There is a remarkable similarity-and one that cannot be taken lightly-between an individual brain and the Internet. The brain has a memory system that allows it to store and retrieve information; the Internet is a massive shared memory, with information being stored and retrieved all the time. Then, there is the network effect: in the brain, it is the network of neurons, with all their connections, that gives rise to its intelligence. The Internet, too, depends on the network effect for its power-the utility of the network increases exponentially as more users are added. Third, both the brain and the Internet have their information protocols.
In an information-driven world, then, think of many intelligent units connected together, with the Internet as the medium. The information is the thought process. The individual nodes- whether mechanical or human-are the neurons. It’s a huge collective brain, no less. But it’s a brain in its infancy: it doesn’t think for itself, it doesn’t learn, it doesn’t solve problems… What several theoreticians have proposed is that the next logical step in the development of the Internet is of it becoming a Global Brain or Overbrain.
On the Internet, each node-an applet, site, server, or whatever -is dumb in itself, but the Net as a whole can exhibit intelligent behaviour. The concept of an Overbrain is based on a few broad ideas: first, that inter-human communication, being as advanced as it is now, is sufficient to make for a global superorganism. Second, that the Overbrain will be more powerful and knowledgeable than individual humans or machines. And third, that it would be something of a grid, with humans as well as machines as nodes on it.
The Web will learn from-and adapt to-the behaviour and needs of its users. No human knowledge “injected” into the Web need ever be lost any more
It’s a very natural idea, actually. People are already acting as nodes. Take Google Answers, for example. You pay and get your query answered-it doesn’t matter to you whether your query has been answered by a bot or a human. The expert who answers your question is a node. You play a game on the Web with a stranger. You’re both nodes. You make a Wikipedia entry. You’re a node.
In what follows, we introduce what people have been calling the Global Brain, which we call the Overbrain. We look at mechanisms by which the Internet could become more intelligent, and ask what could happen thereafter.
Similar Sites, Similar People
Today’s Internet has the trappings of a brain, but not those of a particularly smart one. Think about Amazon’s recommendation system. When you buy a book or CD, you’re prompted with something like “People who bought this item also bought…” This could happen all over the Net! But as of now, when you visit a Web page, you aren’t presented with something like “People who visited this page also visited…”
There could also be links from a page you visited to pages that “similar” people also visited. Web sites such as del.icio.us introduce the concept of “similar people.”
These concepts-of similar sites and similar people-are important. Treating both sites and people as nodes, we’re talking here about clustering. Many aver that in a biological brain, it is clusters of neurons that make up ideas and concepts. Fragmented sets of information points just don’t cut it: there needs to be organisation. In practical terms, an organised Web would be much more useful than the fragmented collection it is today. And in theoretical terms, an Overbrain would need to be organised if it is to learn and think.
The human brain implements what is known as “associative learning.” Neurons that are activated one after the other have the connection strength between them increased. Similarly, concepts that are more frequently used together become more closely interlinked. For example, if you frequently think about “mouse” and “keyboard” at the same time, the two become more tightly integrated together-possibly under a new heading, such as “input devices.”
Applying this to the Web would mean increasing the linkage between two pages depending on how close they are in terms of usage. So if a user views page A and then page B, the linkage between them can be increased, or a new link can be placed between them. If the user views page C after that, there will also be a link from A to C. Now, these links will need a minimum strength in order to be displayed, and so some links could die out altogether. Of course, there’s the issue of unrelated sites being visited in succession, but that’s addressed by the fact that the links between such sites will ultimately die out.
An example of this is at the Global Brain Group (GBG), associated with the Principia Cybernetica Project (PCP). J Bollen, working with F Heylighen, chair of the GBG and AI researcher at the Free University of Brussels, has put up a smart server that does just this. The result is “a dynamic system of strengthening and weakening links between different pages.”
Bollen continues, “These ever-shifting hyperlinks bear a remarkable resemblance to connections that grow and fade in a human brain. On the Principia Cybernetica Web (the Web site of the PCP; http://pespmc1.vub.ac.be), algorithms will reinforce popular links… It’s the first step on the road to the global brain.”
The important thing about this is that the Web is learning from-and adapting to-the behaviour and needs of its users. No human knowledge “injected” into the Web need ever be lost any more. Distant documents could get linked together. Clusters of ideas could form. Take the example of gorillas and conservation; one site could be that of Greenpeace, which mentions gorillas in passing. The other site could be about the social behaviour of gorillas. Mere keyword matching would not link these two as similar; if the Web could learn, it would eventually tie them together.
What if the Web were intelligent enough to learn concepts on its own? The Overbrain could do data mining, where interesting patterns are observed from a mass of data and presented to the user. If a sufficient number of people who consume a certain kind of food (purchased on the Web, of course) develop a vitamin deficiency, the Overbrain could mark this up as a syndrome, and warn buyers of that food about the deficiency.
A Thinking Web
Here, we look at the idea of software agents on the Web, or Web agents. We’ve mentioned agents in The Lizard Of Oz, Digit, January 2006. The Web has thus far been more or less a passive information repository-when you’re looking for information, it’s you that has to be intelligent; you need to supply the search engine with an intelligent query.
On the PCP Web, algorithms will reinforce popular links, while rarely used links will diminish and die. It’s the first step on the road to the global brain
Johan Bollen, Assistant Professor Computer Science dept.,Old Dominion University
Enter the agent. Much like a bat that detects its prey by spreading echo waves and having the prey reflect them, you could let a Web agent “fan out” over the Web and have it return relevant information to you. Say you’re looking for whether gorillas are apes; also, you want to list out what people think are the smartest animals, and whether gorillas figure in that list or not. This is a complex query, so you “weight” your search term: you give “gorillas” and “smart” a high weightage, “apes” and “animals” a lower weightage, and “list” the lowest weightage.
Your agent could make copies of itself and actually have a presence all over the Web, not knowing exactly what it’s looking for, and come up with the best matches. For example, it might spot that a particular domain has interesting information on animal taxonomy and try and apply “ape” in that domain.
Just like regular software agents, Web agents could learn about their users. And agents could learn from each other, exchanging experiences!
This is not the space to analyse what human thinking consists in, but introspection will reveal that when we think about something-as when we try and solve a problem-we start off at a point, and fan our thoughts out to all the concepts and ideas we have. Our thoughts play the role of agents; the concepts and ideas play the role of the clusters we talked about earlier.
Thinking and learning could be combined. The knowledge an agent gains in its journeys across the Web could be passed on to the Web itself. And the learning Web would, in return, facilitate the agents’ job.
For all this to happen, data on the Web needs to be structured better than it is now. It could happen in many ways, but one set of technologies waiting to take off comprise the Semantic Web (www.semanticweb.org), a vision of Sir Tim Berners-Lee, the inventor of the WWW. Berners-Lee, in Weaving The Web, 1999, said: “If HTML and the Web made all the online documents look like one huge book, (Semantic Web technologies) will make all the data in the world look like one huge database.” Documents on the Semantic Web are much more easily processed by machines than HTML documents are, and Web agents are more easily imaginable on something like the Semantic Web.
Eliciting Information
Now here’s where it gets really interesting. The Overbrain could very well become a repository of all the world’s information. Someone high up at Google did recently say that the company would index all the world’s information by 2050-whether he said it in jest or not is not the point; the fact is, we’re certainly moving in that direction. But our point is also that this repository would be a smart repository in a number of ways:
It would be “in touch” with human experts, who would be nodes on it.
It would maintain the consistency of its knowledge, implying that, while being distributed, there would be a centralised check. Think of your brain: you cannot possibly harbour two entirely contradictory beliefs!
To this end, it would elicit information from the experts and from its users.
To elucidate, interaction with the Overbrain need not be one-sided-after all, in the case of our brains, information flows both ways. The Overbrain could keep a check on the consistency of the knowledge it contained; for example, it should not be the case that one node says the plural of “virus” is “viruses,” with another saying it’s “virii.” This is a silly example, but suppose the Overbrain detected such a mismatch. It could consult an expert in biology (or software) and ask for his opinion-using which it would update the data in the mistaken node. This is a massive step: if this were to actually come to pass, the Internet would be 100 per cent reliable, and this alone would make it a hundred times more useful.
The natural question now is, how would an expert be located? We can envisage a scenario in which experts register themselves with the system. Every individual has knowledge unique to him, and that others can benefit from. As more and more people plug in to the Overbrain, we’d have what we like to call a techno-ideosphere. Each one of us would have some form of identification. The domain and level of knowledge that one has to offer would be encoded in the identification, and there could be a sort of “global login” procedure. We can then envisage a system that offers knowledge in return for knowledge-so you log in as a gardening expert, and take help on subjects such as sports. We can also think of knowledge flowing between experts, resulting in debates and discussions that will enrich the usefulness of the Overbrain.
Memes are ideas that can be passed on to others. In The Meme Machine, Susan Blackmore, psychologist and memeticist, has suggested that humans are basically carriers for memes-just like anthropologist Desmond Morris suggested that we are merely carriers for genes. An example of a meme is Darwin’s idea that we descended from the apes. The Internet is, as we can all see, amazingly suited to the propagation of memes. Many memes start off at blogs. An example of a meme currently doing the rounds is the anti-evolutionary idea of Intelligent Design (for more information, visit www.arn.org). This is a good example because a few months ago, nobody had heard about it, and all of a sudden, the Web is abuzz with the phrase.
How do memes tie in with what we’ve been saying about the Overbrain? Well, the Overbrain can be a formal facilitator of meme exchange. Take the case of Idea-X, at http://idea-x.net. From phptr.com, “Idea-X is an online idea exchange. Participants can either propose ideas or ask for ideas to address a specific problem. A suite of tools allows people to see how other members rate each of the ideas and the people proposing them, and to keep track of the best ideas on the site.”
Where Idea-X lacks is that anyone can look at the ideas on the site and use them any way they want. A more rigid system is offered by PLX Systems, “an online market for intellectual property.” At www.pl-x.com, you can buy and sell registered intellectual property. This may sound too commercial, and against the nature of the Internet, but it’s an example of the “expert login” principle we talked about-you give something, you take something.
Idea-X and PL-X are isolated examples; the Overbrain will likely incorporate a vast meme-exchange network. On the Overbrain, memes could sprout all on their own: they could be inferred or detected from documents! They will spread, they will morph, they will be downloaded and uploaded, they will die out-all based on how useful or interesting they are. It happens on today’s Internet, but slowly. Think about the way a good idea hits your brain, activating several concepts and eliciting several responses at the speed of thought. And extend that to an Overbrain, with its rich hyperlinks, its agents buzzing around, its analysis of its own knowledge and ideas!
If we don’t want to be dictated to, we’ll have to be very careful about controlling our dependence, and the evolution of the Global Brain
Daniel C Dennett, Director Center for Cognitive Studies,Tufts University
Giving It The Push
There is an important issue we’ve glossed over. There are certain things the Overbrain would not be able to imbibe all on its own; it would need an initial push-notably in the direction of traditional artificial intelligence. It would have to have some of its links structured as relationships; and it would need to have common sense. Here, we can mention that projects such as Cyc (Refer The Lizard Of Oz, Digit, January 2006) are aiming to build a system that encompasses all of human common sense, for example, that if your father has three sons, you have two brothers. Such a common-sense system could be integrated into the Overbrain.
On the point of relationships, the Overbrain should know, taking the example of a cat, that:
1. It is a member of the mammal class and of the animal class
2. It has feet and a tail
3. It is a pet
And so on. Reciprocally, “tail” should be linked as a member of animal bodies; “pet” should include cats, dogs, and so forth. Semantic Web technologies, which we mentioned earlier, are ideally suited to the marking up of information on the Web in this way.
Today, we interact with the Internet by way of keywords. This is akin to the age of machine language in programming. Humans and the Overbrain will interact in natural language, which means that the Overbrain will need to have natural language understanding capabilities. We can also think about more sophisticated interfaces, and about downloading stuff directly to our brain. But such a tight integration of man and machine is the subject of another discussion altogether!
At this point, we’re talking about an entity that has all the knowledge humans have, that can think, learn and adapt, that “knows what it knows”, and that can easily communicate with us. What sphere of life would be spared from such an entity? Wherever there’s knowledge, information, or intelligence involved, it would be a two-way flow between humans and the Overbrain. Would there be any need for schools and teachers any more? What would universities and think-tanks become but “Overbrain exchange centres”? Would there be a need for non-fiction books any more?
We learn from blogger Stephen Pratt (http://stephenpratt.com) that Heylighen said in 2000: “Whatever problem people have, any kind of question to which they want an answer, it will all become easier because the Web will self-organise and adapt to what people expect of it.” And Heylighen reportedly went on to say that all the technology required to enable an Overbrain is already there -and that “the main stumbling block is the difficulty of convincing the powers behind the Internet to adopt the common protocols that will be needed.”
Being a repository of information is all fine. The Internet already is one. The interesting possibilities arise when the Overbrain begins analysing its knowledge. When it begins forming its own beliefs. When it becomes … aware?
Whether or not the Overbrain will develop consciousness, we’ll leave to the philosophers. But when the Overbrain has beliefs and opinions of its own, it will take on more control. And why not? What sophisticated system in the world today doesn’t depend on computers for its functioning? And when it comes to an Overbrain, who will not hand over control to something that knows much more than what an individual brain can ever know?
Beyond students and learning and books, think of governments, or corporations, and how they might turn to the Overbrain for help in decision making. Would not an overbrain dictate policies? Would it not run everything?
A Skynet?
Skynet is the fictional computer network in the The Terminator series of movies that turns against mankind. If we’re saying we’ll hand control over to an Overbrain, won’t it take over us? Let’s start with a warning from someone in the know. Daniel Dennett is a professional philosopher, and the author of several mind-body-brain bestsellers. From Pratt’s site, Dennett says, “The global communication network is already capable of complex behaviour that defies the efforts of human experts to comprehend. And what you can’t understand, you can’t control. We’ve already made ourselves so dependent on the network that we cannot afford not to provide it with the energy and maintenance it needs.” Dennett goes on, “If we don’t want to be dictated to, we’ll have to be very careful about controlling our dependence, and its evolution.”
It seems that it’s a foregone conclusion for Dennett that the Overbrain will happen. But then again, think of a now-famous article called Why The Future Doesn’t Need Us, in Wired Magazine in 2000: it was written by Bill Joy, co-founder of Sun Microsystems. He began with, “Our most powerful 21st-century technologies -robotics, genetic engineering, and nanotech-are threatening to make humans an endangered species.”
When it comes to speculation, the people up there seem as confused as the rest of us: some predict doomsday, while others speak blithely of better times to come. Regarding Dennett’s comments, we agree that beyond a certain point we will not be able to understand the workings of the Overbrain, but we ask, isn’t the dynamics of the Internet already beyond our comprehension? And as for control and dependence, what would happen to the civilised world if all computers were unplugged?
Coming back to Skynet, what if someone proposed to the Overbrain that it had the power to take over mankind? Wouldn’t it turn evil at that point? Well, the Overbrain would be a reflection of the majority-one idea injected somewhere will not make a difference if the majority thinks (and hopes!) otherwise. But in any case, a doomsday question belongs in the realm of science fiction, where the Overbrain is assumed to have a consciousness and self-awareness and so on: we’re assuming it won’t. Think of the analogy with the human brain: you have a personality, your brain doesn’t. Similarly, humanity might have a collective personality, but its Brain will not. Probably. Maybe.
We hope so…
An important point we’ve just raised is that it’s the view of the majority that will be reflected by the Overbrain. Your current brain state is an “average” of all the thoughts working away inside it; similarly, the Overbrain will reflect the “average” or “most common” view. Is that depressing? Will individualism die? The point here is that this is not radically different from the situation today. It’s the majority that dictates that most people should do a 9 to 5 job, and we accept it as the norm. It’s the majority that likes Microsoft and Windows, and that’s why most of us use it!
As for individuality, it will always stand to be rewarded-recall what we said about the spreading of memes. If you have a great idea-a worthy meme-it will spread faster, more easily than it can today. Our view is that the Overbrain is something that will enrich our lives, much like the Internet has.
As a final point, one can always “opt out” of the Overbrain. Just because Google has a personalised search feature doesn’t mean you have to use it-you can provide your login information only when you want to. Of course, opting out of the overbrain could be a hundred times worse than the situation today when your modem gives up on you, but at least you’ll have the choice!
A Natural Progression
Heylighen said about the Global Brain that it could happen within “just five years.” But he said that in 2000.
AI researchers are notorious for being overly optimistic and often hopelessly wrong with their predictions. We must distinguish well between fact and speculation. What we know for sure is that the Overbrain is the direct evolution of the Internet; that the need for communication can only increase; the need for coping with info overload can only increase; and that the need for sharing human intelligence can only increase.
Whether all this will indeed lead to an Overbrain, we can’t say. But it’s an idea. A meme, we should say.