The $5 million Mind Machine Project (MMP), a patchwork team of two dozen academics, students and researchers, intends to go back to the discipline's beginnings, rebuilding the field from the ground up. With 20/20 hindsight, a few generations worth of experience, and better, faster technology, this time researchers in AI — an ambiguous field to begin with — plan to get things right.

The study of AI is a half a century old, beginning with lofty expectations at a 1956 conference but quickly fragmenting into different specializations and sub-fields. The MMP wants to roll back the clock, fixing early assumptions that are now foundations of the field and redefining what the objectives of AI research should be.

The fundamental problem, it seems, is that the mind, memory and body function both together and separately to solve any number of problems, and the way they work together (and alone) varies from problem to problem. The human mind alone applies various systems and functions to any given problem. Many AI solutions have attempted to solve all the problems with one system or function rather than multiple systems working together as in the human mind, a "silver bullet" approach that hinders real progress.

Likewise, when it comes to memory, researchers have created models that work more like computers, where everything is either one or zero. Real memory is filled with gray areas, ambiguities and inconsistencies, but functions in spite of not always being congruent. MMP researchers also intend to bring computer science and physiology together, forcing computers to work within the confines of physical space and time just like the body does.

The team even proposes discarding the Turing Test, the long-recognized standard for determining artificial intelligence. Instead, MMP researchers want to test for a machine's comprehension of a children's book — rather than a human's comprehension of another human being — to gain a better understanding or the AI's ability to process and regurgitate thought.

It's a big-picture approach to a big challenge, and while it's perhaps unlikely that the team can re-imagine AI in the ambitious five-year window they've given themselves, it very well could shore up some of the loose underpinnings of a discipline that has boundless potential to shape a better world (or, for you SkyNet junkies, limitless potential to destroy it). If nothing else, it's a responsible admission from the scientific community that they simply don't have it quite right, that we need to rethink what we think we know.

Climatologists, take notes.

[MIT News]

Popular Science is your wormhole to the future. Reporting on what's new and what's next in science and technology, we deliver the future now.

Henry Markram, the Blue Brain guy, says in an email to IBM's CTO, that the project is not even close to an ants brain and that the kind of simulations pulled off by IBM are trivial. He also calls the whole thing "stupid", and "extremely harmful to the field." [IEEE via Popsci]

Cats: they're kinda dumb. They only seem smarter than dogs because they're not so friendly, and our society judges kindness harshly. It's true! an interesting theory! Which is why, after mice, simulating a feline-sized brain on a BlueGene/P supercomputer was next on IBM's to-do list. But for all the kitty talk here, this project wasn't specifically about creating a computerized house pet; it's part of a larger, ongoing project to eventually simulate a full human brain. The cat equivalency, derived from the number of virtual neurons and synapses the simulation can manage, at 1.6 billion and 9 trillion, respectively, just gives a sense of how far along the project is: today, despite being the biggest simulated brain ever, it's only capable of simulating the human visual cortex, or as PopMech so delicately puts it, "the wrinkly outer layer" of the human brain.

So how long before a supercomputer can simulate (roughly—since these computer simulations don't have the same neural patterning and learning capabilities of a real brain, among other things) an entire human cortex? Weirdly soon, says the project's lead scientist:

To [simulate a human cortex], he'll need to find 1000 times more computing power. At the rate that supercomputers have expanded over the last 20 years, that super-super computer could exist by 2019. "This is not just possible, it's inevitable," Modha says. "This will happen."

People need to stop getting worked up about the future, honestly: Before we have to worry self-aware robot uprisings, we're going to have to deal with decades of extremely dumb, extremely expensive fake pets. Enforced caution, I believe this is called. [Popular Mechanics]

To put that in perspective, 10 megawatts is the kind of energy a small hydroelectric plant produces—20 watts is only enough juice to power up a weak light bulb. Amazingly, your physical brain runs on this minuscule amount of power, and it's not very efficient. However, embracing this inefficiency could be the key to creating computers that mimic the human brain.

It sounds cockamamy, but it is true. Scientists have found that the brain's 100 billion neurons are surprisingly unreliable. Their synapses fail to fire 30 percent to 90 percent of the time. Yet somehow the brain works. Some scientists even see neural noise as the key to human creativity. Boahen and a small group of scientists around the world hope to copy the brain's noisy calculations and spawn a new era of energy-efficient, intelligent computing. Neurogrid is the test to see if this approach can succeed.

Most modern supercomputers are the size of a refrigerator and devour $100,000 to $1 million of electricity per year. Boahen's Neurogrid will fit in a briefcase, run on the equivalent of a few D batteries, and yet, if all goes well, come close to keeping up with these Goliaths.

So far Boahen has managed to squeeze a million neurons onto his new supercomputer compared to only 45,000 last year. By 2011, he hopes to have 64 million up and running, bringing the project to the equivalent of a mouse's brain.

Ditching reliability and efficiency in favor of organized chaos flies in the face of everything that an engineer holds dear, but the approach does make sense—and reducing the power consumption is the key to upholding Moore's law. But how will this development change our perception of what an artificially intelligent robot might become? Instead of some cold, logical machine that can think for itself, we might end up with robots that are just as stupid and flawed as we are. In other words. it could be a robot on that episode of future Cops running through the bushes with no shirt on after trying to rob a convenience store with a plastic lightsaber. Think about it. [Discover Mag via PopSci]

Officially, this video shows that Carnegie Mellon researchers have given Asimo the ability to detect and avoid moving and stationary objects while in pursuit of a goal. Unofficially, researchers at Carnegie Mellon have fueled the cliched "robot uprising" that we joke about from time to time.

The one positive in all this is that anyone not wearing a large blue dot in the near future will most likely be spared. [YouTube via Plastic Pals via BotJunkie via OhGizmo]

At the conference, the scientists debated research limits on AI, much like their colleagues in genetics and biotechnology have done already with stems cells. Their thoughts were published this weekend under an ominous, dark cloud headline at the New York Times: "Scientists Worry Machines May Outsmart Man"

The location is actually an interesting bit of trivia, as Asilomar was host to a groundbreaking conference on genetics and biology in 1975. At that conference, scientists met to debate their new found ability to reshape life at the cellular level. As the Times notes, the conference ultimately led to guidelines for "recombinant DNA research" and a Nobel Prize for organizer Paul Berg.

Today's scientists are hoping to get similar guidelines into place for AI, although many worried openly that autonomous people-killing robots were here already.

But for every cautionary tale out of Asilomar these days, there's a detractor ready to debunk the warnings with a bit of what they believe to be common sense. Said startup guru and investor Chris Dixon (via Gawker's own Nick Denton, no less), "Is the nytimes serious? AI researchers I know are embarrassed by the lack of progress, not worried about too much."

Indeed, when Wilson chatted with Wired for War author PW Singer during our ominous Machines Behaving Deadly theme week, we learned that a Terminator uprising was unlikely to happen anytime soon because the "preconditions" simply weren't in place—yet.

"The Global Hawk drone may be able to take off on its own, fly on its own, but it still needs someone to put that gasoline in there," he said. Nevertheless, as Wilson added after that comment, "it's not hard to see how this precondition could eventually be overcome." No kidding.

Many of the details from this conference are still coming out, but from what we read today one could definitely infer that there was an ominous, cautious tone present throughout the proceedings. "I went in very optimistic about the future of A.I. and thinking that Bill Joy and Ray Kurzweil were far off in their predictions," said Tom Mitchell, a professor of AI and machine learning at Carnegie Mellon University. "[But] the meeting made me want to be more outspoken about these issues and in particular be outspoken about the vast amounts of data collected about our personal lives."

Sounds like a split decision. Who's afraid of some big bad AI now? [New York Times]

The Turing Test states that to be considered "sentient," an artificial intelligence must achieve a 30% success rate. That means Elbot's accomplishment, while noteworthy, does not an AI make.

Organizer Kevin Warwick from the University of Reading's School of Systems Engineering was excited anyway, and readily compared today's events to the time in the 1997 when IBM's Big Blue defeated chess master Gary Kasparov. "This has been a very exciting day with two of the machines getting very close to passing the Turing Test for the first time," he said.

Perhaps even more impressive was just how believable the computers were, even if the human speaking with them knew they were speaking with a machine.

"Today's results actually show a more complex story than a straight pass or fail by one machine. Where the machines were identified correctly by the human interrogators as machines, the conversational abilities of each machine was scored at 80 and 90%. This demonstrates how close machines are getting to reaching the milestone of communicating with us in a way in which we are comfortable. That eventual day will herald a new phase in our relationship with machines, bringing closer the time in which robots start to play an active role in our daily lives."

Note to Warnick: Read anything on robots written by Isaac Asimov, then get back to me. [Telegraph]

Each match consisted of 500 hands, with Polaris receiving the same cards in one room that the professionals received in another room. The duplicate system was used so that luck would be less of a factor and the game could be played as much on skill as possible.

Historically, computers have been better at playing games where all information is already on the board, such as chess. With poker, players have to make judgments based on incomplete information. Getting artificial intelligence to do that well is a big step, since real life problems are generally more like poker than like chess. [MSN Cosmic Log]

• While machines that could accurately reproduce the sound of human speech, such as Wolfgang von Kempelen's talking head, have been around since the late 1700s, no device has been able to learn the syntactical rules necessary for generating conversation.
• Secondly, the act of hearing and interpreting is more difficult to instill in a machine because of the on-the-fly signal processing that would be required. The complexity of the ear allows it to pick up on the most subtle nuances in sound (according to the article, people can distinguish between hot and cold coffee just by hearing it poured into a glass.
• Roger Schank is a philosopher-programmer who has spent his professional live trying to create a conscious computer that not only has a memory, but can also learn. After years in the field, Schank is skeptical it will ever happen. He says replicating idle chatter and the sheer complexity of speech in general is beyond the abilities of current scientists.
• Steven Pinkner, a Harvard cognitive scientist, says that natural speech could rely on the breadth of one's knowledge, which is "extraordinarily difficult" to endow to a computer.
• R&D efforts in speech recognition began in the 1950s and '60s, but researchers are still hung up on the number of ways to communicate the word yes. Speech engineers for Nuance found that Southerners in the U.S. tend to add "sir" or "ma'am" to responses where as Northerners do not. And "Valley Girl" speak tends to make computers interpret declarative statements as questions.
• Finding it difficult to make a computer able to "learn," scientists turned to brute-force computing and algorithms that relied upon mass amounts of data. But in 1969, high-ranking Bell Labs staffer John Pierce wrote that a speech machine that could recognize, but not understand, was utterly pointless.
• The big emphasis on speech recognition has now moved to emotional analysis, which still uses algorithms to estimate a caller's state of mind. Stanford researcher Elizabeth Shriberg says its impossible to compare emotions in acted speech to emotions in real speech. The escalation of anger, for example, happens in smaller, more subtle increments with authentic speech.
• The most promising breakthrough in emotional recognition is an agression detector that has been deployed through out parts of Europe. Sound Intelligence were able to recreate the processes of the inner ear on a computer, which spawned a device that could learn the sounds of different objects in action and identify them. The Dutch city of Groningen has placed this technology in its pubs, where if it detects excessively aggressive speech in the pub, it will alert the nearest police station. But as Seabrook comments, "This is no HAL."
• Other research labs, like the Speech Analysis and Interpretation Laboratory, have turned to facial recognition to glean emotional insight, but have come up dry. "Emotions aren't discrete," lab chief Shrikanth Narayanan told the New Yorker. "They are a continuum, and it isnt clear to any one perceiver where one emotion ends and the other begins." To add insult to injury, there hasn't been any real demand for emotional recognition outside the call center arena.

So while we might not ever see a robot become a Nobel Laureate, there is one lession to be learned from this New Yorker piece — never talk freely while on hold with customer service. Seabrook learned while at B.B.N. Technologies that they still record the call while you're on hold to assess your emotional state. After a profanity-laced tirade, one annoyed caller took a couple of hits from his bong, waited a little longer, and hung up. [The New Yorker]

Well, we can't really say. The iCub will undergo pretty basic tests, such as shape analysis, nesting different sized cups and stacking wooden blocks, as well as speech development tasks, meaning iCub will be able to name actions it carries out, which will lead to a commentary of its routine. The overall goal was laid out by Angelo Cangelosi, Professor in Artificial Intelligence;

"The outcome of the research will define the scientific and technological requirements for the design of humanoid robots able to develop complex behavioral, thinking and communication skills through individual and social learning."

The team at the Laboratory of Intelligent Systems at the Federal Institute of Technology created the little experimental learning devices to work in groups and hunt for "food" targets nearby while avoiding "poison." Imagine their surprise when one generation of robots learned to signal lies about the poison, sending opponents to their doom.

The little wheeled robots had neural circuitry with about 30 "genes" that determine their behavior, and how much they react to light in the environment. The food sources charged up the robots' batteries while the poison drained them, and by using the genes of the most successful feeders in 50 successive generations, the team was hoping to select the fittest.

Three colonies of bots in the 50th generation learned to signal to other robots in the group when then found food or poison. But the fourth colony included lying cheats that signaled food when they found poison and then calmly rolled over to the real food while other robots went to their battery-death. Eerily wicked, to say the least. Saving the robots' honor, luckily, there were also a few "hero robots" that signalled danger and then rolled to their death to save the others.

Will technology like this make its way into consumer robots sometime? We kind of hope not: not sure I'd like to argue with a Roomba about whether it had or hadn't swept up that mess behind the sofa.
[Discover Magazine]

The trick here is teaching computers to understand the geometric context of objects in a scene. And yes, we hear you. Consider the possibilities for porn. Indeed. – Charlie White

Carnegie Mellon Researchers Teach Computers To Perceive Three Dimensions in 2-D Images [Carnegie Mellon University] Thanks, Thomas!

The whole point of this tournament, where robots play against each other (and by the looks of the photo, torment small children in the penalty box) is to aide in the development of robot technology so that a team of robots will be able to beat the human world champions (c'mon, Andorra!) by 2050. Hopefully the Web will still be around by 2050 so we can report on the tournament's success. – Nicholas Deleon

RoboCup Official Site [RoboCup Federation via Who Ate All The Bratwurst]

"A three-dimensional avatar will lurk inside the cell phone and adjust itself to characteristics of the cell phone carriers. It's just like a sophisticated creature living inside a cell phone. An owner will be allowed to set its first personality by defining the underlying DNA."

The personality of the avatar depends on how you treat it. Something about that term "lurk" is a bit disconcerting. It sounds like this artificial personality can get to be quite a pain in the ass, expressing its loneliness and pleading with its owner to attend to it. Can't you hear it now, whining: We never go anywhere, or You don t bring me flowers any more, or We never talk ? Well, there's always the on-off switch, or as a last resort, uninstall.

Samsung software interacts with phone owner [Textually.org]