IFTF's Future Now

David Brooks, cyborg

David Brooks is now augmenting. The piece is over the top, but is one of those "if he's doing it, it's either really big or is really over" data-points.

I have melded my mind with the heavens, communed with the universal consciousness, and experienced the inner calm that externalization brings, and it all started because I bought a car with a G.P.S....

I had thought that the magic of the information age was that it allowed us to know more, but then I realized the magic of the information age is that it allows us to know less. It provides us with external cognitive servants — silicon memory systems, collaborative online filters, consumer preference algorithms and networked knowledge. We can burden these servants and liberate ourselves....

Memory? I’ve externalized it.... [I]f I need to know some fact about the world, I tap a few keys and reap the blessings of the external mind.

Personal information? I’ve externalized it. I’m no longer clear on where I end and my BlackBerry begins....

Now, you may wonder if in the process of outsourcing my thinking I am losing my individuality. Not so. My preferences are more narrow and individualistic than ever. It’s merely my autonomy that I’m losing.

Technorati Tags: collective intelligence, end of cyberspace, geoweb, memory, mobility

What is Web 3.0?

Some friends of mine sell T-shirts that read, "Everytime you say 'Web 3.0,' a startup dies." In a more serious vein, Wade Roush writes in Technology Review about two strains of research that are reaching for Web 3.0 status, but are still "closer to 'Web 2.1.'"

The first category of projects is related to the Semantic Web, a vision for a smarter Web laid out in the late 1990s by World Wide Web creator Tim Berners-Lee. The vision calls for enriching every piece of data on the Web with metadata conveying its meaning. In theory, this added context would help Web-based software applications use the data more appropriately....

A second category of post-Web 2.0 projects focuses not on helping machines understand the meaning and the uses of existing Web content, but on recruiting real people to add their intelligence to information before it's used. The best known example is Amazon Mechanical Turk, a kind of high-tech temp agency introduced by the online retailer in 2005. The service allows people with tasks and questions that computers can't handle--for example, spotting inappropriate images in a collection of photos--to hire other Web users to help.

Wade is always worth reading. Alas, he seems to have led his very promising Continuous Computing blog go fallow.

Technorati Tags: AI, Web 2.0

When is AI no longer AI? When it works

Adobe Advanced Technology Lab researcher Bill McDaniel asks the provocative question, "why is it that when we build AI systems, the ones that work are no longer considered AI?"

This has plagued the AI arena for years. It seems that as soon as we develop a new technique... it is immediately classed as useful algorithms, semantic technology, vision processing, anything but AI....

I suspect some of that is because we do NOT solve these problems with emergent AI, intelligence that both arises out of some "ghostly signature" in our knowledge bases and tta seeks to extend itself. It is, I contend, the very absence of these traits that makes us so reluctant to embrace wht we HAVE accomplished as Intelligence.

In some region of ourselves, we know that what these techiques represent, as amazingly successful as htey are, is Artificial Smarts, not Artificial Intelligence. We sense that Intelligence will be recognizable and will be emergent, not algorithmic in nature. And we won't be happy with AI until we get that.

Is he right? I don't know enough about the state of AI to answer the question, but it's an interesting one.

Technorati Tags: AI

Rudy Rucker @ IFTF

Rudy Rucker is visiting the Institute today, talking about his new book, The Lifebox, the Seashell, and the Soul-- one of the most intriguing book titles of the year. Should be interesting.

Technorati Tags: computer, IFTF, science

New AI Brainwave

Having done corporate work in AI, I have a strong interest in practical artificial intelligence. This Infoweek article is a good overview of the direction of new AI work, by IBM, Intel, Msoft and PARC. Note the mention of Bayesian Techniques as being key.

... Now a new generation of researchers hopes to rekindle interest in AI. Faster and cheaper computer processing power, memory, and storage, and the rise of statistical techniques for analyzing speech, handwriting, and the structure of written texts, are helping spur new developments, as is the willingness of today's practitioners to trade perfection for practical solutions to everyday problems. Researchers are building AI-inspired user interfaces, systems that can perform calculations or suggest passages of text in anticipation of what users will need, and software that tries to mirror people's memories to help them find information amid digital clutter. Much of the research employs Bayesian statistics, a branch of mathematics that tries to factor in common beliefs and discount surprising results in the face of contrary historical knowledge. Some of the new AI research also falls into an emerging niche of computer science: the intersection of artificial intelligence and human-computer interaction ...

Brain Models for Intelligence

I reviewed Jeff Hawkin's book: On Intelligence here last year. The founder of Palm Computing suggested a model using the architecture of the brain to build artificially intelligent systems. This is a novel approach, since most AI systems do not attempt to use the brain as a model. His book was very good, but I suggested that he had considerable work to do to implement his ideas. Its another commendable attempt at the holy grail of machine intelligence.

Now an article in today's WSJ: Next Case for Palm Pilot Creators: The Brain and another in BusinessWeek: Jeff Hawkins' Bold Brainstorm, suggest that some progress has been made. He linked with Dileep George, who has started to implement some mathematics based on Hawkin's ideas, and formed a company called Numenta to work on practical applications. Both articles quote Intel scientist Gary Bradksi who says "Even if he's wrong, his theory is better than nothing. And it's 'attackable' -- and that's a good thing." Likely early applications are in machine vision and drug discovery. Well worth tracking. See also article in the NYTimes.

Pamela McCorduck on AI, again

I read Pamela McCorduck's book: Machines Who Think: A Personal Inquiry into the History and Prospects of Artificial Intelligence some 25 years ago, when the uptick of commercial Artificial Intelligence hype began. In part, her work and that of Edward Feigenbaum, of Stanford attempted to translate the technology for management.

Even then McCorduck's book was seen as being a bit too full of breathless praise of artificial intelligence and its future. At the time industrial practitioners like us were too willing to agree and gave out copies of the book freely. But practitioners soon found that it would be harder than that. And while McCorduck and others included discussions of how to deal with the broad replacement of human intelligence by machine, the practitioners started to elicit knowledge, one simple rule at a time.

This is a new version of her book, with a long afterword, which reviews AI progress since her 1979 book. I read the afterword of the new book first, recalling my impression of her original volume. I appreciated the fact that early on she admitted she was responsible for some of the over-hyping of AI in the 80s. She filled me in on what happened to Japan's Strategic Computing Initiative and what happened to the once voluminous DARPA AI support. Still, what I have now is a book that is unchanged in its first part, with all its hype intact ... yet down-toned, corrected and updated in its afterword. OK for historical purposes, but confusing, even misleading for the beginner. Its a good journalistic tale at times, but I would have preferred more technical detail about changes the 25-year passing of time. So a good read, with caution. Below, Scientific American's View:

The Value of Genetic Design

In the February MIT Technology Review: Unnatural Selection, which discusses the increasing value of design by genetic algorithms. They quote David Goldberg, director of the Illinois Genetic Algorithms Laboratory at the University of Illinois at Urbana-Champaign. He wrote an excellent book on the subject: Genetic Algorithms in Search, Optimization, and Machine Learning , which introduced me to the topic, and inspired me to write my own genetic codes. Great, gentle introduction. AAAI also has a useful link collection.

We are increasingly seeing problems which combine design and quantitative efficiency, and these are often combinatorially difficult problems that can lead to genetic methods. These are useful because they can provide models for problems whose mathematical statement are too difficult to specify or solve. I am in the process of exploring evolutionary methods, like genetic algorithms, so if there is anyone out there with substantive experience with vendors, I would appreciate comments.

...“Just as the steam engine created mechanical leverage to do larger tasks, genetic algorithms are starting to give individuals a kind of intellectual leverage that will reshape work,” Goldberg says. “By automating some of the heavy lifting of thought, we free ourselves to operate at a higher, more creative level.” Such freedom comes at a price, of course. It requires that engineers recognize the impossibility of peering into each and every “dark corner” and put their trust in yet another layer of mechanical assistance. But more and more of them are taking that leap ...

Biological Design: Ants Solving Problems

A colleague pointed out the below piece from NPR to me, the ant metaphor used for solving cellphone routing problems ... It relates to the previous post, to what degree is biological metaphor useful for designing an algorithm, a system?

(Search for audio link) ... As ants attack a breadcrumb, they appear to run into, over and around each other in complete chaos. But an article in the journal Nature posits that ants actually follow simple geometric rules as they forage your food. ... based on observation of foraging (food gathering) behavior of ants. As the ants begin their trail, they secrete a substance known as pheromone. Each ant is guided on a particular path based on its concentration of this pheromone. As the number of ants following a particular path increase, the concentration of pheromone on the path increases as well, further making the path more attractive for more ants to follow. This way the ants can effectively move around obstacles in its path and quickly adapt their movements to reach the food source. In this way, each agent working towards its individual goal actually drives the organization closer towards its larger goals...

Ant-based problem solving techniques are a form of agent-based problem solving methods called swarm-intelligence, developed in part at the Santa Fe Institute. Although a fascinating example of biologically inspired problem solution methods, their actual application is not common. One of the best known researchers in this area is Eric Bonabeau of Icosystem, who is working with DARPA and Pharmaceuticals to solve problems. See his book: Swarm Intelligence: From Natural to Artificial Systems, for a gentle intro.

This is an example of biologically-inspired solution methods. Neural Nets and Genetic Algorithm methods are other examples. I have some experience in the case of neural nets as a pattern recognition method. Nets are certainly useful for nonlinear applications, but their biological metaphor linkage is very weak. Use of biology as an inspiration for human design is also a related idea. Its reviewed in Steve Vogel's book Cats Paws and Catapults, which suggests that we have looked for biological inspiration after rather than before the fact. The broad meta-idea deserves more study.

Neurocomputing Data Mining Environment

NeuCom - A Neurocomputing Environment for Data Mining, Knowledge Discovery and Intelligent Decision support systems. Based on evolving connectionist systems (ECOS). Here is a PDF overview of ECOS. Brought to my attention by Steve Cayzer of HP Labs UK . Several demo downloads available at the site, have just started to take a look at them. An interesting comparision would be with SPSS/Clementine, another mining environment that includes stat and neural methods.