Will The Computers Really Take Over?
The idea that humans could build machines that could mimic or even extend human thought is not new. In the first scientific publication describing a working computer in 1843, the mathematician, Lady Ada, The Countess of Lovelace described Charles Babbage’s Analytical Engine
“Supposing, for instance, that the fundamental relations of pitched sounds in the science of harmony and of musical composition were susceptible of such expression and adaptations, the engine might compose elaborate and scientific pieces of music of any degree of complexity or extent.”
The Lady, Ada Lovelace also worked unsuccessfully on a mathematical model of the brain that she hoped would give rise to thoughts and nerves to feelings (“a calculus of the nervous system”).
It also wasn’t long before the dark side of computer minds was envisioned. In 1872 Samuel Butler published Erewhon where computer-like machines replace humans.
In recent years, such ideas have been referred to as the Singularity. The term coined by mathematician Vernor Vinge in 1993 describes an event where human-created technological entities posses superior human intelligence.
There is no doubt that computer abilities are already impressive. Computers can already do certain things better than humans. For example, they are less prone to mistakes and they do repetitive calculations more quickly and in greater degrees of complexity than humans are capable of. They also can store vast amounts of information in their memories that aren’t subject to human recall biases. Furthermore, computers don’t need to rest.
However, humans still display many capabilities that computers don’t and in most cases computers must team up with humans to achieve impressive results either through programming or by combining the unique capabilities of human and computer intelligence. For example, in medicine Medical Decision Making combines the art of medicine with complex probability calculations that can improve diagnostic reliability and accuracy.
Chess is an even more interesting case. In timed events, the speed and accuracy of computers can be formidable even for top level grandmasters. However, in postal chess where humans can take weeks to make a single move, humans still come out on top. Among top level players, it’s well know that certain types of chess positions still give computers trouble. Therefore, the combination of a strong player with a strong computer is vastly superior to any computer or human. The combination is referred to as advanced chess.
However, descriptions of the Singularity typically have several major flaws. Generally, the Singularity is hypothesized to occur as a result of increases in the speed and power of computers that surpass human intelligence. However, the way a computer “thinks” is vastly different than the way humans think. Therefore, greater speeds of processing simply don’t translate into intelligence. When futurists imagine computers with super-intellects, they’re speculating on a sort of computer that possesses the abilities of both the human mind and ordinary computing abilities.
A single neuron has the processing power of an average labtop computer. A neuron responds to input through a multitude of channels including biological, chemical and electrical stimuli. Neuronal responses are very computer-like in that neurons respond by varying the strength and quality of biological, chemical and electrical outputs that adjusts according to historical data. Neurons are superior to ordinary computers in that they can communicate, reprogram and cooperate with other neurons in ways that are far more complex than existing ways that computers can interact.
A recent study found that the average adult male brain has about 86 billion neurons but only about 16 billion of those are cortical neurons used for ‘thinking’. Other types of neurons are used for movement and interpreting incoming data from our senses. Thus, the human cortex is similar to a 16 billion-core computer where the characteristics of each core vary according to complex rules and the various cores can share information and respond to input from other cores in ways that existing computer cores can’t. For this reason, some who examine traditional modern computers conclude that humanity is a long way from creating a computer mind. However, three massive research projects have demonstrated that’s not the case and other spin-off projects suggest possibilities that even futurists have rarely considered (e.g., computerized honey bees).
The Blue Brain Project
An entirely different approach to creating a computer mind has begun to yield impressive results. For the past decade, Dr. Henry Markram of the Ecole Polytechnique Fédérale de Lausanne (EPFL), has attempted to reverse engineer the human mind by examining the neuronal connections at the molecular level and then inputting that information into some of the largest existing massively parallel processing computers. A massively parallel computer processes information more like a group of neurons do. The advantage of Dr. Markam’s approach is that it doesn’t require any new technology. Physicists already know how to scan information at the molecular level and they also know out to insert that information into computers. Therefore, the problem of creating a computerized human mind becomes simply a matter of time and money, and waiting for larger massively parallel computers to be developed.
The Blue Brain Project began in 2005. It’s primary funding is through IBM using their Blue Gene computers which are an IBM project aimed at designing supercomputers that can reach operating speeds beyond a quadrillion floating-point operations per second. In 2007, the Blue Brain Project demonstrated that it was able to simulate a cortical column of a rat’s brain. Since then, the project has developed the technology necessary to build brain models for any species at any stage in its development.
A rat’s cortical column contains approximately 10,000 neurons. The project moved onto its next goal, which is to simulate a cortical column in a human brain which may have as many as 100,000 neurons. It’s estimated that the human neocortex has as many as 1-2 million cortical columns.
The video above discusses one of the major findings of the Blue Brain Project that improves our understanding of why people who experience Traumatic Brain Injury(TBI) can recover much of what they’ve lost. The project found that most neuronal connections develop in the same way. That is, our experiences don’t influence neuronal connections. Neurons will simply attempt to connect with anything that is close by. Therefore, human brains can be rewired very easily.
By July 2011, a cellular mesocircuit of 100 neocortical columns with a million cells in total was built. A complete cellular rat brain is planned for 2014 with 100 mesocircuits totalling a hundred million cells. A cellular human brain is predicted to be possible by 2020. In the interim, cat and primate brains are being developed.
Once software is found that perfectly mimics the actions of real neurons at the molecular level of gene expression, the software can be rewritten to make it shorter and more efficient. This process greatly reduces the overall computing power needed to simulate a complete human brain.
A 16 minute documentary about the Blue Brain Project was released last year. It explains the enormous medical potential from being able to simulate the human brain (see below).
A new follow-up documentary was released this year that illustrates the enormous progress that has been made in just one year (see video below).
The Beautiful Brain is a website that created the Blue Brain documentaries and includes many other films about the junction of art and neuroscience.
The Blue Brain Project emphasizes the possibility of learning how various pharmaceuticals influence the mind. Thus far, a question that has been raised is how realistic is the simulation when the brain isn’t receiving the sorts of external input that ordinary brains do. The brains being modeled thus far are fetal brains in development. However, ordinary brains are flooded with environmental and physical stimuli so the researchers will also have to find ways to model environmental and physical stimuli.
Project Human Brain
In 2010, there was a general recognition from scientists working on the Blue Brain Project that more scientists needed to be involved in developing a complete working model of the human brain. Therefore, Project Human Brain was founded which attempts to integrate everything known about the brain into computer models that simulate the actual working of the brain. According to the Project Human Brain website
“The Human Brain Project builds on the work of the Blue Brain Project. The project is led by Dr. Henry Markram whose Blue Brain Project has provided a means to simulate the complete human brain. One major goal of this project is to develop computers that are 1,000 times more powerful than existing computers. Such computers should be able to simulate the cortex of the human brain.”
The project has applied for $1.4 billion dollars in funding. The results of the European funding contest will be known next year. An overview of the project is presented in the video below.
Recreating a human mind doesn’t create super-human intelligence. It just creates ordinary human intelligence. Initial uses of Project Human Brain will simulate how brains react to various deficits and chemicals.
Still, humans will have considerable advantages over any computer that is developed by the Human Brain Project. Humans are connected to some equally fantastic sensory devices capable of receiving vast amounts of input and then acting upon that output. For example, human stereoscopic vision is well beyond the best cameras in the world.
However, other aspects of sensory input may be less difficult to model. For example, technology to move robots through brain scanning has already been developed.
Project Green Brain
The Universities of Sheffield and Sussex are beginning to address the issue of computer brain robotic control by developing a robotic honey bee with the full mind of a bee. The bee is expected to be able to do everything a real bee can do, including finding flowers based on particular odors and is expected to be able to be used for search and rescue missions as well as mechanically pollinating crops. The project is called Green Brain. This project isn’t funded by IBM but one of their competitors NVIDIA using GPU accelerators that currently are the fastest computers in the world. Although not mentioned in the press release, robotic bees that can conduct search and rescue missions could presumably could also play a role in search and destroy missions. That might give the term “killer bees” a whole new meaning.
Dr James Marshall, a Reader in the Department of Computer Science published a mathematical model of bee decision making last year in the prestigious journal Science. He said
“I think that the bees have taught us a valuable lesson about how robust collective decisions can be made even when all the individuals have quite limited information. In the future, I think the rules we found the honeybees follow could prove useful in a variety of engineering applications; getting a swarm of robots to make a collective decision using the honeybee rules is one very exciting possibility.”
The robotic bee won’t just be programmed to complete tasks. Rather, it will act in accordance with its environment. The results of the research may also help in understanding and preventing Colony Collapse Disorder.
In 2008, IBM began working with the U.S. military to develop computers that can specifically simulate neurons. This project is referred to as Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE). As noted above, a human male is estimated to have 86 billion neurons. The goal of SyNAPSE is to build a cognitive computing architecture with 10^10^ neurons. This project is attempting to simulate a modular, scalable, non-von Neumann, ultra-low power, cognitive computing architecture at the scale of the number of human synapses in the human brain. In the past, SyNAPSE has used computing power to simulate cat and monkey brains. However, they’re not designing brains at the molecular level like Blue Brain. Instead, they’re working off of extremely sophisticated brain maps based on tracing studies. SyNAPSE attempts to create a generic brain that will largely ignore the variations between brains within species.
This month, SYNAPSE computers simulated the combined efforts of 530 billion neurons at a speed that is only 1,542 times slower than the human brain. Today, SyNAPSE simulated 2,085 billion neurons at the same speed and demonstrated for human size brains they are now only 388 times slower. The U.S. Defense Advanced Research Projects Agency (DARPA) has awarded approximately $21 million dollars in new funding for the SyNAPSE project. The project leader Dharmendra Modha, Principal Investigator, IBM Research – Almaden, explains his project in the video below and he also provides an explanation of the incredible power and efficiency of the human brain. He also demonstrates how his computers achieve the human-like quality of pattern recognition.
A number critics have argued that computers can’t take over the human race for the following reasons:
1. Computers are incapable of ambiguous thought or similar arguments that human thinking is qualitatively different.
This is a good argument for why ordinary computers can’t become intelligent. However, there are no known reasons why a perfectly simulated human mind wouldn’t be able to do everything a normal human could do.
2. Society will naturally fall-back to avoid such disasters because it will be overwhelmed by the complexity of computers that might threaten our existence.
3. Philosophers and religious leaders haven’t weighed in on this issue. However, they could provide significant push back.
This idea is incorrect as well. The super-intelligent computer will not be developed by society but by a small group of people who are well funded as are the projects listed above. Funding will always be adequate due to the potential of such computers. Therefore, critics who argue that the creation of a human-like computer mind is not possible, have nearly run out of arguments.
So are people creating computers than can take over the world? Not yet? The task of creating an ordinary human-like computer mind is sufficiently daunting at this point. In ten years time with the combined knowledge gained from the Human Brain Project and SyNAPSE, humanity may be on the brink of creating a fully functional human-like computer mind but that mind will not be superior to a human one. No research, to date, appears to be directed toward enhanced computer minds with the computational accuracy and efficiency of existing computers, or the enhanced memory capabilities of ordinary computers so it’s unclear how difficult a task it will be to create a mind that blends the best parts of human and computer capabilities.
Another approach to creating a super-mind is to simply create one with a larger set of virtual neurons. However, more may not translate into better in a direct manner. Much of human intelligence is believed to be derived from our experiences. Therefore, ordinary computer minds will have to learn from experience the same way that we do which would take many years just like it does with human minds. Therefore, humanity appears to be safe at least for the moment. Current research hasn’t begun to address the problem of inputting experiences into a computer mind. Rather, most seem to be believe it would be easier to scan an existing human mind into a computer. This wouldn’t be a transfer of knowledge so much as a cloning of it. People would die but their cloned minds could live on. This gives some futurists hope that humans and advanced computer minds will get along. Alternative scenarios, however, are equally plausible as the world of science fiction is replete with examples.
In the long run, however, it appears quite possible that super-intelligent computer minds can be created. Generally speaking, greater levels of intelligence with repairable bodies that are essentially immortal give computer minds enormous advantages over humans.
If all of the following is disturbing to you, then there is also good news. The robot butler’s are on their way as well. The only question is who will they serve? Also, there is a new TV series based on Battlestar Gallactica debuting on YouTube, so you can sit back and watch humans fight back against their robot creations in the comfort of knowing that it’s just science fiction – at least for now.
by Todd Miller
Sorry! No Links