New Life for Semantic Technologies

Cambridge Semantics provides flexible solutions for the data deluge.

By Kevin Davies

February 4, 2011 | A small software company formed by a group of former IBM staffers is breathing new life into semantic technologies. But don’t look for Cambridge Semantics to harp on the term.

“The world of people well versed in semantic technology is still quite small,” says co-founder Lee Feigenbaum. “It’s important that anyone working with our software should not be IT. You won’t see the word ‘semantics’ anywhere in our software. It’s an enabler for us. We can’t build our software without these technologies, but now we’ve built them, we’ve no interest in preaching that you’re using semantics.” (see, “Masters of the Semantic Web,” Bio•IT World, Oct 2005)

“We don’t lead with ‘Semantic Web’ as a marketing term,” adds senior product manager Rob Gonzalez. “We’d like to see more companies like us trying to solve real-world problems. For us it’s about the problems we’re solving.”

Along with CTO Sean Martin, Feigenbaum was one of a group of about 20 people in an advanced technology group at IBM dating back to 1995. The group’s mission was to research new Internet technologies (including semantic technologies) and potential applications for IBM. An early client was a group of cancer researchers at the Massachusetts General Hospital (the Center for the Development of a Virtual Tumor), for which the IBM team helped to deploy semantic technologies for building and sharing models, data, and literature.

In 2007, Martin and Feigenbaum, together with Simon Martin and Emmett Eldred, established Cambridge Semantics and spent a couple of years building up the engineering team and testing early products before launching its first commercial product in late 2009. Luckily, much of the IBM group’s technology was open source. “People have been [saying] that they can’t build libraries or services that are really reusable or discoverable. We think with semantics, you get these benefits,” says Feigenbaum.

Early customers include Johnson & Johnson, Merck, and Biogen Idec, although Cambridge Semantics’ client base includes Fortune 500 companies in advertizing and the oil industry. “This technology can be used in many industries, but is particularly geared toward life sciences,” says Gonzalez. “The data bonanza isn’t comparable to other industries. Life scientists simply need this flexibility.”

Semantic Sidestep

There’s a saying that Feigenbaum admits is neither new nor particularly funny, but it makes a point: If you put ten Semantic Web advocates in a room, you’ll get 15 different explanations of what the Semantic Web is. “You have a loosely coupled set of technologies that people can use for a million different things. People will latch onto something and say this is the real semantic technology.”

Indeed, Feigenbaum is blunt in his criticism of vendors and users alike who proclaim the magical properties of the Semantic Web. “I’ve seen pharma talk about semantics as the ultimate data integration/analysis tool. That’s all well and good and we might get there in the next 10-15 years, but it’s never been what we’ve seen in semantics.”

For Feigenbaum, the interesting bit of semantic technology is the notion of rebranding data in a flexible and agile way. “The underlying properties of semantic technologies let you build very agile, adaptive software systems as data sources changes. It happens in all industries but especially in life sciences.”

Semantics is about flexibility and having a common data model upon which one can take information from a variety of sources—XML, relational databases, or public clinical trial database—and “map them to a common format not constrained by any a priori database schema or XML structure. We saw this flexibility in 2001, and proved it out at IBM. That’s what we wanted to leverage.”

Cambridge Semantics released its first three products in 2009. “There’s no magic to the software,” says Feigenbaum. Just an easy-to-use interface and set of tools that allows users to point to a particular area in a spreadsheet, for example, and ascribe a meaning, e.g. adverse event, assay result. “You have these common vocabularies and data models, and the system takes care of finding values that match and links them together, without having necessarily considered that way of linking things when you set up the system.”

The Anzo Data Collaboration Server, which sits on the user’s server, is semantic middleware, the plumbing that runs and connects everything else. Says Feigenbaum: “It invokes Web services. It has data services and server services that let you build flexible applications.”

Anzo on the Web is a Web 2.0-style application for self-service reporting of any data connected to the data collaboration server. Typically, when users want to use a new data source, Gonzalez explains, they have to change the database, then the application code, then the web tier. “With Anzo on the Web, you can bring the new data source easily into the data collaboration server, and it propagates throughout the system without requiring a lot of manual changes, so it’s resilient to new types of information being added.” The application is designed for scientists who aren’t necessarily IT experts. “They don’t have to go to IT to build new views; they can do it,” says Feigenbaum.

Anzo for Excel is a plug-in to Microsoft Excel that lets people use spreadsheets more effectively. It makes the collection of ad hoc data trivial, says Feigenbaum. “It turns Excel into a data collection application and lets it serve as a user interface for all this data integrated on the server. Now you can consume the data.” A recently-released second version adds an unnamed component that allows users to collect and integrate data from relational databases.

The company announced in mid-January an agreement with Cray to collectively develop and market solutions, including the Cray XMT system and the Anzo product suite. But Feigenbaum is also using the Amazon Cloud, particularly with new prospects. “The data integration paradigm we’re preaching is anathema to a lot of traditional IT,” says Feigenbaum, particularly in regard to procuring hardware, which can sometimes take months. “Many customers run a proof-of-concept in the Cloud with hosted versions of the software. That lets them prove out the technology and work on the procurement to deploy inside their firewall.”

One of the chief benefits of Cambridge Semantics, says Feigenbaum, is that it affords pharma customers the ability not only to pull in and analyze the data from a traditional database but also “the last 10-15% of their data that might be lurking in a desktop spreadsheet or a public resource such as NCBI. They don’t want to spend millions of dollars and 18 months only to get 90% of the way. They need to handle the heterogeneity of Excel and public data. [The missing data] might only be a small part of the total information but it’s a deal breaker.”

Early users span applications from manufacturing quality control to budgeting, allowing customers such as Biogen Idec to compare their actual spend with budget projections. Merck is using Cambridge Semantics applications to procure time on lab equipment.

Cambridge Semantics is still learning from its early customers where its technology can be leveraged. One promising area is in clinical trial data management. Says Feigenbaum: “When you’ve brought together data that don’t normally talk to each other, there’s a bunch of things you can do, such as looking at data for a drug across trials/phases. But some [historical] trials might have used SAS or Oracle Clinical. This is a good way to bring data together,” perhaps to identify reporting discrepancies for regulatory purposes.

An alternative term for semantic technologies that is growing in popularity is “linked data.” “It’s fine,” shrugs Feigenbaum. “It’s just another name. It’s had some success in life sciences, but I don’t care what it’s called.” •

An Immortal Lesson in Design

The Mac's Inventor's Deathbed Gift: An Immortal Lesson in Design For His Son

The man who created the Mac interface gives his son Aza Raskin a final gift that that testifies to the beauty and power of simplicity.

Twenty five days before my father Jef died, on my birthday exactly six years ago, he gave me a present. He had the sparkle back in his eye -- the one that had been reduced by pancreatic cancer to an ashen ember -- when he gave it to me. It was a small package, rectangular in shape, in crisp brown-paper wrapping. Twine neatly wrapped around the corners, crisscrossing back and forth arriving at a bow crafted by the sure hands of a man who built his first model airplane at age seven.
This small brown package was to be the final gift my father ever gave me.
My family does gifts strangely. For instance, we have our own mangled interpretation of Hanukkah, where each person of the family has a night to give out presents. If we have five people home for Hanukkah, we celebrate only five of the eight nights. The joy of gifts are in the giving, not receiving, so before opening your present you must first guess what’s inside. This tradition is "plenty questions," a more forgiving version than the standard twenty questions.
“Animal, vegetable, or mineral?”, I ask.

I stare at the package. In it is my father. The man who invented the Mac.

We are in it for the game of teasing the gift out of the gifter. It's like extracting a ball of yarn from a kitten. The tugs, pulls, and misdirections are the fun. The question must answerable by a simple "yes" or "no." Naturally, the later into the questions we get, the more liberal this rule becomes. We don't break the rule exactly, but answers become a series of "not-exactly"s and "yes-but"s. In past years, the givers have often spent hours creating elaborate disguises for the gifts. I've shaped styrofoam into a fantastic reptilian shape to disguise a pair of earrings for my mother. She guessed them perfectly anyway. There may be collusion going on.
"Mineral," my father says.
We often waste questions on silly asides. We ask about refrigerators and ostrich eggs when the gift is clearly book shaped. But my father is sick. Where there was once the thought that a cure might be found, only fleeting misplaced hope remains like a high school summer fling dissipating in the face of college. We know there isn't much time. Still I ask.
I stare at the package in my hands. In it is my father. The man who invented the Macintosh and misnamed what should be "typefaces" as the "fonts" menu. He never forgave himself for his incorrect usage of English. He groomed me to use language exactingly and considered that mistake a failure of being young and reckless with semantics. The man who invented click-and-drag was now the man who could hardly keep his gaze focused on his son. The box is, of course, smaller than a bread box. It's a question we always ask. My family smiles only out of habit.
"No," my father says. A long pause. "No," he says again, "it is smaller than a bread box. Smaller and sharper." He speeds the guessing game along. Time.

The gift was a message about an entire way of thought.

"Sharper?" I ask. A knife? The box is too small for a typical kitchen knife. It could be a Swiss Army knife. Jef always carries one. The big blade is for food, the little blade for everything else. He gets a bit indignant if you borrow it and use the wrong blade. I have a Swiss Army knife, but I haven't carried it since airport security theater ramped up after 9/11. It probably isn't a knife. Maybe a razor? One can't just ask outright, that doesn't give enough information when you are wrong. Something sharp could be many things. Seeking something more strategic I ask, "Can it be found in a bathroom?"
Long pause.
"Yes."
Three days before he passed, Jef had an accident. He needed to use the restroom, so -- stooped under his arm -- I supported his weight as he hobble to his business. There was something quietly unsettling about escorting my father to a toilet that had been taller than me when we first moved into the house twenty years earlier. I sat him down, walked out, and closed the door. Moments later, a crash jolted the house. I slammed the door open. The metallic smell of water fresh from a pipe whipped my nose and water flooded the floor. The toilet was dislocated from its base like an arm from its socket, and lodged between the toilet and the wall was my father. Despite his size, he looked small and meager. He stared up at me with eyes full of innocent surprise. Why am I on the floor, they asked? Why am I wet? The shocked curiosity in his wide-open eyes is the single most haunting image I have of my father. In the dark space between closing my eyes and falling asleep, that image sometimes steals in and taints me. When it does, there is no help for it. I have to get out of bed and go for a run. Otherwise, sleep will be overshadowed by those confused, guileless eyes.
"It must be a razor?" I ask. He nods his assent with a satisfied smile. He gestures for me to open it. Carefully undoing the knot, the twine, and the paper reveals a cardboard box on which he has written "For Pogonotomy." Of course there is a word for beard trimming, and of course my father knows and uses it. In high school, I played a trick on my teachers: in every essay I used my own made-up word. I used "indelic" to mean something between "endemic" and "inextricably entwined." No matter how many times I trotted it out, not one of my teachers caught me. I used it once in passing with my father and he immediately but gently pointed it out as a non-word. Some men spend time meticulously trimming their beard. My father trimmed his vocabulary. Language is communication, and my father was fastidious about it. Often when we got into particularly deep conversations, he'd pause and continue the rest of the discussion in written form where he could distill his thoughts into a sharp crystalline relief.
The razor itself was a vintage safety razor. Looking at it, I understood the allure. It is an inventive and simple design. The razor takes a flat blade and arches it under a metal shield, giving the blade both greater mechanical strength as well as a protective sheath that keeps you safe. It's the kind of clear insight for which all designers and inventors strive: beauty in turning constraints into advantages.
That razor is a message, rendered in steel and wood, about an incorporeal way of thought. That was my father's final gift to me: A way of looking at the world through the lens of playful questioning, which reveals more than just an answer.Twenty five days later, the razor remained but my father did not.
Jef, I miss you.

Artificial intelligence based on Darwin’s idea

There are too few ethicists contemplating scenarios of a future populated by out-of-control, “differently sentient’’ beings — a job science fiction writers have been doing for generations.

But when machines start thinking for themselves, there are no guarantees robots will be thinking about anyone else but themselves.

Indeed, consider the robots at the University of Vermont that have already begun to evolve.

In an engineering first, and using the same processes of natural selection that made humans so clever, UVM roboticist Josh Bongard has created robots — both real and simulated — whose body shapes change as they learn to walk.

www.youtube.com/watch?v=ckwsvmf3slU

Bongard said that like animals, his robots’ artificial brains evolved not in isolation, but in conjunction with their changing bodies and individual environmental challenges.

The evolving robots in Bongard’s experiment started with only a few moving parts, like tadpoles. But over time, they became creatures with four legs that walked faster, and with steadier gaits than those that were stuck with “fixed body forms’’ (as demonstrated by their responses to being knocked with a stick, for example), according to a UVM announcement.

The robots exist inside a computer program that looks like a 3-D video game, and as prototypes made from Lego kits.

Bongard expects that by using evolutionary processes, engineers will make smarter and more efficient robots, capable of cleaning up construction sites and maintaining roads.

I expect we will see more sophisticated personal robots coming out of evolutionary robotics, as well, including ones that can walk dogs or serve as companions to the elderly.

Research

Softer components make robots safer to be around

Nothing seems quite so chilling as the thought of being touched by robotic hands.

I am of course thinking of the robotic hands of fiction, such as Robby’s pincers on “Forbidden Planet.’’

Now scientists are developing softer hands for the machines that in the future will perform delicate tasks.

Scientists working with The Whitesides Research Group (gmwgroup.harvard.edu/) at Harvard University recently developed a soft plastic gripper for a robot, one that can grasp a raw egg without cracking it or hold a mouse without crushing the wee creature.

The starfish-shaped grippers are embedded with elastic plastic channels that when inflated with air, expand in those areas that are the most yielding.

The result is a grip that is firm enough to lift a fragile object, but which requires none of the programming that a robot with a hard mechanical hands needs to avoid clamping down too hard on an object.

A robot with a pair of the Whitesides starfish-shaped mitts may not be as attractive as Maria, the charismatic female robot in the 1927 classic film “Metropolis,’’ but it would be capable of holding flowers without crushing them or gripping the arm of a patient without breaking bones and bruising flesh.

Robots with a lighter touch might even make good surgeons one day

Humans vs IBM Artificial Intelligence

An Israeli IBM team put the brains into a supercomputer that will take on Jeopardy champs in a human vs. artificial intelligence televised match.

Can a machine beat some of the best contestants on Jeopardy?

A team of about a dozen IBM employees from four countries -- the United States, Israel, China and Japan - have built an artificial intelligence (AI)-powered supercomputer, "Watson," which could be the world's smartest question-and-answer machine.
On February 14, 15 and 16, Watson will take on Jeopardy champs on national TV in North America. The long running, prime-time program poses answers to which contestants must provide the correct trivia question.
Watson, though he's just a machine, will attempt to win a $1 million prize by playing against two of the brainy game show's most celebrated contestants, Ken Jennings and Brad Rutter, in two matches over three days. IBM has pledged that if Watson wins, all the prize money will go to charity.
Named after Thomas J. Watson, the founder of IBM, and the assistant to Sherlock Holmes, the supercomputer will have a fan club watching. Dafna Sheinwald from the IBM Haifa Lab in Israel will be at the taping, excited to see how man will compete against machine.
It was a huge mission to develop a computer that could rival a human's ability to answer spoken questions posed as answers. Sheinwald and her research partner, David Carmel, say the contribution from the Israeli team was search algorithms that help sort out meaningful information from reams of heterogeneous data. That's their specialty at the IBM R&D facilities.
Organizing and linking data
An organization as big as IBM, the largest information technology company in the world, has a whole lot of data it needs to keep straight, from what's on the Internet to its own Intranet, emails, white papers and blogs.
This is the job of computer specialists like Carmel and Sheinwald. And the experience of programming Watson has applications in their day-to-day work. Although it was an exciting and fun challenge, the team members are also, after all, employees of a commercial company.
"We were asked to look for future directions," Sheinwald says. "One interesting application is related to AI in machine natural-language processing. People are trying to make a smart machine and gradually build its knowledge."

Researchers David Carmel and Dafna Sheinwald from IBM Israel helped build Watson, the supercomputer that will play Jeopardy in February.

But the computer needs to put all the data out there in the world into proportion. It needs to build relationships between words. Sheinwald poses one example: There is a woman. She is a reporter. She works for the media. She is a daughter. All the relationships between this data must be strung together with using connecting words and phrases such as "is," "or," "part of," "a [child] of," or "contains," says Sheinwald.
"We look at two objects and their relationships, so when you ask something, [Watson] will know the answer."
Revving the mind of the machine
Carmel says the Israeli team is highly focused on search engine technology and developing it for enterprise networks. "Question-and-answer is not a new area, but the efforts done in this project have been aggressive," he says. "We hope it shows good performance."
To flex the mind-muscles of Watson, the team applied it in more than 50 trial run games against Jeopardy Tournament of Champions contestants this fall. The computer has also taken the same initial screening test that humans need to pass before they are invited to be contestants on the show.
Since the game of Jeopardy is not so straightforward, it became the ultimate challenge for the Watson developers. The answers demanding correct questions from the contestants can be riddled with irony, and subtle meanings and inferences are often not picked up even by smart people, let alone smart computers. Where computers typically fail miserably at these kinds of questions, humans have always had the edge. But maybe not anymore.
Not so trivial applications
The Israeli team is thinking realistically however, and from what they have seen in trial runs, they believe that Watson will not win the prize. However, there is more at stake than winning: they hope their work will be applied to new and novel advances that are much more than games and trivia.
The technology developed for Watson can be used in healthcare to manage patient data, and help doctors make more accurate diagnoses. It could be applied in call-center technology so that talking to a computer is more like talking to a human.
It could also be used by cell phone companies to deliver customer services and coupons as people pass by a certain location.
This could all be thanks to Watson, and the contribution from IBM Israel, which currently employs about 1,000 people -- a quarter of whom have advanced degrees in computer sciences, math and related fields.

Marvin Minsky AI Pioneer.

Artificial Intelligence Pioneer

Marvin Minsky has long been one of the great human intelligences working in the field of artificial intelligence (AI). A professor at MIT, where he has worked since 1957 and cofounded the AI laboratory in 1959, Minsky is also an inventor, philosopher, and author. In recent years, Minsky has focused his formidable talents on trying to impart the human capacity for commonsense reasoning to machines. In this interview, hear Minsky's take on why it's important to recreate human intelligence, what a five-year-old can do that even the smartest machine cannot, and whether someone will ever invent a computer that laughs at Seinfeld.

Marvin Minsky says that when it comes to designing a smart machine, "you mustn't look for a magic bullet"—that is, just a single way to solve all problems. Enlarge Photo credit: © Louis Fabian Bachrach

MIND OF THE MACHINE

NOVA: You were one of the attendees of the original Dartmouth Summer Research Conference on Artificial Intelligence in 1956. Back then, what was the dream? Was the goal really to build a human intelligence?

Marvin Minsky: Well, the goal was to build something that could do everything we do. [The English mathematician] Alan Turing was perhaps the first person to write intelligible articles about this. He discussed the most complicated processes that we know of and explained in a famous 1936 paper ["On Computable Numbers, With an Application to the Entscheidungsproblem"] the idea that you could build one machine that could imitate any other kind of machine, even one more complicated than it. That was the idea of the "universal Turing machine." So here is this great man in 1936 writing about what could happen in the next 100 years, and the rest of us later read this paper and said, "Let's be part of that."

When you say the goal is to tell a machine to do everything that we can do, what does that mean?

Well, a typical person goes through childhood, learns a language; some people learn two or three languages. That's a wonderful thing. Then they learn a profession. They get good at architecture or street cleaning or baseball or something like that, but nobody gets good at many things. The smartest person might be an expert in four or five fields. It's been estimated that to be an expert at something you have to know maybe 20,000 fragments of knowledge or skills. And you can only learn a few of those a day, so it takes a few thousand days to become an expert. But why can't a person learn a hundred fields or a thousand specialties? Why is everyone so limited?
So one of the ideas is maybe we could build a machine or some gadgets to add to our brains so that we wouldn't have to spend 10 years getting good at something. Rather we could spend five minutes getting good at 20 things.

Becoming skilled at baseball or any other endeavor takes years. Could smart machines, Minsky wonders, help us get there in minutes? Enlarge Photo credit: © Matthew Brown/iStockphoto

A NARROWING FIELD

At that conference in 1956, you all thought we would probably have a true artificial intelligence within about 10 or 15 years, is that right?

Well, I think maybe 30 or 40 years, within a human lifetime, we thought maybe we would have machines that would be more or less as smart as a person. And I still think that could have happened.
My picture of what happened, at least in the United States and certainly in most other countries, is that this kind of progress of trying new experiments with computers kept happening in the 1960s and '70s and part of the '80s, but then things tightened up. The great laboratories somehow disappeared, economies became tighter, and companies had to make a profit—they couldn't start projects that would take 10 years to pay off.

"There aren't any machines that can do the commonsense reasoning that a four- or five-year-old child can do."

In the 1950s, '60s, and '70s, almost all of my students became professors teaching other students. But after the 1970s, almost none of my students became professors, because the universities in the United States were filled. Since 1950 the average lifespan in the developed countries has increased one year every four. It's 60 years since 1950, so people are living on average 15 years longer, including the professors. So today, in 2010, very few professors are retiring, and the students have no place to go. Basic research is sort of dying out because there are no new jobs.

With people living longer, professorships that might entice young new talent remain filled far longer. Basic scientific research suffers because of this, Minsky says. Enlarge Photo credit: © Vyacheslav Shramko/iStockphoto

That's sobering. But what you and other AI researchers have found is that it's actually pretty difficult to build intelligence, right?

How hard is it to build an intelligent machine? I don't think it's so hard, but that's my opinion, and I've written two books on how I think one should do it. The basic idea I promote is that you mustn't look for a magic bullet. You mustn't look for one wonderful way to solve all problems. Instead you want to look for 20 or 30 ways to solve different kinds of problems. And to build some kind of higher administrative device that figures out what kind of problem you have and what method to use.
Now, if you take any particular researcher today, it's very unlikely that that researcher is going to work on this architectural level of what the thinking machine should be like. Instead a typical researcher says, "I have a new way to use statistics to solve all problems." Or: "I have a new way to make a system that imitates evolution. It does trials and finds the things that work and remembers the things that don't and gets better that way." And another one says, "It's going to use formal logic and reasoning of a certain kind, and it will figure out everything." So each researcher today is likely to have one particular idea, and that researcher is trying to show that he or she can make a machine that will solve all problems in that way.
I think this is a disease that has spread through my profession. Each practitioner thinks there's one magic way to get a machine to be smart, and so they're all wasting their time in a sense. On the other hand, each of them is improving some particular method, so maybe someday in the near future, or maybe it's two generations away, someone else will come around and say, "Let's put all these together," and then it will be smart.

Commonsense reasoning, which comes naturally to young children, is challenging for a computer, even one as advanced as Watson is. Enlarge Photo credit: Courtesy of IBM

WHAT ABOUT WATSON?

"Watson," the computer that plays Jeopardy!, is doing very much what you described. Its creators at IBM are using formal logic and machine learning and databases—basically a kitchen-sink approach—to develop a computer that can answer questions about a wide variety of things. They wouldn't say that they are building an artificial intelligence but rather the best question-answering machine ever built.

There are some projects that have tried to do commonsense reasoning, but none of them can solve difficult problems yet because they're all using one-way—one or another kind of pattern-matching. There aren't any machines that can do the commonsense reasoning that a four- or five-year-old child can do. No machine that I've heard of yet can answer a question that involves, for example, knowing that you can pull something with a string but you can't push something with a string—a simple thing like that.

But imagine a machine that's playing a game, and the category is "rhyme time." And the clue is: a politician's rant and a frothy dessert. And within two seconds, the machine comes back with "meringue harangue." Now to me that seems like magic. I mean, it's got to be smart, right?

Well, the average person only knows 20,000 words or so. In one-hundredth of a second, a modern computer can find all possible rhymes in those 20,000 words. Then maybe there are 20 other things it can do, like will a certain phrase connect with a certain year. And if you take about 20 of those, maybe you can answer most Jeopardy! questions. I don't know.

"It would be nice to have a machine that could make the next 10 string quartets that Beethoven didn't quite get to do."

But if we're impressed by somebody's program that plays Jeopardy!, then we have to ask, is this because it's taking a lot of data and doing something really stupid like the chess programs do, having no knowledge of chess itself but only knowing how to do, say, 20 of a certain kind of search and that's all there is to it? If that's the answer, then yes, ignorant people will be impressed, but people who understand how it works won't be impressed.

If Watson is little more than a glorified chess-playing computer, than AI experts will not be wowed, Minsky says. Enlarge Photo credit: © Karl Dolenc/iStockphoto

Now, the minute the Watson people publish a scientific paper saying how they did it, then we'll have something to discuss, because maybe some of us will say, "Yes, that is a good new idea, I'm really interested." Or, as in the case of chess programs, we'll say, "Now, I see, this is just another worthless, stupid trick that answers the kinds of questions that most people are interested in for no particular reason"—like what date did a certain baseball player make a certain kind of play. That doesn't require any intelligence to answer if you have the answer in a list.

But Watson has to understand the question, right? That's hard.

Well, you don't have to understand the question if just fitting and matching five keywords will give you an 80 percent chance of getting the answer without understanding either the question or the answer.
I have a good human example of this. My friend Joe Weizenbaum, who was one of the pioneers of AI, wrote a program that appeared to have a lot of common sense. It was called ELIZA, after the character in that wonderful [George Bernard] Shaw play Pygmalion. Joe said he got the idea because he had an aunt who was considered the wise woman of the neighborhood. People would come and tell her their problems—their daughter did this and that and some terrible thing happened and so forth—and Joe's aunt would listen. And after a while she'd say, "Yes, things like that happen." That's all she did that Joe could remember, but he noticed that it was this kind of reaction of appearing to understand that gave her this reputation in the neighborhood.

A five-year-old child knows you can pull but not push something using a string, but does Watson? Enlarge Photo credit: © Adrian Assalve/iStockphoto

I hear what you're saying, but I think, Well, there are computer systems that can understand what I'm saying, ones that can answer questions, ones that are almost beginning to see, ones that can basically begin to move through the world. Is it possible that out of all of those we're going to get an artificial intelligence?

I don't think it will happen without a good architecture. It won't evolve from any particular program. It's a tough one. The problem is that there are some things that impress people, and there are some researchers who, for economic or other reasons, work on things that get an excited reaction from the public.
I'm afraid Watson is that, and I can't tell whether it will ever understand why you can't push something with a string. If you ask the average person why you can't push something with a string, he or she might find it very hard to explain that the string will bend and it won't transmit any force because when it comes to a curve the force will go off the end of the curve [laughs]. No one knows how to think about that.

EMOTIONAL MACHINES

What about music? Do you think that we'll ever have a computer that appreciates music?

There are a lot of good reasons why it would be nice to understand how music affects people and even to make machines that can produce music that affects them. For example, I like some things that Beethoven wrote toward the end of his life. There is a series of four or five string quartets in which he was getting new ideas, and three or four piano sonatas, the last ones he wrote. In the last sonata, he invents jazz in the second movement and shows some tricks you could do in the future with jazz. It was another 100 years before people like Thelonious Monk went further with that. Well, it would be nice to have a machine that could make the next 10 string quartets that Beethoven didn't quite get to do. And it would be nice to make machines that could listen to the old ones and see things that none of us can see.

Someday, Minsky says, computers might be sophisticated enough to hear things in Beethoven's late string quartets that we humans cannot. Enlarge Photo credit: © David Rehner/iStockphoto

I've heard about a French computer program that listened to all of Thelonious Monk and came up with new pieces in his style, and they were remarkably good given the fact that a computer was creating them. Probably it will get better and better at this, but when I hear something really beautiful, I tear up, I show a reaction. The question is: Will we ever have a machine that reacts the way we do?

I don't see why we should doubt that we can make machines that do anything that people do if we understand that people are very complicated machines. In the last 400 million years the nervous system has evolved, and as you can see in any neurology book, there are several hundred specialized organs. Now, when a person reacts to music, maybe 20 or 30 of these little computers are doing particular things, and we don't know what those things are, and maybe each of them is very complicated. Well, it's very hard to understand 20 or 30 complicated things at a time, but maybe someday we'll have a big computer for which that's child's play, and it will understand exactly why most people react to certain kinds of music in such and such a way, and it will say it's obvious.

"Something will seem very mysterious just because you're ignorant, not because it's terribly complicated."

But there's a difference between understanding why someone reacts that way and having the emotional response itself.

I feel emotional responses are much simpler than intellectual responses. It's an unfortunate thing that's happened in most human civilizations that people think that processes like getting angry or jealous or uncomfortable are more complicated than thinking of why triangles with three sides have to have three angles. This idea that emotions are profound and complicated I think comes from the fact that they're very hard to describe. But the reason they're very hard to describe is that they involve lots of rather simple processes that we don't know about. And something will seem very mysterious just because you're ignorant, not because it's terribly complicated.

Will there ever be a computer that laughs at Seinfeld?

The answer is probably somebody, some graduate student, will program a computer that only laughs at Seinfeld.

Minsky takes the long view: In five billion years, the sun will grow so large as to incinerate the Earth, so we should act now, he says, to design smarter machines that could perhaps help us avoid this distant but inevitable fate. Enlarge Photo credit: © Andreas Karelias/iStockphoto

A NEW KIND OF INTELLIGENCE

So when do you think we'll see a true AI?

I think everything depends on the future of the economy. After World War II, in the 1950s, things looked very good from the vantage point of a young person who liked the idea of doing research. The universities were growing, the budgets were high. Every time you invented something it might increase productivity, and the world would get richer. Then something happened around 1980 that I don't understand, when things stopped growing and universities stopped expanding. I know in Taiwan, they started 20 or 30 new mathematics departments in the last decade. I don't think in the United States anything like that has happened.

And why is it so important to recreate human intelligence?

For most people it's not important. For people who have a larger view, one answer is that we may be the only intelligent species in the universe for all we know, and there is likely to be an accident. We know that in five billion years the sun will turn into a red giant and everything will be fried, so everything goes to waste. We also know that in a few trillion years the stars will go out and, in the current theory of physics, the universe will end. Now, we're not smart enough to fix this yet, but maybe if we were a little smarter we could.

By building a new intelligence?

By building a new kind of universe and jumping into it. Otherwise everything is a waste. So from this point of view everything that people do right now is worthless and useless, because it will just end without a trace. Making one of us or replacing us with something intelligent enough to fix the universe or make a better one and jump into it—that should be our main priority.
Now, if you don't live in the world of science fiction that might sound silly. But if you live in my world, everything else seems a little silly.

Intelligence Test for Extraterrestrials, Robots, Humans, Non-Human Animals

We have developed an 'anytime' intelligence test, in other words a test that can be interrupted at any time, but that gives a more accurate idea of the intelligence of the test subject if there is a longer time available in which to carry it out", José Hernández-Orallo, a researcher at the Polytechnic University of Valencia (UPV), tells SINC.

On the hunt for the universal intelligence test.

This is just one of the many determining factors of the universal intelligence test. "The others are that it can be applied to any subject – whether biological or not – at any point in its development (child or adult, for example), for any system now or in the future, and with any level of intelligence or speed", points out Hernández-Orallo.

The researcher, along with his colleague David L. Dowe of the Monash University, Clayton (Australia), have suggested the use of mathematical and computational concepts in order to encompass all these conditions. The study has been published in the journal Artificial Intelligence and forms part of the "Anytime Universal Intelligence" project, in which other scientists from the UPV and the Complutense University of Madrid are taking part.

The authors have used interactive exercises in settings with a difficulty level estimated by calculating the so-called 'Kolmogorov complexity' (they measure the number of computational resources needed to describe an object or a piece of information). This makes them different from traditional psychometric tests and artificial intelligence tests (Turing test).

Use in artificial intelligence

The most direct application of this study is in the field of artificial intelligence. Until now there has not been any way of checking whether current systems are more intelligent than the ones in use 20 years ago, "but the existence of tests with these characteristics may make it possible to systematically evaluate the progress of this discipline", says Hernández-Orallo.

And what is even "more important" is that there were no theories or tools to evaluate and compare future intelligent systems that could demonstrate intelligence greater than human intelligence.

The implications of a universal intelligence test also impact on many other disciplines. This could have a significant impact on most cognitive sciences, since any discipline depends largely on the specific techniques and systems used in it and the mathematical basis that underpins it.

"The universal and unified evaluation of intelligence, be it human, non-human animal, artificial or extraterrestrial, has not been approached from a scientific viewpoint before, and this is a first step", the researcher concludes.

Artificial intelligence tackles stupid commutes

By Luke Rosiak

A team of researchers today discussed technology that could dynamically control an urban street grid, rerouting traffic in the way most efficient for any given conditions. Described as an automatic architecture for real-time traffic, the street grid would learn from experience, remembering which strategies were most effective in the real world.

The scientists were part of a panel on artificial intelligence in transportation at the annual Transportation Research Board meeting, a gathering of 10,000 policymakers, professionals, and academics.

The system was inspired by the human body, a researcher said, and like humans, is fundamentally about "reflection, routine and reaction."

It would default to operating on a timeline and pattern that it has found to be most efficient for a given circumstance in the past. Sensors would monitor road conditions such as the weather and how heavy the traffic is and detect obstructions and accidents and their severity.

When a problem arose, it would respond by changing patterns in the smallest possible area--similar, the researcher said, to the way your body picks up a pen with its hands--no need to disrupt the legs. Similarly, as soon as that unit is freed up, the unit can begin maximizing some other aspect of traffic management, just as your hand could go on to do something else.

Specifically, the system's architecture divides the grid into small rectangular units to respond to conditions. Where it's necessary or most efficient, those units can then interact with neighboring units. That interaction can expand as necessary until many blocks are interconnected--in the maximally efficient way.

And it would evaluate how successful that strategy was, for use in the future.

While much of the math is in place, this is mostly all still theory, however.

One problem involves where to define the boundaries for each small unit, and another is how to detect incidents.