Ray Kurzweil Speaks About His Role at Google

Late last year, Google announced that it had teamed up with AI guru Ray Kurzweil to collaborate on AI applications both in search and elsewhere.  Last week, Kurzweil hosted a Google Hangout, and then sat for an interview with Wired Magazine.  Among the highlights was his description of what he is doing for Google:

My mission at Google is to develop natural language understanding with a team and in collaboration with other researchers at Google. Search has moved beyond just finding keywords, but it still doesn’t read all these billions of web pages and book pages for semantic content. If you write a blog post, you’ve got something to say, you’re not just creating words and synonyms. We’d like the computers to actually pick up on that semantic meaning. If that happens, and I believe that it’s feasible, people could ask more complex questions.

Wired Magazine article

China Gets in on the Self-Driving Boom

Joining Toyota, Google, and Audi, China's Army (PLA) announced earlier this month that it has tested its own self-driving car. Code named "Lion No. 3", the car apparently hit a top speed of 105/kmh (65 mph), and overtook 33 cars in 85 minutes in recent tests on China's roads.

Interacting with the Driverless Car

Alexis Madrigal of The Atlantic had an interesting "back to the future" moment when filming a demonstration of the Audi driverless-car system; he was reminded of the reaction that early 20th-century farmers had when confronting an automobile for the first time. See the video for more.

AI Creates Classical Music

Spanish researchers are developing a new kind of AI software that can make classical music on its own. Iamus (named after Greek god of music) uses the 12-note "Western" scale currently, but its programmers are currently adding "Eastern" (Hindu and Arab) scales, and the expectation is that the program will eventually merge the different sounds into something new.

New Information Emerges About Air France Flight 447

More than two years after the Airbus 330 crashed, new information has emerged thanks to locating and recovering the "black box" flight data recorder. In short, human error caused the crash of AF 447. But the human error was in the face of -- and in some cases, enhanced by -- advanced autopilot functions that the human pilots were relying upon. Because they trusted the Airbus to "self-correct", the inexperienced pilots continued to fly the plane in the face of stall alarms. If they had been flying a much less sophisticated vehicle, it seems unlikely that they would not have taken corrective action and avoided the catastrophe. The Popular Mechanics article shows one of the problems that can emerge when "artificial intelligence" is relied upon in the face of other empirical data. Popular Mechanics article

All Your Information Are Belong To Us

Interesting story in today's Globe about Google decision to allow its users to quickly (and easily) see what information the company's servers are storing. Putting this information together in one place (and on one screen) is (at least for Globe author Hiawatha Bray) somewhat disconcerting.

From the looks of the comments, most readers are in contrast "underwhelmed."

AP has written about privacy and security on the net before, so the set of information that Google has accumulated (through voluntary decisions by users) is in fact, somewhat limited.

But the article serves as a good reminder of the scope of electronic surveillance, and its impact on individual privacy. And for that matter, individual and identity security.

Wired for War Comes to Cambridge

PW Singer, author of Wired for War, spoke yesterday at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL).

Picking up on his book, Singer spoke to a full lecture hall that seemed evenly split between undergrad, grad, and faculty. He emphasized the fact that development of military robotics -- whether designed for defensive or offensive capabilities -- was outstripping (by far) legal, ethical, and scientific analysis of the impact that such new devices have on human's capacity to make war. Instead, he spoke of scientists and robotic developers who believed that such "non-scientific" concerns were not their concern, or that the developers would be able to maintain control over the robots (rather than policy-makers.)

Singer also spoke of the speed of this change, and the impact it has been having on both young soldiers (who can fly a drone over Pakistan from North Dakota, for instance, without even having a pilot's license) to senior policy-makers (whose understanding of the rate of deployment of such weapons is, at times, woefully behind the curve.)

There was some discussion of a recent New Yorker article about the Predator and Reaper (a more-heavily armed Predator) programs. The article expressed special concern about the drones operated by the CIA, which author Jane Mayer maintains, does not have the same experience or procedural limits on indiscriminate targetting as does the military. Indeed, the UN (through its Committee on Extrajudicial, Summary or Arbitrary Executions) recently demanded that the US make showing that such CIA-operated drones are within the bounds of international law.

Singer was joined on stage by MIT Professor Missy Cummings, a former Navy fighter pilot. Her work is in the field of unmanned flight, and she discussed a few of her ongoing projects(*).

(*) - More to come on this.

Finally, it is clear that in the post-9/11 era, funding for robotics and related projects will be dominated by the Defense Department. From the Navy (like Japan) studying baseball-playing robots to self-driven cars, it clear that military uses (and plain old military research) will drive this field. And as Singer noted, such research funding comes with costs.

A recording of the Forum is expected to posted online shortly at the MIT Technology and Culture site.

Wired for War

"Wired for War", by P.W. Singer, is a far broader book that the title indicates (although the subtitle -- The Robotics Revolution and Conflict in the 21st Century does give some indication.) Rather than limit his inquiry into the the current state of the robotics world, Singer covers the entire gamut of how wars will be fought in the next century.

But he does spend some (first) exploring the current state of the world. He identifies two current robots: PackBot, built by iRobot; and Talon and its "pissed-off big brother" SWORDS (Special Weapons Observation Reconnaissance Detection System) built by Foster-Miller.

Singer considers himself a child of pop culture, and references to film, books, and even video games are littered throughout the text. Some of the references are direct, as when he wonders whether Star War's C-3PO will be the future shape of robots; others are more obscure. But Singer is at his best when he can merge the influence of science fiction -- from Asimov to the original Star Trek; from James Cameron (director of Terminator) to J.K. Rowling (the 'invisability cloak' described in Harry Potter is inspiring scientists today -- to show how culture influences scientists and their expectations of and for robotics.

Singer also wonders and writes about the role that robots will play in the warrior culture. Separation of the warrior from his opponent has long been a theme of technologic advance (after all, even Japanese samurai eventually acknowledged that their swords were no match for a peasant with a gun). But Singer notes that the very fact of "going to war" -- heading off to a distant place where the laws of normal society are suspended -- no longer applies when a Predator drone can be flown over Afghanistan by an operator who can finish his mission and be home in time for dinner.

The book also raises the topic related to robotics: artificial intelligence. As systems become more and more sophisticated, human oversight becomes an temporal impediment to resolving a conflict successfully. In other words, if machines are waiting for human approval of a pre-emptive attack, the opportunity will be lost, and the machine (and perhaps one or more humans) will suffer as the result. But the alternative: humans ceding control to machines without "human fail-safe" is a topic that, as Singer records, is not one very many scientists working in the field wish to discuss.

Singer does a admirable job of identifying issues that the new technology will raise. While he doesn't try to predict outcomes, he also points to a future that will continue to challenge us. The book is well-worth reading for those who are looking to quickly grasp a state of the robotics/AI world, as it applies to war-making. And in a post-9/11 environment, defense budgets and political resistance to casualties means that more, not less, resources will be available for placing machines "in harm's way."

MSM Discovers AI

Another sign that concepts around artificial intelligence (AI) and robotics are breaking through to the mainstream media (MSM): an article by John Markoff in the NYT's Week in Review.

While Markoff does not break much new ground -- he summarizes the rise of the concepts of "Singularity" and the ethical issues that AI may raise -- TWIR is a marker on the way to reaching opinion-leaders. In the old world of newspapers and magazines, a cover story on AI/Robotics in Time or Newsweek would not be far behind.

But in the brave new world of the MSM, there's no longer a clear formula for moving from techie journals, to creeping into the Science section of the NYT, to 'breaking out' into popular culture. And the fact that it's Memorial Day weekend -- and that much of Manhattan is out in the Hamptons (or wishing they were there like last year) -- may mean that AI is not quite ready for its coming-out party.

Ripped From the Headlines?

It's become a classic TV crime drama teaser: "The new CSI(*): Ripped from the Headlines"

But with today's news that the "Electricity Grid in U.S. Penetrated By Spies" in the WSJ, it's now the headlines that appear to be 'ripped' from drama.

The drama in question is a (sort of) new book, called "Daemon" by Daniel Suarez. It's self-billed as a techo-thriller, and it lives up to the hype.

The book is clearly from the Clancy- / Grisham-type genre, so anything beyond a high-level description will be giving away too much of the plot. However, the main idea of the book: a distributed 'daemon' (or a computer program -- or series of programs -- that run in the 'background' where they are oblivious to the user) that has malicious intent. And because this daemon was programmed by a genius computer-game developer, the Artificial Intelligence (AI) component makes it particularly dangerous.

The author comes from an IT background, and it shows with the technological accuracy of the book. The thin lines and close connection between hackers, game-players, and IT security personnel all play a role; the result is a world of 'reality' that exists -- but is oblivious to those over the age of 40 whose memories of computer games were formed by "Pong" and "Space Invaders." (It is informative that computer games now out-earn Hollywood, although like Hollywood, games are driven by blockbusters that support the equivalent of "straight to video" bombs.)

One of Suarez' themes is the degree to which law enforcement -- from local police to the FBI to even the NSA -- is outmanned, outwitted, and outgunned by rampant distribution of technology. It does take much imagination to think of the possible ramifications for national security, and economic security.

There are some problems with the book: the body count is exceedingly high, and Mr. Suarez' descriptions of bloody endings reads (at times) like the script for "Friday the 13th Part XIX." But the stuff he knows -- the IT and its vulnerabilities -- are far more chilling than the worst sort of horror movie.

The reliance on technology in the modern world -- the vulnerabilities of that reliance to attack -- makes this a must-read for anyone interested in the future of technology. Security is only as strong as its weakest link, and as the WSJ story today shows, malicious agents are always searching for vulnerabilities.

Finally, the book itself has an interesting back story. Suarez had to essentially self-publish it in 2006 under an alias ("Leinad Zeraus"!) It apparently kicked around Silicon Valley and caught the eye of a number of high-profile individuals, including Rick Klau of Feedburner, which was shortly thereafter acquired by Google. As a result, as highlighted by a Wired Magazine article last year, the print-on-demand manuscript became an underground success.

And Penguin/Dutton eventually caught on. The book was re-released this January.

(*) - or "Cold Case" or "Law & Order" or "CSI: Miami" or...

Aptly Named

The London-based Financial Times reports that a self-directed robot (called Adam) at Aberystwyth University (located in Wales) working with scientists from Cambridge University developed and conducted an experiment without human intervention.

Although the experiments were relatively simple -- involving baker's yeast -- the results, and the means of obtaining, were published in the journal Science.

The scientific team is already working a more "elegant" successor to the robot, to be named Eve.

Mapping the Brain

While Honda is working on mapping and interpreting 'brain waves' from outside the scalp, the Allen Institute is taking a more 'inside' approach.

As highlighted in next month's Wired Magazine, Paul Allen has sponsored a $55M, 9-year project that is expected to take until 2012: mapping the DNA sequences throughout the human brain.

Modelled on the Human Genome Project (which itself took 13 years, and was completed in 2003), the Allen Brain Atlas has developed its own robots to efficiently test and classify the 20,000 genes that comprise the average human brain.

The Allen Institute has already completed one related project: mapping the DNA of a mouse brain, located here. (Mouse brains are -- perhaps somewhat disappointingly -- are not dissimilar to human brains.)

Completion of the brain DNA sequencing may open up many possibilities to address mental illnesses and disorders, which often contain a significant genetic component. Indeed, high-profile conditions like autism and Alzheimer's (and perhaps related conditions like Chronic Traumatic Encephalopathy (CTE)) may become more treatable once the genetic codes are unlocked.

More evidence that technological advancement continues even as the world's finances remain unsettled.

"These aren't the droids you're looking for..."

Pretty interesting video from Honda Motors this morning (accompanying story found here) showing a man controlling an Asimo robot -- with just his thoughts! (Jump to 2:32 of the video)

The helmet-like device measures electrical currents and/or blood flow on the scalp based on brain activity; it then transmits that information to the robot. Note that the company did not allow a "live" demonstration of the technology because of concerns about distractions or other difficulties in 'reading' the subject's thoughts.

While the demonstration above is rudimentary, the implications have far-ranging impact. And not just for Stars Wars fans.

It's also more evidence of the importance that Japan is placing on robotics as the key to its technological future, from introduction of fashion model-robots, robo-nurses to other household uses.

Recent Writing

A recent article co-authored in the American Bar Association's SciTech Journal.