It was only a matter of time until Samsung launched a full-fledged virtual assistant of its very own — "S Voice" just never quite cut it. Today the company unveiled Bixby, a new assistant that’ll debut with the Galaxy S8 on March 29th. Naturally, it’s meant to help Samsung differentiate itself from Apple’s Siri and Google’s Assistant. Bixby seems different on a conceptual level: It’s meant to serve as a new voice-based interface for controlling your apps, rather than just something that you can ask a few questions.
"Instead of humans learning how the machine interacts with the world (a reflection of the abilities of designers), it is the machine that needs to learn and adapt to us," Samsung senior vice president InJong Rhee wrote in a blog post. "The interface must be natural and intuitive enough to flatten the learning curve regardless of the number of functions being added. With this new approach, Samsung has employed artificial intelligence, reinforcing deep learning concepts to the core of our user interface designs."
Yes, that all reads a bit like a buzzword salad, but there’s definitely a need for a different sort of virtual assistant. Siri has mostly been a disappointment, and Google Assistant is still getting off the ground. The big problem for Samsung is that, historically, it hasn’t been known for innovating with software as much as it has with hardware. Just look at how disappointing S Voice has been for the past few years.
As the company explains it, Bixby will be able to control almost everything you could do in an app with a touchscreen. It will reportedly also be contextually aware, allowing you to jump between voice and touch commands as you’re trying to accomplish something. Finally, Samsung claims Bixby can understand incomplete commands, so you hopefully won’t be forced to memorize rote phrases.
It’s still unclear what’s powering Bixby at this point. Samsung picked up the AI assistant company Viv last year, which showed off some intriguing capabilities when it was first demoed. But according to the WSJ, Bixby is actually based on an upgraded version of S Voice. We’ll most likely see Viv’s technology integrated into Bixby down the line, though.
Watch out, birds. The drones are coming for your jobs. Researchers at BMT Defence Services (BMT) and the University of Bristol in Britain have built a fixed-wing UAV that can land as well as its avian counterparts, reports Popular Mechanics. Although BMT’s project is currently part of a wider defense program called Autonomous Systems Underpinning Research, the team believes their drone could one day be used for other tasks like putting out fires or delivering packages.
When birds land, they perform a "deep stall," meaning they swoop in at low altitude and angle their wings upward before landing. BMT’s drone does this too, thanks to a new morphing wing that can sweep forwards and backwards to create a pitching moment, or twist to allow the aircraft to roll. With this kind of high maneuverability, researchers envision a future where UAVs can easily fly through urban environments, dodging lampposts and power lines.
But, it takes more than high-tech wings for a drone to safely do a deep stall. The team also had to build it a bird-like brain, one that could compensate for slight changes in speed, wind, angle and wing position. They achieved this through something called "Q-learning," a technique where an artificial intelligence learns an optimal course of action by raising its "Q," or satisfaction level. Like a kid in an arcade, it kept trying to beat its high score, until it worked out how to get from its starting point to its destination. After about 5,000 practice attempts, BMT says the drone pulled off a soft landing without a runway.
AIDS patients must endure a lifetime of drugs because the virus conceals itself in the immune system and reactivates with a vengeance once the treatment stops. However, French scientists have discovered a marker that makes it possible to identify dormant, HIV-infected T-cells from healthy ones. That could lead to drugs that target those "reservoir cells," eradicating the virus completely and curing the patient.
Expensive HIV antiretroviral drug regiments massively knock down levels of the virus, letting patients lead relatively normal lives. However, a tiny number of the virus cells remain in around one in a million disease-fighting T-cells. Once the therapy is stopped, the virus rapidly multiplies, meaning the patient can never be totally cured.
However, while working with infected cells in a lab, the team from Montpellier University noticed a biomarker protein called CD32a that wasn’t present in healthy cells. They then studied blood samples from 12 HIV patients living under treatment and isolated cells expressing that marker. Using an antibody that sticks to CD32a, the team pulled cells expressing the protein from those samples and, as expected, they were laden with hidden HIV. "You absolutely could not have done that before now," says lead author Monsef Benkirane.
The fact that this work has been done by such competent investigators, and the data looks good, makes me optimistic.
It turned out that almost all the T-cells expressing the same protein were "reservoir cells" loaded with the virus. By contrast, neither normal T-cells or those that carried active HIV virus (that can be killed by antiretroviral drugs) did not show the same marker.
Unfortunately, CD32a was not present on all the T-cells caching latent HIV, so drugs targeting the marker wouldn’t kill enough of the virus to cure a patient. However, it’s still a huge breakthrough, marking the first time researchers have been able to identify latent virus cells after trying since 1996. The technique could be used to augment "kick and kill" treatments that activate latent virus and then kill it with antiretroviral drugs.
Next, the Montpellier team will try to duplicate the findings on a more diverse group of patients and test tissue that HIV usually infects from the stomach and lymph nodes. Tony Fauci, director of the US National Institute of Allergies and Infectious Disease, is cautiously hopeful that it will lead to something. "The fact that this work has been done by such competent investigators, and the data looks good, makes me optimistic," he told Nature.
The Gmail app on Android could be your new payment hub, assuming you are a fan of Google Wallet and need to send cash to people on the regular through email. Just like you have been able to on the web for some time, Google is adding an option in the Compose menu to send cash without much work.Â
Once the feature hits your Gmail account, you’ll send money with Google Wallet sort of like you do an attachment. As you hit the option to “Send money,†you’ll walk through how much to send and with which account, before the task then attaches to the email you are sending. You can include memos, send cash to people who don’t even have Gmail, and let the person receiving money direct the funds to their bank account, instead of Wallet.
The two GIFs below show you exactly how this will work. Anyone get the option yet?
AT&T on Friday promised a “truly unlimited plan” for its prepaid phone customers, but the new offering limits video quality to standard definition, caps overall speeds at 3Mbps, and prohibits mobile hotspot use.
For $60 a month, GoPhone (AT&T’s prepaid brand) now offers a plan without traditional data caps and overage charges. “We’re excited to offer a truly unlimited plan for our AT&T GoPhone customers,†AT&T VP Bob Bickerstaff said in the announcement. But the plan is always limited to an overall speed of 3Mbps, and video is limited to 1.5Mbps (about 480p), AT&T said. The cost is $65 a month for customers who don’t sign up for automatic payments.
The plan is basically identical to one AT&T recently introduced for postpaid customers. But while this is the only unlimited option for AT&T prepaid customers, AT&T’s postpaid customers have the option of paying $90 a month for an unlimited plan without the 3Mbps speed limit. The $90 postpaid plan also includes 10GB of high-speed tethering data a month and lets customers enable HD video.
Like other AT&T unlimited plans, GoPhone unlimited customers who use 22GB a month can be throttled during periods of network congestion for the rest of the month. If you’re not in a congested area, you keep receiving speeds of up to 3Mbps. This is similar to other carriers, although T-Mobile USA sets its throttle point at 30GB.
AT&T prepaid customers can also opt for a 6GB-per-month plan that doesn’t limit speeds to 3Mbps. Video is limited to 1.5Mbps by default on the 6GB plan, but customers can enable HD video, which uses more data. There are no overage fees after you hit 6GB in the month, but data speeds are slowed to 128kbps. The 6GB plan costs $40 a month, or $45 without autopay.
Data storage needs to keep up with our desire to snap pictures, download clips from the Internet, and create new digital documents. Since the early stages of computer technology, magnetic storage has been the method of choice to handle digital data. It has stayed that way because of our ability to continually shrink the area used to hold a single magnetic bit.
But we’re closing in on the limits of this approach, as clusters of three to 12 atoms have been used as a functional system. Last week, however, scientists demonstrated the ability to magnetically store data in a single atom.
The basics of magnetic storage
Magnetic storage requires the magnetization of a ferromagnetic material to record data. These materials rely on the atom’s electrons, which themselves behave like tiny magnets. The electrons carry a magnetic dipole moment that is determined by the direction the electron spins and the shape of the path the electron travels (quantum mechanical spin and orbital angular momentum, to be technical). There are only two directions the electron can spin, either “up†or “down.â€
Although this system is dynamic, its two stable equilibrium states, “up†or “down,†provide what’s often referred to as magnetic bistability. Having a bit stored indefinitely usually involves a cluster of atoms all set to the same state. This provides a bigger signal, ensuring the bit is maintained even if any given atom doesn’t behave stably.
So, while there were many advances in the miniaturization of magnetic bistability, there were some obvious questions about the limits it could reach.
A single-atom approach
In this investigation, scientists worked with holmium atoms (Ho) supported on magnesium oxide (MgO). Although many Ho atoms formed clusters on the surface, the researchers identified single atoms located atop oxygen to use as a magnetic storage material.
Using a scanning tunneling microscope, the researchers applied current pulses to the Ho atoms to switch the direction of the magnetic moments, demonstrating the ability to control the magnetic behavior of individual Ho atoms. They were then able to read the magnetic patterns by placing another magnetic material in close proximity, which enabled electrons to tunnel from one magnet to another (tunnel magnetoresistance).
Magnetic storage requires the ability to read and write information, but it also requires information to be retained over time. To gauge the storage retention time, the scientists observed how long a Ho atom would remain in a single state after being switched using a pulse of current. They found that the bits would last for hours before starting to randomize.
The researchers performed a clever trick to test that what they were seeing was truly due to the magnetic moment of the Ho atom. They used the microscope tip to place a single iron atom near the Ho atom. In this experimental set-up, the iron atom functions as a local magnetometer, since it has an external out-of-plane magnetic field that’s influenced by nearby magnetic materials. When the scientists applied current pulses to the Ho atom, they saw corresponding shifts in the magnetic field of the iron atom, demonstrating that the individual Ho atoms did in fact possess two distinct magnetic orientations.
Finally, the researchers explored the ability of an array of two Ho atoms to store two bits of information, again using a nearby Fe atom to locally read the magnetic state. They were also able to use other advanced techniques to remotely read the magnetic states.
These experiments demonstrate that high-density magnetic storage at the atomic level is possible, though significant research is still needed to further develop this technology and understand the practical feasibility. After all, something that’s only stable for a few hours would require a very different approach than past magnetic media.
Alarms and flashing lights have begun blaring in a hotel meeting room, much to the chagrin of CIA Senior Collection Analyst David Clopper. The officer, along with a team of his colleagues, is in the middle of demonstrating his department’s training materials, and he has to account for these materials before leaving the room. In some cases, this might require picking up a few stacks of paper, or some pamphlets and flyers.
Clopper, on the other hand, has to pick up dozens of 10-sided die, over 100 colored gems, and hundreds of custom-printed cards that describe hypothetical crises across the globe. That’s nothing compared to the mess of cards on the other side of the room, which, moments ago, were being used to track and capture the elusive drug kingpin El Chapo.
They were this close.
Uh-oh, Venezuela’s causing trouble again!
The two groups of South By Southwest attendees split up in this conference room hesitate to get up. They were testing out the weirdest training exercise the CIA has ever publicly revealed: board games. These aren’t off-the-shelf games; instead, CIA officers designed and assembled these elaborate tabletop games to reflect the realities of the CIA’s day-to-day operations.
Don’t expect to see CIA training game Collection in board game shops anytime soon.
Each acronym on the board represents a real-life intel collection agency.
I didn’t get a shot of a card that explains what each role in the game is, but this explains how the dice-rolling process works when you attempt to gather intel for a particular campaign.
These are some of the campaigns you may be saddled with during a game of Collection. "Venezuela Causing Trouble is easily our favorite.
Almost every "crisis expansion" card ratchets up one of the three "fire" meters on the board, but these can also make collection agencies ramp up their efforts accordingly. Kind of like in real life, when shit hits the fan and agencies start actually taking your phone calls and getting things done in response.
You’re going to want the "collector response" numbers as high as possible, because in the board game, these represent which dice rolls will count as intel successes.
Clopper recalls one day in 2008 when his "boss’s boss" called him into a meeting and asked him to develop new internal training exercises. Normally, these exercises test whether recent lessons and seminars have been absorbed by officers, and they usually involve "teams, flip charts, and briefings," Clopper says. "Incredibly boring." But Clopper had now been at the CIA long enough to reshape its exercises, his boss said, and he got excited: "I’m a gamer. I enjoy games, video games, tabletop games. Could we bring games into learning?"
He used SXSW to present three board games made for his training exercises over the span of a four-year period, one of which is still in development. The first is the one we got the most hands-on time with during SXSW: Collection. If that dry-as-a-desert name isn’t a good indicator, rest assured—this is not a game meant for retail or for the highest ratings at BoardGameGeek.
Collection compares favorably to the popular cooperative game Pandemic. In Clopper’s game, a group of players must work together to resolve three major crises across the globe. The object is for players, who each represent different types of CIA officers, to collect enough relevant intel to resolve all three crises. If any one of the three impending disasters boils over (as represented by three increasing "fire" meters), the team loses. Every game must have at least three players to fill the roles of "political advisor," "military advisor," and "economic advisor." Those three are only able to collect intel in their specific fields, while additional players (up to seven on a team) have their own specialties.
The difficulty comes from the low number of actions each player can do per turn, along with how quickly the fire meter ratchets up. Players can take two actions per turn: move your agent around the globe, foster new relationships in your current location, or attempt to acquire intel. The latter move is a dice roll. Those rolls are bolstered by having more relationships in a certain zone, by having fellow agents in that same zone (if they offer a specific "colocator" bonus), by having bonus cards, and by how good your relationship is with a specific agency.
Let me be blunt: These boosts and bonuses are not doled out in ways that game designers would call "fair." This game is hard. That’s how Clopper likes it.
"This game is really about value of collaboration," Clopper says. "We saw [game sessions] where people took the time to talk to one another, talk about your [individual] capability, how we can work together, or thought ahead, strategized, ‘I go first, you go next.’ They tended to win. The tables where someone would go on their own and do what they wanted, or do their own thing, or didn’t collaborae until too late, they couldn’t catch up to the crises. It was a simulation of what we do, but also teaching the importance of working together."
Thanks to detailed cards about department descriptions, our civilian test group was able to squeak some fun out of this game by role-playing. I enjoyed knowing that one of my turns revolved around me working as a political analyst in collaboration with the CIA’s Department of Operations. I peppered my table-talk with descriptions of my very "clandestine" efforts.
Magic: The (covert intel) Gathering
One of Clopper’s other game designs, which we didn’t get to demo.
Clopper’s unfinished game about satellite grid creation and management. Should your team use fewer satellites and make them more powerful? Use more to account for failures? Either way presents different challenges (which is great for training scenarios, the CIA says).
Another Clopper game game, Collection Deck, focuses less on collaborative work and more on the sheer act of collecting intel. It also differentiates between "things that are secret and not secret." This collectible card game plays like Magic: The Gathering, Clopper says. Multiple players work to resolve intelligence problems (represented by cards laid on the table) while dealing with out-of-nowhere issues (represented by "reality check" cards that players can use against each other).
"You try to use a card representing an overhead satellite—you want to use that to take a picture," Clopper says. "Another player throws down a ‘ground station failure’ card. Now you can’t use that one."
The game was built to simultaneously deliver context about intelligence-gathering methods and to help officers understand how real-world hiccups can get in the way while in the act. "People would come up to me after [a session] and say, ‘David, I learned bout something I didn’t know existed before,’" Clopper says. "’I think we can use this on a real intelligence problem I’m tracking.’ It’s a game, but it had real mission impact."
His unfinished project, tentatively titled Satellite Construction Kit, will have players cooperate to manage resources, budget, and time to build and maintain connected satellites. While managing costs and deciding whether to build redundancies in this satellite network, Clopper says players will deal with challenges like the Department of Defense demanding certain capabilities, or Congress slashing your budget by 10 percent. "That never happens," Clopper says with a laugh about the latter.
This is one of two giant, confusing boards for Kingpin: The Hunt for El Chapo.
CIA officers refused to answer Ars’ questions about where these photos came from.
Sadly, I didn’t get to play this one and explore exactly the game’s mechanics or how chits get shuffled around the board.
At least I got some photos before the fire alarm went off.
The other board game at the event, which I didn’t get to test thanks to that unfortunate fire alarm, has a far cooler name: Kingpin: The Hunt for El Chapo. This one was co-developed by CIA Intelligence Educator Volko Ruhnke, who happens to have designed his fair share of publicly available board games, as well. "I have two Golden Geeks," he brags.
Kingpin, which was made in collaboration with the Defense Intelligence Agency, is used "to train analysts who might work with law enforcement and other partners around world to find a well-armed, well-defended, well-protected bad guy," Ruhnke says. "It’s a game about finding a fugitive from justice who, if not found sooner than later, will likely do harm to innocent people and harm US interests."
This is a two-team versus affair, Ruhnke says, because analysts need to be aware of how their targets work. "It’s a real brain-on-brain affair," he says. "The fugitives we’re analyzing, they’re analyzing right back." While I, unfortunately, didn’t get to explore all of the methods that each team uses to react to the other in this game, I did spy a "facial surgery" option on the board for Chapo’s side—which goes to show that Ruhnke’s game has made room for a vast range of response possibilities.
Bonkers data models like these are better represented in board-game form, the CIA argues.
Sam Machkovech
Ruhnke says that board games are particularly useful in helping officers and analysts synthesize giant criss-crossing charts’ worth of data to understand complex real-world situations. He points to modern insurgency and counter-insurgency engagement in Afghanistan as an example. First, he presents a bonkers-confusing chart while listing off its myriad elements and actors on all sides. Then he shows an image of people playing a board game.
"People playing a game, together they’re experiencing the designers’ mental model of insurgency in Afghanistan and sharing that model," he says. "They are learning it, very quickly, because they’re inside, operating in it. Pushing levers, pulling cords, seeing what happens. Stories are very sticky, and they’ll remember their own stories. " He also insists that this is a great way for outside perspectives to critique an office’s biased model of how insurgency may play out and receive more dynamic feedback. "The greatest power of simulation games is that players have to operate these games themselves and know the rules."
With that in mind, as I briskly walked out of the hotel with fire-alarm alerts ringing in my ears, I couldn’t help but wonder: Had we seen too much? Had the officers noticed a few SXSW attendees with particularly thick foreign accents and decided enough had been disclosed? Had we, as unwitting board game players, suddenly found ourselves as part of a meta game?
The whole thing certainly made me think of another CIA officer’s comments during a SXSW panel—in which game-based exercises wound up teaching CIA staffers more than they expected. One game about global map management included giant touch-panel screens, which required an IT staffer to routinely visit two competing teams and keep the systems working. In one session, the "blue" team soundly won—an unusual result, considering the game was specifically built for the other "red" side to have better odds. A post-game debrief with both teams of players didn’t turn up any interesting details… but a throwaway comment by the IT pro to the educators spilled the truth.
"That game was awesome," he said, which surprised the staff. Turns out, the blue team had recruited the IT pro to snoop—and did so in such a way that didn’t make him think he was assisting any cheating. The blue team managed to keep its debrief stories straight to hide the sneaky move. The game’s runners laughed it off, because the insight proved useful.
"That’s real human behavior," CIA Chief Strategy Officer Rachel Grunspan says. "If you design a game right, you’ll see a lot of complexity organically emerge. That’s what you want."
