nilsson would write a second book on how technology, in the lingo of the healthy memory blog transitive memory, might be used to address this problem.  During the Cold War there was a movie, Collosus:  The Forbin Project.  At this time there was a realistic fear that a nuclear exchange could occur between the United States and the Soviet Union that would obliterate life on earth.  In the movie the United States has built a complex sophisticated computer, the Collosus to manage the country’s defenses in the event of a nuclear war.  Shortly after Collosus becomes operational it establishes contact with a similar computer built by the Soviet Union.  These two systems agree that humans are not intelligent enough to manage their own affairs, so they eventually take control of the world.

Perhaps we are not intelligent enough to govern and we need to turn the job over to computers.  Kurzweil has us becoming one with silicon in his Singularity, so we would be as intelligent as computers.  Suppose, however, that computers were infected with human frailties.  In Bill Joy’s the World Without Us, we are eliminated by intelligent machines.  But perhaps he is projecting human desires on computers.  Perhaps they would be motivated to dominate, but rather to assist.  Or perhaps this feature would be incorporated by AI developers offering this solution to a country or the world, locked in gridlock.

So here is my plea to AI researchers and Sci-fi authors.  Please take this concept and run with it.

© Douglas Griffith and healthymemory.wordpress.com, 2015. Unauthorized use and/or duplication of this material without express and written permission from this blog’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Douglas Griffith and healthymemory.wordpress.com with appropriate and specific direction to the original content.

Limits to Human Understanding

May 20, 2014

This blog post was motivated by an article in the New Scientist1, “Higher State of Mind” by Douglas Heaven.  It raised the question of limits to human understanding, a topic of longstanding interest to myself.  The article reviews two paths Artificial Intelligence has taken.  One approach involved rule-based programming.  Typically the objective here was to model human information processing with the goal of having the computer “think” like a human.  This approach proved quite valuable in the development of cognitive science, as it identified problems that needed to be addressed in the development of theories and models of human information processing.   Unfortunately, it was not very successful in solving complex computational problems.
The second approach eschewed the modeling of the human and focused on developing computational solutions to difficult problems.  Machines were programed to learn and to compute statistical correlations  and inferences by studying patterns in vast amounts of data.  Neural nets were developed that successfully solved a large variety of complex computational problems.  However, although the developers of these neural nets could describe the neural net they themselves had programmed, they could not understand  how the conclusion was made.  Although they can solve a problem, they are unable to truly understand the problem.  So, there are areas of expertise where machines can be said to know not only more than we do, but also know more than we are capable of understanding.  In other words, what we can understand  may be constrained by our biological limitations.
So, what does the future hold for us?  There is an optimistic scenario and a pessimistic scenario.  According to Kurzweil a singularity will be achieved by transcending biology and we shall augment ourselves with genetic alterations,  nanotechnolgy, and machine learning.  He sees a time when we shall become immortal.  In fact, he thinks that this singularity is close enough that he is doing everything to extend his life so that he shall achieve this immortality.  This notion of a singularity was first introduced in the fifties by the mathematician John von Neuman.
A pessimistic scenario has been sketched  out by Bill Joy.  I find his name  a bit ironic.  He has written a piece titled, “Why the Future Doesn’t Need Us”  where he argues that technology might be making us an endangered species.
So these are two extremes.  A somewhat less extreme scenario was outlined in the movie, Collosus:  the Forbin Project, which was based on a novel by Dennis Feltham Jones, Collus.  The story takes place during the cold war with the confrontation between the United States and the Soviet Union.  The United States has built a complex sophisticated computer, the Collosus to manage the country’s defenses in the event of a nuclear war.  Shortly after the Collosus becomes operational, it established contact with a similar computer built by the Soviet Union.  These two systems agree that humans are not intelligent enough to manage their own affairs, so they eventually hey take over the control of the world.
So what does the future hold for us?  Who knows?

© Douglas Griffith and healthymemory.wordpress.com, 2014. Unauthorized use and/or duplication of this material without express and written permission from this blog’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Douglas Griffith and healthymemory.wordpress.com with appropriate and specific direction to the original content.

How Much Information Is There and What Does It Mean?

September 27, 2012

A recent article by Martin Hilbert was published in the Big Data Special Issue of the publication Significance: statistics making sense titled “How Much Information Is There in the Information Society”? Hilbert together with his collaborator Priscila Lopez tackled the task of estimating the world’s technological capacity to store, communicate, and compute information over the period from 1986 to 2007/2012. The complete collection of these studies can be accessed free of charge at

http://martinhilbert.net/WorldInfoCapacity.html

In 1949 the father of information theory, Claude E. Shannon, estimated that the largest information stockpile he could think of was the Library of Congress with about 12,500 megabytes (106). The current estimate for the amount of storage for the Library of Congress has grown to a terabyte 1012. During the two decades of their study the amount of information quadrupled from 432 exabytes (1018) to 1.9 zetabytes (1021). For our personal and business computation we are familiar with gigabytes (109). Next are terabytes (1012), then petabytes (1015), the aforementioned exabytes, and zetabytes. Yottabytes (1024) await us in the future.

Although these are measures of information in the technical sense, I prefer to think of them as data. I think of information in technical transactive memory as data. When it is perceived by a human it becomes information. When it is further processed into the human information processing system, it becomes knowledge. Suppose we all disappeared and the machines kept remembering and processing. What would that be? Perhaps sometime in the future machines will become intelligent enough to function on their own. There is a movie, Colossus: the Forbin Project in which intelligent machines take over the world because they have concluded that humans are not intelligent enough to govern. Then there is Ray Kurzwiel‘s concept of the Singularity, when humans and technology become one. However, coming back to reality, I think there would just be machines storing and processing information absent true knowledge. We need to use technology to help us cope with all these data and fortunately according to Hilbert computation is grown at a faster rate than storage.

Hilbert makes some interesting comparisons between technical processing and storage of information and biological processing and storage of information. In 2007, the DNA of the 60 trillion cells of one single human body would have stored more information than all of our technological devices together. He notes that in both cases information is highly redundant. One hundred human brains can roughly execute as many nerve pulses as our general purpose computers can execute instructions per second. Hilbert asks the question why we currently spend 3.5 trillion dollars per year on our information and communication technology but less than $50 dollars per year on the education of many children in Africa? I think what he is proposing is that we not lose sight of human potential. Although our brains and DNA have phenomenal processing and storage capacities, we only have access to a very small percentage of this information in our conscious awareness. The healthymemory blog makes a distinction among potential transactive memory, available transactive memory, and accessible transactive memory. Potential transactive memory is all the information about which Hilbert writes as well as information held by our fellow humans. Available transactive memory is that information we are able to find. And accessible transactive memory is that information we are able to access readily. The goal is that this accessible transactive memory grows into knowledge, understanding, and insight, as it is in these final stages where its true value is realized.

© Douglas Griffith and healthymemory.wordpress.com, 2012. Unauthorized use and/or duplication of this material without express and written permission from this blog’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Douglas Griffith and healthymemory.wordpress.com with appropriate and specific direction to the original content.