Tip Trick Here

12 December 2018

Review: Amazon Echo Input is the easiest way to stream media to speakers

This is the Echo I’ve been waiting for.

Throughout my house, I have Amazon Echo Dots connected to stereo systems. In my office, I have a Dot connected to an Onkyo receiver and amp. In my basement, I have one hooked up to a small bookshelf system. Outside on the deck, a Dot serves audio to a small amp that powers outside speakers. There’s more, but the point is made. The Dot is a great device to add voice services to existing speakers. But with its built-in speaker, I’m paying for features I’m not using.

That’s why Amazon made the Echo Input.

The premise is simple: The Input is a Dot without a speaker. It has a mic, two buttons, and most importantly, a 3.5mm output. This output lets the Input serve media to amps and powered speakers — just like I’m doing so with a Dot.

Since the Input doesn’t have a speaker, it’s much smaller. It’s only a half an inch thick. It’s a tiny thing, and I found it does the job as well as a Dot

Plug it in, set it up, and the Input adds voice services to speaker systems. From Bluetooth speakers to bookshelf speakers, it’s a great way to bring the convenance of Alexa to speakers.

The device is basic. To be clear this is not a Hifi device. To me, that’s okay on most speaker systems since I’m just streaming Spotify and NPR. Hopefully Amazon makes good on producing the Input’s HI-Fi cousin, the $199 Echo Link. This device was announced a few months back and does the same job as the Input, but features TOSLINK and coaxial digital audio outputs for connections to a proper DAC. The $299 Echo Link Amp does the same but features a built in amplifier to directly power a set of speakers. The Input is great for smaller speaker, but the Echo Link should provide a higher fidelity experience — and now that Tidal is available on the Echo, there’s a proper source too.

The Echo Link is said to be released on December 13.

For $35 the Echo Link does its job well. However, during the holidays, the Echo Dot is only $29.99 or less and features the same 3.5mm output. Unless size is a concern, I would recommend buying the Dot while it’s on sale just in case you need the speaker at a later time.

Read Full Article

Why Does Everyone Hate YouTube Rewind 2018?

Every year, YouTube compiles a video designed to represent the past 12 months on YouTube. It’s called YouTube Rewind, and it traditionally features famous YouTubers, video clips, music, and cultural trends that defined YouTube during that year.

2018 is no exception. However, YouTube Rewind 2018 is currently the second most-disliked YouTube video of all time. And it’s expected to sail past Justin Bieber’s Baby within the next few days. So, why exactly is YouTube Rewind 2018 universally hated?

The YouTube Rewind 2018 Video

The first time I watched YouTube Rewind I failed to recognize most of the people in it. But I quickly realized I’m not the generation YouTube is targeting with this. However, the number of dislikes and negative comments suggest YouTube missed the mark here.

At the time of writing, YouTube Rewind 2018 has racked up 8.6 million dislikes over the course of four days. Justin Bieber’s Baby has racked up 9.7 million dislikes in the eight years since it was uploaded to YouTube in 2010. YouTube Rewind is crushing it.

Why All the Hate for YouTube?

There are several reasons for the hate, but the biggest is the disconnect between what YouTube thinks represents YouTube in 2018, and what YouTubers and their audiences think represents YouTube in 2018.

YouTube views 2018 as being about Will Smith, Fortnite, the Royal Wedding, and other mainstream fare. There was no mention of Logan Paul and KSI boxing, PewDiePie battling T-Series for the most subscribers, and Shane Dawson experimenting with the format.

PewDiePie’s non-appearance, while justified from a business perspective, has fuelled much of the hate. Like him or loathe him, the YouTube star boasts 76 million subscribers, which makes him a force to be reckoned with. Unless you’re YouTube.

What YouTube Aspires To Be…

YouTube Rewind 2018 isn’t so much a representation of what happened on YouTube in 2018 as what YouTube wants to be in 2019. It’s a happy-go-lucky place where fresh-faced YouTubers mix with actual celebrities, and where brands spend money on advertising.

This disconnect between what YouTube aspires to be, and what YouTubers and their fans actually do on the platform is set to be a running battle for years to come. Which should at least make reading YouTube comments more fun than usual.

Read the full article: Why Does Everyone Hate YouTube Rewind 2018?

Read Full Article

You Can Now Send Voice Messages on Instagram

Instagram now lets you send voice messages to other users. Given that voice messaging is a thing on virtually every other app of Instagram’s size and stature, this has been a long time coming. However, it also moves Instagram further away from its origins.

How to Send a Voice Message on Instagram

To send a voice message on Instagram, just open an existing conversation or start a new conversation. Then, press and hold the microphone icon, and record your message. When you’ve finished recording, your voice message will appear in the chat as a waveform.

Voice messages can be up to one minute long, and you can send them in both private conversations and group chats. You’ll have to be very sure the message is appropriate for everyone in a group before sending it though, or you may regret recording it.

Starting today, you can send voice messages in Direct. Talk the way you want to be heard, whether by whispering what you’re up to or shouting a compliment. pic.twitter.com/3rkdQneNXO

— Instagram (@instagram) December 10, 2018

Be warned that a voice message will be sent automatically when you release the microphone button. So, if you change your mind while recording it, or aren’t entirely happy with what you’ve said, be sure to slide your finger to the trash can instead.

This feature could be useful for people who prefer to message rather than speak live, who don’t have their hands free to type, or who don’t have time to get into a full-blown conversation. But it will probably be mostly used for fun and frivolous reasons.

Every App Now Does the Exact Same Thing

It has taken Instagram a long time to add voice messaging. Facebook, Instagram’s parent company, added voice messaging to Messenger several years ago. But then unlike Messenger, WhatsApp, et al, Instagram was never intended to be about messaging.

All of these popular apps seem to be copying off each other to the point that it’s difficult to tell them apart. The only discernible difference is their respective attitudes to privacy and security, which led us to suggest Telegram is the only messaging app you need.

Read the full article: You Can Now Send Voice Messages on Instagram

Read Full Article

Podcast industry aims to better track listeners through new analytics tech called RAD

Internet users are already being tracked to death, with ads that follow us around, search histories that are collected and stored, emails that report back to senders when they’ve been read, websites that know where you scrolled and what you clicked, and much more. So naturally, the growing podcast industry wanted to find a way to collect more data of its own, too.

Yes, that’s right. Podcasts will now track detailed user behavior, too.

Today, NPR announced RAD, a new, open sourced podcast analytics technology that was developed in partnership with nearly 30 companies from the podcasting industry. The technology aims to help publishers collect more comprehensive and standardized listening metrics from across platforms.

Specifically, the technology gives publishers – and therefore their advertisers, as well – access to a wide range of listener metrics including downloads, starts and stops, completed ad or credit listens, partial ad or credit listens, ad or credit skips and content quartiles, the RAD website explains.

However, the technology stops short of offering detailed user profiles, and cannot be used to re-target or track listeners, the site notes. It’s still anonymized, aggregated statistics.

It’s worth pointing out that RAD is not the first time podcasters have been able to track engagement. Major platforms, including Apple’s Podcast Analytics, today offer granular and anonymized data, including listens.

But NPR says that data requires “a great deal of manual analysis” as the stats aren’t standardized nor as complete as they could be. RAD is an attempt to change that, by offering a tracking mechanism everyone can use.

Already, RAD has a lot of support. In addition to being integrated into NPR’s own NPR One app, it has commitments from several others who will introduce the technology into their own products in 2019, including Acast, AdsWizz, ART19, Awesound, Blubrry Podcasting, Panoply, Omny Studio, Podtrac, PRI/PRX, RadioPublic, Triton Digital, and WideOrbit.

Other companies that supported RAD and participated in its development include Cadence13, Edison Research, ESPN, Google, iHeartMedia, Libsyn, The New York Times, New York Public Radio, and Wondery.

NPR says the NPR One app on Android supports RAD as of now, and its iOS app will do the same in 2019.

“Over the course of the past year, we have been refining these concepts and the technology in collaboration with some of the smartest people in podcasting from around the world,” said Joel Sucherman, Vice President, New Platform Partnerships at NPR, in an announcement. “We needed to take painstaking care to prove out our commitment to the privacy of listeners, while providing a standard that the industry could rally around in our collective efforts to continue to evolve the podcasting space,” he said.

To use RAD technology, publishers will mark within their audio files certain points – like quartiles or some time markers, interview spots, sponsorship messages or ads – with RAD tags and indicate an analytics URL. A mobile app is configured to read the RAD tags and then, when listeners hit that spot in the file, that information is sent to the URL in an anonymized format.

The end result is that podcasters know just what parts of the audio file their listeners heard, and is able to track this at scale across platforms. (RAD is offering both Android and iOS SDKs.)

While there’s value in podcast data that goes beyond the download, not all are sold on technology.

Most notably, the developer behind the popular iOS podcast player app Overcast, Marco Arment, today publicly stated his app will not support any listener-tracking specs.

Yes. I understand why huge podcast companies want more listener data, but there are zero advantages for listeners or app-makers.

I won’t be supporting any listener-behavior tracking specs in Overcast. Podcasters get enough data from your IP address when you download episodes. https://t.co/mplhnrmCsc

— Marco Arment (@marcoarment) December 11, 2018

“I understand why huge podcast companies want more listener data, but there are zero advantages for listeners or app-makers,” Arment wrote in a tweet. “Podcasters get enough data from your IP address when you download episodes,” he said.

The developer also pointed out this sort of data collection required more work on the podcasters’ part and could become a GDPR liability, as well.

In addition to NPR’s use of RAD today, Podtrac has also now launched a beta program to show RAD data, which is open to interested publishers.

Read Full Article

Bumble bees bearing high-tech backpacks act as a living data collection platform

There’s lots of research going into tiny drones, but one of the many hard parts is keeping them in the air for any real amount of time. Why not hitch a ride on something that already flies all day? That’s the idea behind this project that equips bumble bees with sensor-filled backpacks that charge wirelessly and collect data on the fields they visit.

A hive full of these cyber-bees could help monitor the health of a field by checking temperature and humidity, as well as watching for signs of rot or distress in the crops. A lot of this is done manually now, and of course drones are being set to work doing it, but if the bees are already there, why not get them to help out?

The “Living IoT” backpack, a tiny wafer loaded with electronics and a small battery, was designed by University of Washington engineers led by Shyam Gollakotta. He’s quick to note that although the research does to a certain extent take advantage of these clumsy, fuzzy creatures, they were careful to “follow best methods for care and handling.”

Part of that is minimizing the mass of the pack; other experiments have put RFID antennas and such on the backs of bees and other insects, but this is much more sophisticated.

[gallery ids="1757459,1757477,1757479,1757480,1757481,1757482"]

The chip has sensors and an integrated battery that lets it run for seven hours straight, yet weighs just 102 milligrams. A full-grown bumblebee, for comparison, could weigh anywhere from two to five times that.

They’re strong fliers, if not exact ones, and can carry three quarters of their body weight in pollen and nectar when returning to the hive. So the backpack, while far from unnoticeable, is still well within their capabilities.

“We showed for the first time that it’s possible to actually do all this computation and sensing using insects in lieu of drones,” explained Gollakotta in a UW news release. “We decided to use bumblebees because they’re large enough to carry a tiny battery that can power our system, and they return to a hive every night where we could wirelessly recharge the batteries.”

The backpacks can track location passively by monitoring the varying strengths of signals from nearby antennas, up to a range of about 80 meters. The data they collect is transferred while they’re in the hive via an energy-efficient backscatter method that Gollakotta has used in other projects.

Technique to beam HD video with 99 percent less power could sharpen the eyes of smart homes

The applications are many and various, though obviously limited to what can be observed while the bees go about their normal business. It could even help keep the bees themselves healthy.

“It would be interesting to see if the bees prefer one region of the farm and visit other areas less often,” said co-author Sawyer Fuller. “Alternatively, if you want to know what’s happening in a particular area, you could also program the backpack to say: ‘Hey bees, if you visit this location, take a temperature reading.’ ”

It is of course just in prototype form right now, but one can easily imagine the tech being deployed by farmers in the near future, or perhaps in a more sinister way by three-letter agencies wanting to put a bee on the wall near important conversations. The team plans to present their work (PDF) at the ACM MobiCom conference next year.

Read Full Article

Grasp2Vec: Learning Object Representations from Self-Supervised Grasping

Posted by Eric Jang, Software Engineer, Robotics at Google and Coline Devin, Berkeley PhD Student and former Research Intern

From a remarkably young age, people are capable of recognizing their favorite objects and picking them up, despite never being explicitly taught how to do so. According to cognitive developmental research, the ability to interact with objects in the world plays a crucial role in the emergence of object perception and manipulation capabilities, such as targeted grasping. By interacting with the world around them, people are able to learn with self-supervision: we know what actions we took, and we learn from the outcome. In robotics, this type of self-supervised learning is actively researched because it enables robotic systems to learn without the need for large amounts of training data or manual supervision.

Inspired by the concept of object permanence, we propose Grasp2Vec, a simple yet highly effective algorithm for acquiring object representations. Grasp2Vec is based on the intuition that an attempt to pick up anything provides several pieces of information — if a robot grasps an object and holds it up, the object had to be in the scene before the grasp. Furthermore, the robot knows that the object it grasped is currently in its gripper, and therefore has been removed from the scene. By using this form of self supervision, the robot can learn to recognize the object by the visual change in the scene after the grasp.

Building on our prior collaboration with X Robotics, where a series of robots learn in parallel to grasp household objects using only monocular camera inputs, we use a robotic arm to grasp objects “unintentionally”, and that experience enables the learning of a rich representation of objects. These representations can then be used to acquire “intentional grasping” capabilities, where the robot arm can then pick up user-commanded objects.

Constructing a Perceptual Reward Function
In the framework of reinforcement learning (RL), task success is measured via a “reward function”. By maximizing that reward, robots can teach themselves diverse grasping skills from scratch. Engineering a reward function is easy when success can be measured by simple sensor measurements. A simple example of this is a button that supplies rewards directly to a robot when it is pushed.

However, engineering a reward function is much more difficult when our success criteria depends on perceptual understanding of the task at hand. Consider the task of instance grasping, where a robot is presented a picture of a desired object being held in the gripper. After the robot attempts to grasp that object, it inspects the contents of the gripper. The reward function for this task comes down to answering the question of object recognition: Do these objects match?

On the left, the gripper is holding the brush and there are some objects (yellow cup, blue plastic block) in the background. On the right, the gripper is holding the yellow cup and the brush is in the background. If the left image was the desired outcome, a good reward function should “understand” that the two images above correspond to different objects.

In order to solve this recognition problem, we need a perception system that extracts meaningful object concepts from unstructured image data (without any human annotations), learning the visual perception of objects in an unsupervised fashion. At their core, unsupervised learning algorithms work because they make structural assumptions about data. It is common to assume that images can be compressed into a low-dimensional space, and that frames in a video can be predicted from previous frames. However, without further assumptions on the content of the data, these are usually insufficient for learning disentangled object representations.

What if we used a robot to physically disentangle objects from each other during data collection? The field of robotics presents an exciting opportunity for representation learning because robots can manipulate objects, thus providing the factors of variation needed in data. Our method relies on the insight that grasping an object removes it from the scene. This yields 1) an image of the scene before grasping, 2) an image of the scene after grasping and 3) an isolated view of the grasped object itself.

Left: Objects before the grasp. Center: Objects after the grasp. Right: The Grasped object.

If we then consider an embedding function that extracts “the set of objects” from images, it should preserve the following subtractive relation:

objects_before_grasp - objects_after_grasp = grasped_object

We implement this equality relation using a fully convolutional architecture and a simple metric learning algorithm. At training time, the architecture shown below embeds the pre-grasp images and post-grasp images into a dense spatial feature map. The maps are mean-pooled into vectors and the difference between the “before grasp” and “after grasp” vectors represents a set of objects. This vector and the corresponding vector representation of the grasped object are pushed to equivalence via the N-Pairs objective.

Add caption

Once trained, two useful properties emerge naturally from our model.

1. Object Similarity
The first property is that a cosine distance between vector embeddings allows us to compare objects and determine whether they are identical. This can be used to implement reward functions for reinforcement learning, and allow robots to learn instance grasping without human-provided labels.

2. Localizing Target Objects
The second property is that we can combine scene spatial maps and object embeddings to localize a “query object” in image space. By taking the element-wise product of spatial feature maps and the vector corresponding to the query object, we can find all the pixels in the spatial map that “match” the query object.

Using Grasp2Vec embeddings to localize objects in a scene. The image on the top left shows the objects in the bin. On the bottom left is the query object we wish to grasp. By taking the dot product of the query object vector with the spatial features of the scene image, we get a per-pixel “activation map” (top right image) of how similar that region of the image is to the query. This response map can be used to approach the object for grasping.

Our method also works when there are multiple objects that match the query object, or even if the query consists of multiple objects (the average of two vectors). For example, here is a scenario where it detects multiple orange blocks in a scene.

The resulting “heatmap” can be used to plan the robot approach to the target object(s). We combine Grasp2Vec’s localization and instance recognition capabilities with our “grasp anything” policies to obtain a success rate of 80% on objects seen during data collection and 59% on novel objects the robot hasn’t encountered before.

Conclusion
In our paper, we show how robotic grasping skills can generate the data used for learning object-centric representations. We then can use representation learning to “bootstrap” more complex skills like instance grasping, all while retaining the self-supervised learning properties of our autonomous grasping system.

Besides our own work, a number of recent papers have also studied how self-supervised interaction can be used to acquire representations, by grasping, pushing, and otherwise manipulating objects in the environment. Going forward, we are excited not only for what machine learning can bring to robotics by way of better perception and control, but also what robotics can bring to machine learning in new paradigms of self-supervision.

Acknowledgements
This research was conducted by Eric Jang, Coline Devin, Vincent Vanhoucke, and Sergey Levine. We’d like to thank Adrian Li, Alex Irpan, Anthony Brohan, Chelsea Finn, Christian Howard, Corey Lynch, Dmitry Kalashnikov, Ian Wilkes, Ivonne Fajardo, Julian Ibarz, Ming Zhao, Peter Pastor, Pierre Sermanet, Stephen James, Tsung-Yi Lin, Yunfei Bai, and many others at Google, X, and the broader robotics community who contributed to improving this work.

With Congress focused on political bias, Google’s CEO gets off easy

Google may have evaded the first few rounds of grill-a-tech-exec, but its CEO Sundar Pichai wound up in the hotseat on the Hill after all. Pichai appeared today, weathering a three and a half hour-long hearing solo with no competitors to offload congressional ire onto.

To his credit, Pichai performed very well. Instead of adopting Facebook’s flavor of over-rehearsed confidence and robot-like answers, Pichai came across as human, very competent and fairly sympathetic. That demeanor held, even as Pichai dodged here and pushed a handful of misrepresentations and mistruths there.

In a normal hearing, Pichai’s task would be tough, but this wasn’t a normal hearing. Pichai wasn’t called to Capitol Hill to testify on Google’s role in Russian disinformation, its controversial plan to re-enter China or recent news that Google+ exposed private data. Rather, he was summoned to Washington as the last gasp of a Republican led House Judiciary Committee hell bent on performing its own preoccupation with perceived political bias against conservatives.

The hearing, titled “Transparency & Accountability: Examining Google and its Data Collection, Use, and Filtering Practices” hewed closely to that aim in its mission to “examine potential bias” supposedly perpetrated by the search giant to the detriment of right-leaning voices. Over more than three hours of testimony, no proof surfaced that Silicon Valley’s left-leaning ideals had polluted Google’s search algorithms, which shouldn’t come as a surprise. No substantial evidence has surfaced indicating that conservatives are unfairly treated by Google to date, though that doesn’t keep many of the committee’s Republicans from treating those accusations as fact.

Unfortunately for them, perhaps, none of this flustered Google’s chief executive. Pichai patiently explained that Google was aware of many of the accusations from the right but that its own investigations had turned up nothing of substance. He gently disputed the questionable methodologies behind research purporting to show evidence of Google’s bias.

“I lead this company without political bias and work to ensure that our products continue to operate that way,” Pichai said. “To do otherwise would go against our core principles and our business interests.”

The partisan-fueled line of questioning led to a weird role dynamic in which Democrats on the committee pushed back on Republican talking points, defending Google instead of laying into the company on the myriad relevant issues of substance. For a political body ostensibly committed to holding power to account — and wielding its regulatory power in doing so — it was disappointing. Pichai mostly slipped through their hands.

Some of the committee’s members even acknowledged more substantial issues — anti-competitive behavior, Google’s position on China — before skipping back to the Republican-led committee’s agreed upon throughline. Still, there were a few moments of clarity.

When asked point blank by Rhode Island Rep. David Cicilline if Pichai would “rule out launching a tool for surveillance and censorship in China” during his tenure as CEO, Pichai dodged noticeably. Washington Rep. Pramila Jayapal cornered Pichai into a follow-up commitment on ending its forced arbitration practices for issues extending beyond sexual harassment. And a number of members of the committee pressed Pichai on Google’s invasive Android location-sharing practices in light of a recent New York Times report raising the issue, so it’s good to see Congress uncomfortable with that knowledge even if some members of this particular committee still don’t know the basics of how tech works.

In spite of these bright spots, Pichai got off easy. He managed to make it through more than three hours without any questioning about Google’s interest in providing technology for military applications. And no one quite pinned him down to any kind of commitment around Android location sharing concerns or Google’s inscrutable privacy policies, though that doesn’t come as a surprise. Under Democratic leadership next term, the committee hinted that it would try again.

“I look forward next year to working with you on some of the very serious questions we face,” Rep. Zoe Lofgren said. “It’s pretty obvious bias against conservative voices is not one of them.”

Tech executives, so insulated within massive, layered companies and cocooned in wealth, are rarely if ever accountable, on a personal level, to much of anyone. The opportunity to press them with questions in real time is almost never afforded to members of the press. Unfortunately, especially right now, there are many, many legitimate things to hold Google to account about that were not the premise of today’s hearing.

It’s all the more disappointing then that Pichai’s time on the stand was mostly squandered.

Google CEO won’t rule out relaunching in China

Read Full Article

Lift Aircraft’s Hexa may be your first multirotor drone ride

We were promised jetpacks, but let’s be honest, they’re just plain unsafe. So a nice drone ride is probably all we should reasonably expect. Lift Aircraft is the latest to make a play for the passenger multirotor market, theoretical as it is, and its craft is a sleek little thing with some interesting design choices to make it suitable for laypeople to “pilot.”

The Austin-based company just took the wraps off the Hexa, the 18-rotor craft it intends to make available for short recreational flights. It just flew for the first time last month, and could be taking passengers aloft as early as next year.

The Hexa is considerably more lightweight than the aircraft that seemed to be getting announced every month or two earlier this year. Lift’s focus isn’t on transport, which is a phenomenally complicated problem both in terms or regulation and engineering. Instead, it wants to simply make the experience of flying in a giant drone available for thrill-seekers with a bit of pocket money.

This reduced scope means the craft can get away with being just 432 pounds and capable of 10-15 minutes of sustained flight with a single passenger. Compared with Lilium’s VTOL engines or Volocopter’s 36-foot wingspan, this thing looks like a toy. And that’s essentially what it is, for now. But there’s something to be said for proving your design in a comparatively easily accessed market and moving up, rather than trying to invent an air taxi business from scratch.

“Multi-seat eVTOL air taxis, especially those that are designed to transition to wing-borne flight, are probably 10 years away and will require new regulations and significant advances in battery technology to be practical and safe. We didn’t want to wait for major technology or regulatory breakthroughs to start flying,” said Chasen in a news release. “We’ll be flying years before anyone else.”

The Hexa is flown with a single joystick and an iPad; direct movements and attitude control are done with the former, while destination-based movement, takeoff and landing take place on the latter. This way people can go from walking in the front door to flying one of these things — or rather riding in one and suggesting some directions to go — in an hour or so.

It’s small enough that it doesn’t even count as a “real” aircraft; it’s a “powered ultralight,” which is a plus and a minus: no pilot’s license necessary, but you can’t go past a few hundred feet of altitude or fly over populated areas. No doubt there’s still a good deal of fun you can have flying around a sort of drone theme park, though. The whole area will have been 3D mapped prior to flight, of course.

Lifting the Hexa are 18 rotors, each of which is powered by its own battery, which spreads the risk out considerably and makes it simple to swap them out. As far as safety is concerned, it can run with up to 6 engines down, has pontoons in case of a water landing, and an emergency parachute should the unthinkable happen.

The team is looking to roll out its drone-riding experience soon, but it has yet to select its first city. Finding a good location, checking with the community, getting the proper permits — not simple. CEO Matt Chasen told New Atlas the craft is “not very loud, but they’re also not whisper-quiet, either.” I’m thinking “not very loud” is in comparison to jets — every drone I’ve ever come across, from palm-sized to cargo-bearing, has made an incredible racket and if someone wanted to start a drone preserve next door I’d fight it tooth and nail. (Apparently Seattle is high on the list, too, so this may come to pass.)

In a sense, engineering a working autonomous multirotor aircraft was the easy part of building this business. Chasen told GeekWire that the company has raised a “typical-size seed round,” and is preparing for a Series A — probably once it has a launch city in its sights.

We’ll likely hear more at SXSW in March, where the Hexa will likely fly its first passengers.

Read Full Article

Google CEO won’t rule out relaunching in China

Members of the House Judiciary committee have today closely questioned Google’s CEO about the company’s intentions in China, following reports this summer it’s planning a controversial return to the market despite local censorship of Internet services.

Mountain View pulled its search engine out of China back in 2010, under pressure over censored search results. It also cited cyber attacks originating from the country as factoring into its decision to leave.

So Congress wanted to know what exactly had changed that Google is considering reversing that exit, and how doing so wouldn’t contradict its own “core values”.

A none-too-comfortable Pichai stuck closely to a qualified line on the topic, telling Congress Google has no plans “right now” to launch in China, thereby leaving the door open for a launch at a future date.

At the same time he pointed to Google’s “mission”, of making information digitally accessible, to justify what he couched as an exploratory, internal effort at this stage — saying the company mission underpins its ongoing interest in the market.

He made no mention of the 800M+ Chinese Internet users that Google could potentially add to its business if it returned to the market as factoring into his calculations.

“Right now there are no plans for us to launch a search product in China,” Pichai told the committee. “We are, in general, always looking to see how best — it’s part of our core mission and our principles — to try hard to provide users with information.

“We always have evidence, based on every country we’ve operated in, us reaching out and giving users more information has a very positive impact. And we feel that calling but right now there are no plans to launch in China. To the extent that we ever approach a position like that I will be fully transparent, including with policymakers here. And engage and consult widely.”

A little earlier in the session the committee was briefly interrupted by a person trying to enter the room holding a poster depicting the Google logo superimposed on the Chinese flag.

The chair asked for the individual to be removed and the door to be shut.

Hearing is interrupted by someone holding this sign pic.twitter.com/5Pyb72KP87

— Daisuke Wakabayashi (@daiwaka) December 11, 2018

Asked directly whether he would “avoid launching a tool of censorship and surveillance in China while you are CEO of Google”, Pichai trod carefully in his response, avoiding making any such categorical commitment — but saying he would be “very thoughtful” about any relaunch.

“Congressman I commit to engaging,” he said, beginning his answer. “One of the things which is important to us as a company, we have a stated mission of providing users with information, and so we always — we think it’s in our duty to explore possibilities. To give users access to information… I have that commitment but as I said earlier on this we’ll be very thoughtful and we will engage widely as we make progress.”

He was also asked directly whether there are any current discussions with any member of the Chinese government on launching the app. “This effort currently is an internal effort,” he replied to that, reiterating again that he would be “happy to consult/be transparent should we take steps towards launching a product in China”.

Asked who is leading the China project, Pichai also kept things vague — saying it’s being undertaken by Google’s “search teams”, before adding: “But these are distributed efforts. It’s a limited effort internally currently.”

One of the policymakers who questioned Pichai about Google’s intentions in China, congressman David Cicilline, suggested returning to the market would be “completely inconsistent” with Google’s recently announced AI principles.

But ending his five minutes of allotted question time with the Google CEO, Cicilline said concern about technologists working with anti-democratic regimes “goes beyond Google — and frankly beyond China”.

“At a moment of rising authoritarianism around the world when more leaders are using surveillance, censorship and repression against their own people we’re in a moment that we must reassert American moral leadership,” he said, asking also for an open letter from a coalition of human and civil rights organizations, opposing the launch of a censored Google search engine for the Chinese market, to be submitted to the committee for the record.

Read Full Article

Google CEO admits company must better address the spread of conspiracy theories on YouTube

Google CEO Sundar Pichai admitted today that YouTube needs to do better in dealing with conspiracy content on its site that can lead to real world violence. During his testimony on Tuesday before the House Judiciary Committee, the exec was questioned on how YouTube handles extremist content which promote conspiracy theories like Pizzagate and, more recently, a Hillary Clinton-focused conspiracy theory dubbed Frazzledrip.

According to an article in Monday’s The Washington Post, Frazzledrip is a variation on Pizzagate that began spreading on YouTube this spring.

In a bizarre series of questions, Rep. Jamie Raskin (D-MD) asked Pichai if he knew what Frazzledrip was.

Pichai replied that he was “not aware of the specifics about it.”

Raskin went on to explain that the recommendation engine on YouTube has been suggesting videos that claim politicians, celebrities and other leading figures were “sexually abusing and consuming the remains of children, often in satanic rituals.” He said these new conspiracist claims were echoing the discredited Pizzagate conspiracy, which two years ago led to a man firing shots into a Washington D.C. pizzeria, in search of the children he believed were held as sex slaves by Democratic Party leaders.

He also explained the new Frazzledrip theory in more detail, which he read about in The Washington Post’s report about the still rampant hateful conspiracies being hosted by YouTube. This newer conspiracy claims that Hillary Clinton and longtime aide Huma Abedin sexually assaulted a girl and drank her blood.

The Post said some of the video clips were removed after first appearing in April and had been debunked, but its review of the matter found dozens of videos where the claims were still being discussed. Combined, these videos had been viewed millions of times over the past eight months. In addition, the investigation found that YouTube’s search box would highlight these videos when people typed in terms like “HRC video” or “Frazzle.”

YouTube’s policy doesn’t prevent people from uploading falsehoods, The Post’s report noted.

Raskin asked Pichai about this type of extremist propaganda.

“What is your company policy on that? And are you trying to deal with it?,” he questioned.

Pichai admitted, essentially, that YouTube needed to do better.

“We are constantly undertaking efforts to deal with misinformation. We have clearly stated policies and we have made lots of progress in many of the areas where over the past year – so, for example, in areas like terrorism, child safety, and so on,” said Pichai. “We are looking to do more,” he said.

In terms of the Frazzledrip theory, he said it was more of a recent happening.

“But I’m committed to following up on it and making sure we are evaluating these against our policies,” the CEO promised.

The issue with videos like Frazzledrip is that YouTube’s current policies don’t fully encompass how to handle extremist propaganda. Instead, as The Post also said, its policies focus on videos with hateful, graphic and violent content directed at minorities and other protected groups. Meanwhile, it seeks to allow freedom of speech to others who upload content to its site, despite the disinformation they may spread or their potential to lead to violence.

The balance between free speech and content policies is a delicate matter – and an important one, given YouTube’s power to influence dangerous individuals. In addition to the Pizzagate shooter, the mass shooter who killed 11 people at the Pittsburgh synagogue in October had been watching neo-Nazi propaganda on YouTube, the Post’s report pointed out, in another example.

Asked what YouTube was doing about all this, Pichai didn’t offer specifics.

The CEO instead admitted that YouTube struggles with evaluating videos individually because of the volume of content it sees.

“We do get around 400 hours of video every minute. But it’s our responsibility, I think, to make sure YouTube is a platform for freedom of expression, but it’s responsible and contributes positively to society,” Pichai said. He added that its policies allow it take down videos that “insight harm or hatred or violence.” But conspiracy videos don’t always directly insight violence – they just radicalize individuals, who then sometimes act out violently, as a result.

“It’s an area we acknowledge there’s more work to be done, and we’ll definitely continue doing that,” Pichai said. “But I want to acknowledge there is more work to be done. With our growth comes more responsibility. And we are committed to doing better as we invest more in this area,” he said.