Robots taught to deceive

Researchers from the Georgia Institute of Technology have developed algorithms that allow a robot to determine whether it should deceive a human or other intelligent machine and which techniques to use for the best deceptive strategy. The new work, published in the International Journal of Social Robotics, is believed to be the first detailed examination of robot deception.

Georgia Tech's Alan Wagner (right) says that robots capable of deception will likely be valuable for military and search and rescue operations. A search and rescue robot may need to deceive in order to calm or receive cooperation from a panicking victim. Robots on the battlefield with the power of deception will be able to successfully hide and mislead the enemy to keep themselves and valuable information safe.

Wagner and his co-researchers focused on the actions, beliefs and communications of a robot attempting to hide from another robot to develop programs that successfully produced deceptive behavior. Their first step was to teach the deceiving robot how to recognize a situation that warranted the use of deception. The researchers used interdependence theory and game theory to develop algorithms that tested the value of deception in a specific situation. A situation had to satisfy two key conditions to warrant deception -- there must be conflict between the deceiving robot and the seeker, and the deceiver must benefit from the deception.

Once a situation was deemed to warrant deception, the robot carried out a deceptive act by providing a false communication to benefit itself. The technique developed by the Georgia Tech researchers based a robot's deceptive action selection on its understanding of the individual robot it was attempting to deceive.

To test their algorithms, the researchers ran 20 hide-and-seek experiments with two autonomous robots. Colored markers were lined up along three potential pathways to locations where the robot could hide. The hider robot randomly selected a hiding location from the three location choices and moved toward that location, knocking down colored markers along the way. Once it reached a point past the markers, the robot changed course and hid in one of the other two locations. The presence or absence of standing markers indicated the hider's location to the seeker robot.

"The hider's set of false communications was defined by selecting a pattern of knocked over markers that indicated a false hiding position in an attempt to say, for example, that it was going to the right and then actually go to the left," explained Wagner.

The hider robots were able to deceive the seeker robots in 75 percent of the trials, with the failed experiments resulting from the hiding robot's inability to knock over the correct markers to produce the desired deceptive communication.

"The experimental results weren't perfect, but they demonstrated the learning and use of deception signals by real robots in a noisy environment," said Wagner. "The results were also a preliminary indication that the techniques and algorithms described in the paper could be used to successfully produce deceptive behavior in a robot."

The researchers are not unaware of the ethical issues that robotic deception creates. "We have been concerned from the very beginning with the ethical implications related to the creation of robots capable of deception and we understand that there are beneficial and deleterious aspects," explained co-researcher Ronald Arkin. "We strongly encourage discussion about the appropriateness of deceptive robots to determine what, if any, regulations or guidelines should constrain the development of these systems."

MOVE OVER HYDROGEN, MAKE WAY FOR MAGNESIUM POWER

Magnesium burns as a white hot flame in fireworks.

•      Magnesium could work better than hydrogen in fuel cells.

•      There's enough magnesium in seawater to provide energy for 300,000 years.

 "When people talk about alternative energy, hydrogen often comes up. How about magnesium? We’ll see. Today, on Engineering Works! Listen to the podcast."

 

Magnesium is nifty stuff. Pure magnesium is a silvery metal, and you probably remember from high school chemistry that it burns with a hot white flame.

 

While a lot of research has already gone into using hydrogen to store energy, either directly as a fuel or as part of fuel cell systems, some researchers think we should be looking at magnesium as a way to store energy. Magnesium stores about 10 times as much energy as hydrogen. And there’s enough magnesium in seawater to provide energy for 300,000 years

 

Engineers at a Canadian company are working on a fuel cell that uses magnesium, air and water to produce electricity. An Israeli researcher has come up with a magnesium-based battery sort of like the rechargeable lithium-ion batteries we all know about. And a California researcher is working on a way to use magnesium to produce hydrogen for fuel.

All of this sounds good, but there’s a problem. It takes a lot of energy to purify magnesium to a form we can use. Maybe more than we’d get back. One researcher in Japan thinks he has the answer: solar energy to power a laser that would give us the almost 6,700° F. heat needed. We’ll see how that turns out.

Our magnesium power is somewhere in the future, so we’re done. See you next time.

Engineering Works! is made possible by Texas A&M Engineering and produced by KAMU-FM in College Station.

 

The human process behind Google's algorithm !

Matt Cutts, Google's antispam search engineer and part of the team that decides the strategies behind Google's algorithms.

(Credit: Stephen Shankland/CNET)

MOUNTAIN VIEW, Calif.--Google is famous for evangelizing the power of the algorithm. It spends less time talking about the several hundred human beings who influence its algorithm.

To work at Google is to believe in the power of automation. "We've found the best approach to search is algorithmic, it's the only thing that works at scale," said Matt Cutts, a principal engineer at Google responsible for keeping spam out of search results. "We want to use computers all the time."

The obvious reality that never seems to come up quite as often is that even algorithmic Internet search is very much driven by humans. Computers and algorithms may be what handle incoming queries and generate search results hundreds of times a second, but just as Porsche engineers design their engines with slightly different requirements than Ford engineers, Google engineers are constantly tinkering with the recipe for search results that can make or break Web businesses.

Earlier this year, Cutts said that Google "tends to make a change to our core search algorithms at least once a day." In a recent interview with CNET, he reiterated that that pace continues: just last week Google search engineers met to consider 27 separate changes to the more than 200 factors that Google uses to rank search results.

In many ways, this is a natural evolution of Google's quest to organize information. The Web changes quickly and dramatically, and a Google search recipe left unaltered would quickly grow stale and choked with spam. Yet the constant tweaks show that the internal debates conducted by a relatively small number of people can have a significant impact on the way the Internet is presented to millions.

Cutts and Google representatives wouldn't disclose exactly how many people work on search quality at Google, but he did say there are "hundreds" involved in the process at a company that employs more than 20,000 people. They are entrusted with making sure Google keeps its place atop the search world, which affords the company a steady supply of cash allowing it to pursue a seemingly limitless number of other interests.

They range from someone like John Mueller, a Webmaster trends analyst who patiently answers questions from confused site owners in Google's Webmaster Central forums, to Udi Manber, a legendary engineer listed next to Google executives such as co-founders Larry Page and Sergey Brin as the person "responsible for core search." They must balance the desire to provide the most accurate results with the griping of Webmasters knocked down a peg by algorithm changes: not to mention the constant battle by spammers to infiltrate search results.

The Webmaster forums are the first level of Google search where human representatives of the company make their presence felt. A lot of the content in the forums involves tips provided by volunteers for getting your site to rank within Google's search index, Google's trademark approach to customer service. However, representatives like Mueller wade into the discussion threads to provide a more official answer to questions about site tags or inbound links.

Mueller and his counterparts (Google is trying to keep the forums staffed around the clock) also serve as an early-warning sign for engineers higher up the food chain when considering what changes might need to be made to the way Google ranks sites. Cutts considers this a competitive advantage for Google, in that it has a constant human "debugging" system generated by the number of people who care passionately about where they rank on Google as well as engineers who closely monitor feedback in the press, social networks, and Google's forums.

Most of the time the feedback comes from those who could use a basic lesson or two in site design, ordinary small business owners who may know their business cold, but are out of their depth when it comes to search-engine optimization except that they know it's important to rank high in Google results. But on a fairly regular basis, issues generated by complaints from the outside world work their way up to Google's weekly search quality meetings. (Wired ran a good article on the behind-the-scenes process earlier this year.)

Before any such changes are formally proposed, they are tested among small groups of searchers to gather data on the impact, an essential part of the decision making process. A committee then discusses the impact of those changes: at the last meeting, 16 proposed changes required formal discussion while an additional 11 were considered uncontroversial enough to hash out over e-mail.

Each of those changes has the potential to affect search results in a big way. Take for example a recent change to present more results from a single domain when processing a search query like "apple iPhone." Google now presents results from Apple.com sites as six of the crucial top 10 results, believing someone searching that query demonstrated intent to find Apple-affiliated sites, whereas before it wanted to encourage "diversity" within that first page of results. Reviewers, retailers, and others with relevant content related to Apple's prize product suffer the consequences of being pushed farther down the page, or onto the second page.

Those affected by changes to Google's algorithms can file a "reconsideration request" if they believe they've done nothing wrong in seeing their rankings plummet, or if they've made changes to bring their sites back into compliance with Google's antispam rules. That, of course, requires a human to review the site in question and make a determination about the viability of the site.

So behind every algorithm, and therefore behind every search result, is a team of people responsible for making sure Google search makes the right decisions when responding to your query. Obviously, there's no other way it could have happened: Google is a living example of what's possible when brilliant people devise a smart algorithm and marry it to limitless computing resources.

But it also means that search results are dictated by far more than unemotional machines. No Googler sits there and responds to individual requests for "wedding venues in Lake Tahoe." They do decide, however, how Google's algorithms consider content to be relevant to such a query, and which sites are eligible to be presented in those results.