Hawking, Musk and Others Call for a Ban on AI based Autonomous Weapons; but Will it Work?

As reported by Engadget: If you don't like the thought of autonomous robots brandishing weapons, you're far from alone. A slew of researchers and tech dignitaries (including Elon Musk, Stephen Hawking and Steve Wozniak) have backed an open letter calling for a ban on any robotic weapon where there's no human input involved. They're concerned that there could be an "AI arms race" which makes it all too easy to not only build robotic armies, but conduct particularly heinous acts like assassinations, authoritarian oppression, terrorism and genocide. Moreover, these killing machines could give artificial intelligence a bad name. You don't want people to dismiss the potentially life-saving benefits of robotic technology just because it's associated with death and destruction, after all.

There's nothing legally binding in the letter, but it lends weight to the United Nations' preliminary talk of a global ban on deadly automatons. If officials, academia and the tech industry are all against removing humans from the equation, it's that much more likely that there will be rules forbidding lethal bots. While that doesn't preclude rogue nations and less-than-ethical companies from forging ahead with their own equipment, you might not see a world full of AI-driven warriors.

Perhaps we only need a refresher on Issac Asimov's, Three Laws of Robotics:
1. A robot (or AI system) may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

Later, Asimov added a fourth or zeroth law that preceded the other in terms of priority:  0. A robot may not harm humanity, or, by inaction, allow humanity to come to harm.

Issac Asimov was one of the first to foresee AI/Robotics in the 1960's, and the need for a code of 'morals' for the machines to operate by in order to work cooperatively and safely with humans.  His book of short stories - 'I Robot', was related to the implications of tampering with these basic laws, and their inherent pitfalls.

Any attempts to disable, or circumvent these basic functions should render the system useless; which from a design standpoint is easier said than done.  That is why it may be difficult or impossible to engineer or design into future AI systems in a reliable way.  Keep in mind that 1% (or more) of the human population can and do violate the first law regularly without regard to the social or moral contract with those around them; and they do this even when it's not in their own self-interest; and in some cases simply because it's 'fun'.  

HUM∀NS does a good job of portraying humanoid AI systems
within the 'uncanny valley' of creepiness.

If we are unable to abstract reason within ourselves to uphold such social contracts, embedding them in an intelligent machine that one day could meet or exceed our own level of consciencness our world will evolve in ways that will be fraught with danger previously unimagined in the history of humanity. Without something like the above, we run the risk of creating an intelligence that could be considered purely sociopathic by human standards, while being superior in many ways (think a robotic Hannibal Lecter); which is why we're so fascinated of late with tales of Terminator (Skynet), iRobot, The Matrix, Transcendence, HUM∀NS, The Age of Ultron, and Ex Machina.

However, even if the western societies are in agreement regarding limits on AI, can we depend on other societies with a different view of this technology such as China or Russia to adhere to these rules; especially if it gives them access to a highly competitive technology (think Atomic Bomb)?  What about potential tech savvy terrorist organizations with a desire to destroy any opposing society standing against them (ISIL comes to mind)?  It also seems possible that at some point even benign organizations may consider advanced defensive AI technology out of fear or distrust; and thanks to modern filmmakers, we all have some idea of how that may turn out.

Maybe that is why Elon Musk is shelling out millions to study how to potentially mitigate AI related disasters in the future, as well billions in a technological space-race to establish a Martian colony as quickly as possible.

China Launches Two Satellites as it Builds a GPS Rival

As reported by Physics.org: China launched two new satellites into space Saturday, state media reported, as it builds a homegrown satellite navigation system to rival the US's Global Positioning System.

A rocket carrying the satellites was launched from the Xichang Satellite Launch Center in southwestern Sichuan province at 8:29 pm (12:29 GMT), the official Xinhua news agency said.

The satellites are the 18th and 19th launched by China as it develops its domestic navigation system Beidou, or Compass. They take the total number launched this year to three.

Beidou is currently centred on the Asia Pacific region but is slated to cover the whole world by 2020.

"The successful launch marks another solid step in building Beidou into a navigation system with global coverage," the satellite launch centre was quoted by Xinhua as saying.

Beidou—named after the Chinese term for the plough or Big Dipper constellation—was announced in 2012, joining the US's GPS, Russia's GLONASS and European Union's Galileo.

It is already used by several Asian countries including Laos, Pakistan and Thailand.

The new satellites will be deployed in "testing a new type of navigation signalling and inter-satellite links" as well as providing navigation services, Xinhua said.

The Beidou system is currently used for civilian services such as navigation and messaging, as well as in the transportation and weather forecasting sectors. It also has military applications.

The Beidou system is one of several GPS like systems deployed from various countries around the world.  While the US Global Positioning System was the first and most widely used system, the Russian GLONASS system has been operating for several years as well.  

The EU is currently deploying their version called the GALILEO system.  Japan has deployed the Quazi-Zenith Satellite System (QZSS), a regional tracking system which only covers parts of Asia.  India too is deploying a regional navigation satellite system called the (IRNSS).

Tracking devices that use multiple satellite systems have some advantages over individual or dedicated satellite tracking devices, including redundancy and potential higher position accuracy.  Providers such as the Swiss u-blox can provide devices that will use multiple satellite systems in order to provide higher accuracy positioning for items like cell-phones that can allow for greater E911 location services.

Satellite systems have been known to go down for periods of time.  The Russian GLONASS system suffered a system wide outage in 2014.

In spite of the advantages of using multiple systems, out of concern the FCC in 2014 issued a requirement that all multi-constellation receivers certified for U.S. use ensure that non-GPS constellations were properly licensed in the United States, in spite of the fact that many have been in operation for several years prior to the proclamation.  

A typical Chinese GPS jamming device
for use in a vehicle.  They are illegal to
purchase or use in the USA.

Concerns over interference and potential jamming and 'spoofing' of GPS signals has also prompted the US and the EU to begin deployment of eLoran systems in critical locations throughout the world as a backup to GPS location technology.

The eLoran and GPS systems have come full circle as GPS was designed to provide higher precision location throughout the world by replacing antiquated Loran marine location technology.  Now eLoran technology is coming back to potentially rescue the satellite navigation and location technology which has become so critical to daily life throughout the world.

Security Experts Demonstrate Ability to Remotely Crash a Jeep Cherokee

As reported by TechXplore: A pair of cybersecurity experts has demonstrated to a writer for Wired magazine, an ability to remotely hack into a Jeep Cherokee and take over some of its functions, and at least in one case, to cause the vehicle to run into a ditch. The demonstration was staged by security hackers Charlie Miller and Chris Valasek and Wired writer Andy Greenberg—the purpose was to showcase the increasing vulnerability of modern cars and trucks to hacking.

Hacking Demonstration

Over the past several years car and truck makers have introduced cellular technology into vehicles to offer customers over-the-internet services, such as automatically monitoring vehicle systems and offering assistance if it appears it is needed. Such systems can alert service workers who can in turn alert authorities for example, if it appears a vehicle has crashed. But that has led to a new type of danger Miller and Valasek insist, drivers and their vehicles are becoming more vulnerable to hacking of the type that can put them at risk of physical harm.

In the demonstration, Greenburg drove the vehicle on a public highway, while Miller and Valasek hacked into its onboard systems from a remote location. They took over the A/C system and the radio, and at one point squirted fluid on the windshield. Then, they caused the accelerator to stop working which of course caused the vehicle to slow suddenly on a freeway. Continuing the experiment, Greenburg drove to an abandoned lot where the hackers disabled his brakes, causing him to drive into a ditch.

The hack was possible due to a lot of work on the part of the security experts, they have been studying onboard vehicle systems for a couple of years and are set to give a talk at this year's Black Hat conference outlining what they have found. They have also been "working" with car makers, keeping them abreast of their findings—Chrysler for example has already put together a patch to protect vehicles such as the Jeep Cherokee that come with the company's Uconnect infotainment system. Owners of vehicles have to download the patch and apply it via a USB stick or have a dealer do it for them. But that is beside the point, the security duo claim, the real issue is that vehicle makers are adding vulnerabilities to vehicles without doing the work required to keep hackers from taking them over and either bricking them, or causing harm.

In an additional report by BBC News, a European security group was able to show that several car infotainment systems are vulnerable to a hack attack that could potentially put lives at risk.  NCC Group said the exploit could be used to seize control of a vehicle's brakes and other critical systems from the infotainment systems.

The Manchester-based company told the BBC it had found a way to carry out the attacks by sending data via digital audio broadcasting (DAB) radio signals.  It coincides with news of a similar flaw discovered by Chris Valasek and Charlie Miller.  However, NCC's work - which has been restricted to its labs - points to a wider problem.

NCC Group was able to transmit the DAB signal using a laptop and a box made from easy-to-source parts

The UK's Society of Motor Manufacturers and Traders has responded by saying that car companies "invest billions of pounds to keep vehicles secure as possible".

Breached brakes
NCC demonstrated part of its technique to BBC Radio 4's PM program at its offices in Cheltenham.

By using relatively cheap off-the-shelf components connected to a laptop, the company's research director, Andy Davis, created a DAB station.  Because infotainment systems processed DAB data to display text and pictures on car dashboard screens, he said, an attacker could send code that would let them take over the system.

Once an infotainment system had been compromised, he said, an attacker could potentially use it as a way to control more critical systems, including steering and braking. Depending on the power of the transmitter, he said, a DAB broadcast could allow attackers to affect many cars at once.

"As this is a broadcast medium, if you had a vulnerability within a certain infotainment system in a certain manufacturer's vehicle, by sending one stream of data, you could attack many cars simultaneously," he said.  "[An attacker] would probably choose a common radio station to broadcast over the top of to make sure they reached the maximum number of target vehicles."

Mr Davis declined to publicly identify which specific infotainment systems he had hacked, at this point.

Lab simulation
In many ways, modern cars are computer networks on wheels.  Mike Parris, of SBD, another company that specializes in vehicle security, said modern cars typically contained 50 interlinked computers running more than 50 million lines of code.

By contrast, he said, a modern airliner "has around 14 million lines of code".

The addition of automated car controls is creating new opportunities for hackers

Such technology allows the latest cars to carry out automatic maneuvers. For example, a driver can make their vehicle parallel park at the touch of a button.

Mr Davis said he had simulated his DAB-based attack only on equipment in his company's buildings because it would be illegal and unsafe to do so in the outside world.

But he added that he had previously compromised a real vehicle's automatic-braking system - designed to prevent it crashing into the car in front - by modifying an infotainment system, and he believed this could be replicated via a DAB broadcast.

"If someone were able to compromise the infotainment system, because of the architecture of its vehicle network, they would in some cases be able to disable the automatic braking functionality," he said.

Tesla Model X Crossover Will Get 'Ludicrous' Mode, Too

As reported by Popular Mechanics: On Friday, Tesla rocked the automotive world by announcing "Ludicrous Mode," a battery and circuitry upgrade that ​makes the new top-of-the-line Model S P90D capable of 0-60 in just 2.8 seconds. All that fun's not just for the sedan set, though:Bloomberg reports that Elon Musk promises the upcoming Model X crossover will also have the option of Ludicrous acceleration.

Thanks to some added mass, the Model X will "only" be capable of sprinting 0-60 in an estimated 3.3 seconds in Ludicrous Mode, Musk said. "We haven't tested it yet, to that's just a guess," he admitted, adding, "that's mad for an SUV, obviously."

It sure is: The next-fastest crossovers we can think of are the Porsche Macan Turbo and Mercedes-Benz GLA45 AMG, both of which take a full 4.2 seconds to reach 60, and neither of which offer the Model X's seven-passenger capability and goofy/awesome gullwing doors.

Of course, you can walk into a Porsche or Mercedes store and buy either of those hot crossovers today. Meanwhile, the oft-delayed Model X is now promised "in a few months," per Musk. We've been hearing that tale for some time now.

As for the long-promised Model 3, the affordable family sedan from Tesla? Musk says that one is "on track in just a few years."

Big Automakers Need Nokia's Maps for their Driverless Car Future

As reported by Popular Science: The future landscape of the autonomous car industry is beginning to take shape with the impending acquisition of Nokia’s mapping software, Here, which already has a focus on “humanised driving.”

German companies Audi, BMW, and Daimler are reported to be making a deal with Nokia to buy the mapping software for more than $2.7 billion.

With this software, these automakers would be positioning themselves to compete in a market soon to be flooded with autonomous taxis by Uber and Google, and be able to offer comparable or premium autonomous driving products. Apple is also reportedly building their car, supported by their hires of a robotic car expert and senior executive from Chrysler, and ex-Tesla designers. (Not to mention potentially having the most-used map data in the world.)

It's also worth noting that Audi has already been developing its own autonomous, or "piloted," cars. In April, an Audi A7 named "Jack" drove a 550-mile span from Silicon Valley to Las Vegas, although the underlying map program wasn't made clear. Another company, Cruise Automation, announced a product that adds driverless capabilities to specifically the Audi A4 or S4, that works on certain California highways.

A robust map data set is crucial to operating an autonomous car--if the car doesn’t know where to go, it can’t go anywhere. Nokia identified this early on, and has already used its LIDAR-equipped cars to create high-definition maps of certain roads and highways, accurate down to 20 centimeters, according to the company. Their LIDAR (Light Detection And Ranging) system features 32 lasers mounted on a car, concurrently collecting 700,000 data points per second at distances of more than 200 feet. (Think the Terminator of remote mapping cars.)

Here's a look at what Here LIDAR data looks like, when put to use in a city like Philadelphia:

Nokia's Here has already used this to maps streets in Silicon Valley, Michigan, France, and Germany, which they announced would be made public yesterday. They’ve also announced a partnership with Mcity, the University of Michigan’s 32-acre faux city solely populated by autonomous vehicles.

Nokia’s Here is is far more accurate than even Google’s Street View car (or Trekker or Trolley or Snowmobile). And Uber, while it recently acquired Bing’s mapping services, has similar, albeit less robust, mapping data to Google. Uber was initially reported to be bidding on Nokia’s services, but dropped out a few weeks ago, according to Reuters.

The ridesharing company has a wealth of data from their drivers, but given that the whole point of the app is a decentralized marketplace for rides, don't expect your Uber driver to sport a LIDAR array anytime soon. Apple, on the other hand, also has mapping vans out on the road. All of which is to say: for companies looking to provide cars in our increasingly autonomous future, maps are a necessary cost of doing business — either making maps of their own from the ground up, or buying them.

Home Made Handgun Drone Attracts FAA Investigation

As reported by GizMag: If you've ever fumbled at the controls of a drone, then the following might make you a little uneasy. An 18-year-old mechanical engineering student has affixed a semi-automatic handgun to a custom-built drone and demonstrated its firepower in a video that has quickly attracted the attention of the authorities.

The video simply titled "Flying Gun" appeared on YouTube on July 10 and has since attracted more than two million views. It shows the home made multi-rotor floating at about chest height amongst the trees and firing off four shots, recoiling and repositioning itself after each discharge.

According to NBC, the flying weapon was created by Austin Haughwout, a student at Central Connecticut State University, and the video was shot on his family's property in Clinton, Connecticut. Haughwout and his father Brett are adamant that they haven't broken any laws, a position that has been backed up by local police. But that doesn't mean authorities are entirely happy with their little experiment, either.

"Our number one job in law enforcement is public safety, first and foremost, the protection of the citizens of the community we serve," Sgt. Jeremiah Dunn, of the Clinton Police Department is quoted as saying. "That’s alarming."

The flying gun has also drawn the attention of the FAA, which has announced it will launch an investigation into the event to ascertain whether or not any rules around air safety were broken.

"The FAA will investigate the operation of an unmanned aircraft system in a Connecticut park to determine if any Federal Aviation Regulations were violated. The FAA will also work with its law enforcement partners to determine if there were any violations of criminal statutes," said Jim Peters, a spokesperson for the FAA.

University of Michigan Opens 32-Acre Driverless Car Test Center

As reported by Mashable: Detroit has long been the home of all things automotive in the U.S., and it doesn't want to cede its position to Silicon Valley when it comes to driverless cars.

The University of Michigan announced Monday it was opening a 32-acre testing facility at its Ann Arbor campus for the development of driverless cars called Mcity.

The facility, which is about 45 miles away from Detroit, will be an immersive simulation of different driving environments. Mcity will feature varying road conditions, construction hazards, road signs deteriorated with graffiti and building facades. It was designed by the Michigan Department of Transportation and the University of Michigan Mobility Transformation Center (MTC).

Mcity received large investments from a variety of automakers including Honda, Ford, GM, Toyota and Nissan as well as tech companies like Verizon and Delphi. Ford has already been testing autonomous cars at the facility, according to a Bloomberg report.

IMAGE: UNIVERSITY OF MICHIGAN

MTC has also expressed interest in testing vehicle to vehicle (V2V) communications at the facility, where connected cars "talk" to each other and to city infrastructure. One of the goals of the project is to have autonomous vehicles testing on Ann Arbor streets by 2021.

This new facility comes at a time when the testing of autonomous cars is spreading out all over the U.S.: Daimler is testing self-driving semi trucks in Nevada; Google recently brought a pair of its cars to Austin; Virginia is opening its public highways for testing.

Cars with autonomous functions will have to be incredibly well-developed before they can be sold to the general public, so extensive testing in a variety of environments is a must. Driverless cars already face much skepticism from the public, and the press hasn't helped matters.

Mcity should also be a boon to the many automakers who are either based in Detroit or have their U.S. operations headquartered there.

“The Google folks are kind of strutting their stuff. They’ve got nothing on us. This is the center of the universe. This is Michigan, it’s not California. We’re not going to let Silicon Valley take this technology, ” Michigan Senator Gary Peters said Monday, as reported by Bloomberg.

A grandiose statement, but one which does reflect the long history of making cars in Michigan."