Editor’s note: What would happen if computers could understand the words we type online and turn them into a language that expresses their true meaning? Analyzing data from social media can provide insight into deeper questions about our true motivations and feelings, rather than just statistics.
Pierre Levy, a French philosopher who has been writing about cyberspace since the 1990s, The Canada Research Chair in Collective Intelligence at the University of Ottawa, is working on software that can do just that. He tried to annotate an entire French dictionary in a hyper-language, Because they generate their own language — he calls it the Information Economy MetaLanguage (IEML). All that’s left to do is code it into an automated system.
The principle of IEML is that each word in a given language is represented by a symbol, and the permutations and combinations of these symbols can represent several quadrants of meaning: empty, virtual, actual, things, beings, and symbols (empty, Virtual, actual, Things, and signs). Algorithms that recognize and compute these symbols construct a semantic network from IEML text and work out the relationship between this text and other texts. Writing software that converts natural language into this code, and allows computers to interact with this code, would completely change the nature of online communication and the results we can get from analyzing it.
No doubt everyone, from the NSA to online advertisers, likes this capability because it allows computers to search the web for meaning. But in terms of artificial intelligence, what are the implications of computers being able to communicate real meaning to each other? Let’s hear from a philosopher.
Q: Can you tell me a little bit about what collective intelligence on the Internet means?
Pierre Levy: Ok. Collective intelligence is rooted in animals. Bees and ants are known to have collective intelligence. Social animals also have collective intelligence, signaling each other about danger and where food is. But, in the case of humans, our collective intelligence is much more powerful because we have language, we have technology, we have complex social institutions, and this is a higher level of collective intelligence, which is fundamentally based on symbolic operations.
Throughout human history, collective intelligence has been enhanced by the presence of media that enhance our ability to manipulate symbols. Right now, we are at the stage of manipulating algorithmic notation, but this new era is just beginning, and I am committed to using this new medium to advance collective intelligence.
There are a lot of misconceptions about collective intelligence. The first is that collective intelligence is created, but it’s not. Collective intelligence already exists. The second idea to avoid is that collective intelligence is groupthink. Collective intelligence is the opposite of groupthink. In a philosophical sense, collective intelligence is an integration of diversity and singularity, not Kurzweilian. It’s not uniformity, it’s thinking together.
Giiso, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, Giiso’s research and development products include editing robots, writing robots and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.
Finally, the most common mistake is “Huh, collective intelligence? But there’s so much stupid stuff online.” A philosopher once said to me, you can imagine, “More like collective ignorance!” I would argue that swarm intelligence is not the opposite of stupidity, it’s the opposite of artificial intelligence.
Someone said, “We’re going to make computers smarter,” and they got all the grants. But how do you make people smarter? Collective intelligence is a research project to make humans smarter with the help of computers, not to make computers smarter than humans. This is the true definition of collective intelligence.
You’ve noticed that a question has a long answer, so choose your questions well.
Pierre Levy, photo from Flickr
Q: Some people worry that AI will one day wipe out the human race. What do you think about that?
Pierre Levy: I disagree. Just because a lot of celebrities say this doesn’t mean we should repeat it in platitudes, I strongly disagree. Computers or intelligent software or artificial intelligence programs will never take power, ever. If there is no one to maintain them, they fall apart. This is technically impossible.
And it’s a subtle way of denying responsibility. These machines are made by people, these software is programmed by people, they are the carrier of our will, intention, they have no will and intention, they just extend our intention, thought. They have no responsibility.
Q: What if human knowledge were encoded into machines?
Pierre Levy: Ok, so you studied philosophy?
Q: Yes, a little bit.
Pierre Levy: Ok. So Plato’s Phadro is a very important dialogue. He came up with the same thing you did, but not about machines, but about writing. He said, “What, you’re going to put everything we know in the library?” So in the library, let’s say the Library of Alexandria, where all human knowledge is compiled, we don’t need to teach anymore. Everyone is out of work. I’m so scared! You know, in the 18th century there were 3,000 men in Paris who carried water professionally. Then the water pipe was invented. Ah, they lost their jobs.
So, no, I don’t think we should look at algorithms in such a negative way. I myself, have been and still do research in artificial intelligence, and I have also designed expert systems. I interviewed a group of experts and tried to understand their expertise. I then formalize their knowledge in a creative way, forming hierarchical rules so that the algorithm can understand the calls. They have not become obsolete. Rather, they become masters of the expert system. Once they set up the system with my help, they became self-sustaining. The dissemination of their practical knowledge through this organization is very beneficial.
Like printing or writing, they facilitate the dissemination of knowledge. Converting knowledge into software is a good way to disseminate knowledge and should be encouraged rather than feared.
Q: How does IEML make people smarter with computer tools?
Pierre Levy: You have the World Wide Web, this huge database. They share an address system at the physical level, and you can access all the information from the server’s address, but there is no general classification system. What makes a library valuable is its classification system, otherwise how would you find the right documents?
We have many different classifications, we have ontology, we have many ways to classify languages, and different natural languages have different internal classifications. In fact, information-gathering can be quite difficult when you don’t know what you’re looking for. Google is useful when you know what you are looking for. But when you’re trying to navigate knowledge, it gets a little harder.
A good solution is to turn to Wikipedia, but it is organised in the same way as the 19th century Encyclopaedia Britannica, based on the same divisions and classifications. These interdisciplinary distinctions are not real, and natural language is a problem.
My idea is to have a general classification system that is as flexible as natural language, so that when you want to describe a document you don’t have to follow any rules that say one thing or the other. But all descriptions are in the same language, and you don’t need to learn the language because you can communicate with it in natural language. One of the curious properties of the language is that it is an algorithmic code in which each phrase shows its internal semantic network, and it also computes semantic relationships between one text and other related texts.
If you use this system (where all the data is organized around semantic relationships) to sort the data, to manage the data, and to describe what you’re doing online, new ideas emerge. People co-create these ideas, co-create the way ideas are organized, and through their communication, an ecosystem of ideas emerges. This is the collective intelligence of reflexivity that I preach, and collective intelligence is already out there, especially on the Internet, but we don’t really know what we’re doing together.
You’re familiar with this famous picture of the Internet, huge networks of connections like neurons, and look at the colors. Nodes represent locations and links represent traffic between nodes, but you don’t know what it really means. What we’re going to achieve is a kind of map where nodes represent thoughts and links represent calculations, and that makes a lot of sense.
Many people confuse calculation with quantity, not so much. Calculation is talking about quantity. Even mathematics is talking about quantity. If we can mathematize these structural relations, we can calculate semantic relations. But we need the right code.
Q: I have one more question. You’re building a theory of organizing information that seems to fundamentally change how information is gathered and encoded on the Internet. This is going to give everyone who operates this system unprecedented insight into the data on the Internet at the same level. Do you think there’s a political calculation here?
Pierre Levy: Of course. The power to analyze big data is currently in the hands of some powerful entities, these multinational corporations, Google, Facebook… I don’t know.
Q: NSA, The General Administration of Communications……
Pierre Levy: These governments and big companies, which are the main ones using these big data algorithms, seem to know a lot, but they don’t really have that much insight. I know what’s behind it. It’s mostly data. Data can give you information, but there really isn’t that much.
On a political level, MY idea was to empower people in the same way that the radicals in Silicon Valley in the 1970s wanted to give everyone a computer. And they did. Now everybody has computing power, everybody should have the ability to analyze and understand big data, that’s one of the main points.
How do you do that? There are two main tools. The first is the language itself, and the second is the software that implements the language, which must be open and free. This will be released under the Third Edition of the GPL (the famous free software license). As Richard Stallman told me, IEML is transparent at every step.
Not everyone will be able to contribute to the IEML dictionary because you’ll need expertise — language skills, math skills, etc., but everything at this level will be transparent rather than hidden. The creation of new labels and so on will be open to all. Of course, applications that semantically label data will, by definition, be completely free and open.