×
Google

Google Promises Ethical Principles To Guide Development of Military AI (theverge.com) 154

An anonymous reader quotes a report from The Verge: Google is drawing up a set of guidelines that will steer its involvement in developing AI tools for the military, according to a report from The New York Times. What exactly these guidelines will stipulate isn't clear, but Google says they will include a ban on the use of artificial intelligence in weaponry. The principles are expected to be announced in full in the coming weeks. They are a response to the controversy over the company's decision to develop AI tools for the Pentagon that analyze drone surveillance footage.

Internal emails obtained by the Times show that Google was aware of the upset this news might cause. Chief scientist at Google Cloud, Fei-Fei Li, told colleagues that they should "avoid at ALL COSTS any mention or implication of AI" when announcing the Pentagon contract. "Weaponized AI is probably one of the most sensitized topics of AI -- if not THE most. This is red meat to the media to find all ways to damage Google," said Li. But Google never ended up making the announcement, and it has since been on the back foot defending its decision. The company says the technology it's helping to build for the Pentagon simply "flags images for human review" and is for "non-offensive uses only." The contract is also small by industry standards -- worth just $9 million to Google, according to the Times.

AI

AI Better Than Dermatologists At Detecting Skin Cancer, Study Finds (cbsnews.com) 60

An anonymous reader quotes a report from CBS News: For the first time, new research suggests artificial intelligence may be better than highly-trained humans at detecting skin cancer. A study conducted by an international team of researchers pitted experienced dermatologists against a machine learning system, known as a deep learning convolutional neural network, or CNN, to see which was more effective at detecting malignant melanomas. The results? 'Most dermatologists were outperformed by the CNN,' the researchers wrote in their report, published in the journal Annals of Oncology. Fifty-eight dermatologists from 17 countries around the world participated in the study. More than half of the doctors were considered expert level with more than five years' experience. Nineteen percent said they had between two to five years' experience, and 29 percent had less than two years' experience. At first look, dermatologists correctly detected an average of 87 percent of melanomas, and accurately identified an average of 73 percent of lesions that were not malignant. Conversely, the CNN correctly detected 95 percent of melanomas. The study has been published in the journal Annals of Oncology.
NES (Games)

Hacker Gets Super NES Games Running On Unmodified NES (arstechnica.com) 43

The latest project from Tom "Tom7" Murphy is an unmodified NES running Super NES games. "Murphy breaks down this wizardry in a pair of detailed videos laying out his tinkering process," reports Ars Technica. "Though the NES hardware itself is untouched, the cartridge running this reverse emulation is a heavily customized circuit board (ordered from China for about $10), with a compact, multi-core Raspberry Pi 3 attached to handle the actual Super NES emulation." From the report: The Pi essentially replaces the PPU portion of the cartridge, connecting to the NES via a custom-coded EEPROM chip that tells the system how to process and display what would normally be an overwhelming stream of graphical data coming from the miniature computer. Only the CIC "copyright" chip from the original cartridge remains unmodified to get around the hardware's lockout chip. Murphy -- you may remember him from previous efforts to teach an AI how to play NES games -- says that the Raspberry Pi actually has too much latency to effectively "stream" tile-by-tile graphical instructions to the NES' cartridge CPU. By the time the Pi manages to "discharge" a set of instruction bits (only 180ns after they were generated), the NES itself has already moved on to the next part of its read-write cycle.

Murphy used a one-cycle delay to compensate for this latency, essentially guessing where the fairly predictable PPU would be writing to next and just sending data to that location ahead of time. That process works pretty well but results in the persistent flickering and graphical noise you see throughout his video demonstrations.

Cloud

Nvidia Debuts Cloud Server Platform To Unify AI and High-Performance Computing (siliconangle.com) 15

Hoping to maintain the high ground in AI and high-performance computing, Nvidia late Tuesday debuted a new computing architecture that it claims will unify both fast-growing areas of the industry. From a report: The announcement of the HGX-2 cloud-server platform, made by Nvidia Chief Executive Jensen Huang at its GPU Technology Conference in Taipei, Taiwan, is aimed at many new applications that combine AI and HPC. "We believe the future requires a unified platform for AI and high-performance computing," Paresh Kharya, product marketing manager for Nvidiaâ(TM)s accelerated-computing group, said during a press call Tuesday.

Others agree. "I think that AI will revolutionize HPC," Karl Freund, a senior analyst at Moor Insights & Strategy, told SiliconANGLE. "I suspect many supercomputing centers will deploy HGX2 as it can add dramatic computational capacity for both HPC and AI." More specifically, the new architecture enables applications involving scientific computing and simulations, such as weather forecasting, as well as both training and running of AI models such as deep learning neural networks, for jobs such as image and speech recognition and navigation for self-driving cars.

The Internet

Mary Meeker's 2018 Internet Trends Report (recode.net) 33

Mary Meeker has published her anticipated internet trends report of 2018. This year, the Kleiner Perkins Caufield & Byers partner released 284 slides in rapid succession, covering everything from smartphone behavior in the U.S. to tech company competition in China. Some takeaways: 1. 2017 was the first year in which smartphone unit shipments didn't grow at all. As more of the world become smartphone owners, growth has been harder and harder to come by. The same goes for internet user growth, which rose 7 percent in 2017, down from 12 percent the year before. With more than half the world online, there are fewer people left to connect.
2. People, however, are still increasing the amount of time they spend online. U.S. adults spent 5.9 hours per day on digital media in 2017, up from 5.6 hours the year before. Some 3.3 of those hours were spent on mobile, which is responsible for overall growth in digital media consumption.
3. Despite the high-profile releases of $1,000 iPhones and Samsung Galaxy Notes, the global average selling price of smartphones is continuing to decline.
4. Mobile payments are becoming easier to complete. China continues to lead the rest of the world in mobile payment adoption, with over 500 million active mobile payment users in 2017.
5. Voice-controlled products like Amazon Echo are taking off. The Echo's installed base in the U.S. grew from 20 million in the third quarter of 2017 to more than 30 million in the fourth quarter.
6. Tech companies are facing a "privacy paradox." They're caught between using data to provide better consumer experiences and violating consumer privacy.
The most popular courses on learning platform Coursera last year were (in descending order): Machine Learning (Stanford), Neural Networks & Deeper Learning (Deeplearning.ai), Learning How to Learn: Powerful Mental Tools to Help You Master Tough Subjects (UC San Diego), Introduction to Mathematical Thinking (Stanford), Bitcoin & Cryptocurrency Technologies (Princeton), Programming for Everybody (University of Michigan), Algorithms, Part I (Princeton), English for Career Development (University of Pennsylvania), Neural Networks / Machine Learning (University of Toronto), and Financial Markets (Yale).
AI

Why Thousands of AI Researchers Are Boycotting the New Nature Journal (theguardian.com) 62

An anonymous reader shares an excerpt from a report via The Guardian, written by Neil Lawrence, the founding editor of the freely available journal Proceedings of Machine Learning Research: Machine learning has demonstrated that an academic field can not only survive, but thrive, without the involvement of commercial publishers. But this has not stopped traditional publishers from entering the market. Our success has caught their attention. Most recently, the publishing conglomerate Springer Nature announced a new journal targeted at the community called Nature Machine Intelligence. The publisher now has 53 journals that bear the Nature name. Should we be concerned? What would drive authors and readers towards a for-profit subscription journal when we already have an open model for sharing our ideas? Academic publishers have one card left to play: their brand. The diversity and quantity of academic research means that it is difficult for a researcher in one field to rate the work in another. Sometimes a journal's brand is used as a proxy for quality. When academics look for promotion, having papers in a "brand-name journal" can be a big help. Nature is the Rolex of academic publishing. But in contrast to Rolex, whose staff are responsible for the innovation in its watches, Nature relies on academics to provide its content. We are the watchmakers, they are merely the distributors.

Many in our research community see the Nature brand as a poor proxy for academic quality. We resist the intrusion of for-profit publishing into our field. As a result, at the time of writing, more than 3,000 researchers, including many leading names in the field from both industry and academia, have signed a statement refusing to submit, review or edit for this new journal. We see no role for closed access or author-fee publication in the future of machine-learning research. We believe the adoption of this new journal as an outlet of record for the machine-learning community would be a retrograde step.

AI

Microsoft Developing a Tool To Help Engineers Catch Bias in Algorithms (venturebeat.com) 239

Microsoft is developing a tool that can detect bias in artificial intelligence algorithms with the goal of helping businesses use AI without running the risk of discriminating against certain people. From a report: Rich Caruana, a senior researcher on the bias-detection tool at Microsoft, described it as a "dashboard" that engineers can apply to trained AI models. "Things like transparency, intelligibility, and explanation are new enough to the field that few of us have sufficient experience to know everything we should look for and all the ways that bias might lurk in our models," he told MIT Technology Review. Bias in algorithms is an issue increasingly coming to the fore. At the Re-Work Deep Learning Summit in Boston this week, Gabriele Fariello, a Harvard instructor in machine learning and chief information officer at the University of Rhode Island, said that there are "significant ... problems" in the AI field's treatment of ethics and bias today. "There are real decisions being made in health care, in the judicial system, and elsewhere that affect your life directly," he said.
Privacy

Amazon Explains Why Alexa Recorded And Emailed A Private Conversation (mercurynews.com) 163

Amazon has issued the following statement about why their Alexa device recorded a woman's private conversation and then emailed it to one of her friends: Echo woke up due to a word in background conversation sounding like "Alexa." Then, the subsequent conversation was heard as a "send message" request. At which point, Alexa said out loud "To whom?" At which point, the background conversation was interpreted as a name in the customers contact list. Alexa then asked out loud, "[contact name], right?" Alexa then interpreted background conversation as "right." As unlikely as this string of events is, we are evaluating options to make this case even less likely.
This apparently didn't satisfy the woman whose conversation was recorded, according to the Mercury News:
Now her family has unplugged all the devices, and although Amazon offered to "de-provision" the devices of their communications features so they could keep using them to control their home, Danielle and her family reportedly want a refund instead.

When reached Friday, an Amazon spokeswoman would not comment about whether the company will issue a refund.

Other smart home speakers carry similar privacy risks. Last year, for example, Google had to release a patch for its Home Mini speakers after some of them were found to be recording everything.

Robotics

AI-Enhanced Weed-Killing Robots Frighten Pesticide Industry (reuters.com) 176

Rick Schumann writes: A Swiss company called ecoRobotix is betting the agricultural industry will be willing to welcome their solar-powered weed-killing autonomous robot, in an effort to reduce the use of herbicides by up to a factor of 20 and perhaps even eliminate the need for herbicide-resistant GMO crops entirely.

The 'see-and-spray' robot goes from plant to plant, visually differentiating the actual crops and weeds, and squirting the weeds selectively and precisely with weed killer, as opposed to the current technique of using large quantities of weed killer like Monsantos' Roundup to spray entire crops.

Weeds are already becoming resistant to such glyphosate-based herbicides after "more than 20 years of near-ubiquitous use," reports Reuters. (The head of one pesticide company's science division concedes that "That was probably a once-in-a-lifetime product.") But AI-based precision spraying "could mean established herbicides whose effect has worn off on some weeds could be used successfully in more potent, targeted doses."

Meanwhile, another Silicon Valley startup has built a machine using on-board cameras to distinguish weeds from crops -- and was recently acquired by the John Deere tractor company. Reuters calls these companies the "new breed of AI weeders that investors say could disrupt the $100 billion pesticides and seeds industry."

The original submission asks: Should we welcome our weed-killing robotic overlords?
Canada

How Canada Ended Up As An AI Superpower 64

pacopico writes: Neural nets and deep learning are all the rage these days, but their rise was anything but sudden. A handful of determined researchers scattered around the globe spent decades developing neural nets while most of their peers thought they were mad. An unusually large number of these academics -- including Geoff Hinton, Yoshua Bengio, Yann LeCun and Richard Sutton -- were working at universities in Canada. Bloomberg Businessweek has put together an oral history of how Canada brought them all together, why they kept chasing neural nets in the face of so much failure, and why their ideas suddenly started to take off. There's also a documentary featuring the researchers and Prime Minster Justin Trudeau that tells more of the story and looks at where AI technology is heading -- both the good and the bad. Overall, it's a solid primer for people wanting to know about AI and the weird story of where the technology came from, but might be kinda basic for hardcore AI folks.
AI

Eric Schmidt Says Elon Musk Is 'Exactly Wrong' About AI (techcrunch.com) 143

At the VivaTech conference in Paris, Alphabet CEO Eric Schmidt was asked about Elon Musk's warnings about AI. He responded by saying: "I think Elon is exactly wrong. He doesn't understand the benefits that this technology will provide to making every human being smarter. The fact of the matter is that AI and machine learning are so fundamentally good for humanity." TechCrunch reports: He acknowledged that there are risks around how the technology might be misused, but he said they're outweighed by the benefits: "The example I would offer is, would you not invent the telephone because of the possible misuse of the telephone by evil people? No, you would build the telephone and you would try to find a way to police the misuse of the telephone."

After wryly observing that Schmidt had just given the journalists in the audience their headlines, interviewer (and former Publicis CEO) Maurice Levy asked how AI and public policy can be developed so that some groups aren't "left behind." Schmidt replied that government should fund research and education around these technologies. "As [these new solutions] emerge, they will benefit all of us, and I mean the people who think they're in trouble, too," he said. He added that data shows "workers who work in jobs where the job gets more complicated get higher wages -- if they can be helped to do it." Schmidt also argued that contrary to concerns that automation and technology will eliminate jobs, "The embracement of AI is net positive for jobs." In fact, he said there will be "too many jobs" -- because as society ages, there won't be enough people working and paying taxes to fund crucial services. So AI is "the best way to make them more productive, to make them smarter, more scalable, quicker and so forth."

Privacy

Zimbabwe is Introducing a Mass Facial Recognition Project With Chinese AI Firm CloudWalk (qz.com) 33

An anonymous reader shares a report: In March, the Zimbabwean government signed a strategic partnership with the Gunagzhou-based startup CloudWalk Technology to begin a large-scale facial recognition program throughout the country. The agreement, backed by the Chinese government's Belt and Road initiative, will see the technology primarily used in security and law enforcement and will likely be expanded to other public programs.

[...] Zimbabwe may be giving away valuable data as Chinese AI technologists stand to benefit from access to a database of millions of Zimbabwean faces Harare will share with CloudWalk. [...] CloudWalk has already recalibrated its existing technology through three-dimensional light technology in order to recognize darker skin tones. In order to recognize other characteristics that may differ from China's population, CloudWalk is also developing a system that recognizes different hairstyles and body shapes, another representative explained to the Global Times.

Privacy

Woman Says Alexa Device Recorded Her Private Conversation and Sent It To Random Contact; Amazon Confirms the Incident (kiro7.com) 272

Gary Horcher, reporting for KIRO7: A Portland family contacted Amazon to investigate after they say a private conversation in their home was recorded by Amazon's Alexa -- the voice-controlled smart speaker -- and that the recorded audio was sent to the phone of a random person in Seattle, who was in the family's contact list. "My husband and I would joke and say I'd bet these devices are listening to what we're saying," said Danielle, who did not want us to use her last name. Every room in her family home was wired with the Amazon devices to control her home's heat, lights and security system. But Danielle said two weeks ago their love for Alexa changed with an alarming phone call. "The person on the other line said, 'unplug your Alexa devices right now,'" she said. '"You're being hacked.'" That person was one of her husband's employees, calling from Seattle. "We unplugged all of them and he proceeded to tell us that he had received audio files of recordings from inside our house," she said. "At first, my husband was, like, 'no you didn't!' And the (recipient of the message) said 'You sat there talking about hardwood floors.' And we said, 'oh gosh, you really did hear us.'" Danielle listened to the conversation when it was sent back to her, and she couldn't believe someone 176 miles away heard it too. In a statement, an Amazon spokesperson said, "Amazon takes privacy very seriously. We investigated what happened and determined this was an extremely rare occurrence. We are taking steps to avoid this from happening in the future."

Further reading: Amazon Admits Its AI Alexa is Creepily Laughing at People.
Facebook

Facebook Asks British Users To Submit Their Nudes as Protection Against Revenge Porn (betanews.com) 303

Mark Wilson writes: Following on from a trial in Australia, Facebook is rolling out anti-revenge porn measures to the UK. In order that it can protect British users from failing victim to revenge porn, the social network is asking them to send in naked photos of themselves. The basic premise of the idea is: send us nudes, and we'll stop others from seeing them .
AI

UK Military Fears Robots Learning War From Video Games (bbc.com) 68

Robots that train themselves in battle tactics by playing video games could be used to mount cyber-attacks, the UK military fears. From a report: The warning is in a Ministry of Defence report on artificial intelligence. Researchers in Silicon Valley are using strategy games, such as Starcraft II, to teach systems how to solve complex problems on their own. But artificial intelligence (AI) programs can then "be readily adapted" to wage cyber-warfare, the MoD says. Officials are particularly concerned about the ability of rogue states and terrorists to mount advanced persistent threat attacks, which can disable critical infrastructure and steal sensitive information.
United States

The US Military is Funding an Effort To Catch Deepfakes and Other AI Trickery (technologyreview.com) 70

The Department of Defense is funding a project that will try to determine whether the increasingly real-looking fake video and audio generated by artificial intelligence might soon be impossible to distinguish from the real thing -- even for another AI system. From a report: This summer, under a project funded by the Defense Advanced Research Projects Agency (DARPA), the world's leading digital forensics experts will gather for an AI fakery contest. They will compete to generate the most convincing AI-generated fake video, imagery, and audio -- and they will also try to develop tools that can catch these counterfeits automatically. The contest will include so-called "deepfakes," videos in which one person's face is stitched onto another person's body.

Rather predictably, the technology has already been used to generate a number of counterfeit celebrity porn videos. But the method could also be used to create a clip of a politician saying or doing something outrageous. DARPA's technologists are especially concerned about a relatively new AI technique that could make AI fakery almost impossible to spot automatically. Using what are known as generative adversarial networks, or GANs, it is possible to generate stunningly realistic artificial imagery.

AI

Microsoft Also Has An AI Bot That Makes Phone Calls To Humans (theverge.com) 61

An anonymous reader quotes a report from The Verge: At an AI event in London today, Microsoft CEO Satya Nadella showed off the company's Xiaoice (pronounced "SHAO-ICE") social chat bot. Microsoft has been testing Xiaoice in China, and Nadella revealed the bot has 500 million "friends" and more than 16 channels for Chinese users to interact with it through WeChat and other popular messaging services. Microsoft has turned Xiaoice, which is Chinese for "little Bing," into a friendly bot that has convinced some of its users that the bot is a friend or a human being. "Xiaoice has her own TV show, it writes poetry, and it does many interesting things," reveals Nadella. "It's a bit of a celebrity."

While most of Xiaoice's interactions have been in text conversations, Microsoft has started allowing the chat bot to call people on their phones. It's not exactly the same as Google Duplex, which uses the Assistant to make calls on your behalf, but instead it holds a phone conversation with you. "One of the things we started doing earlier this year is having full duplex conversations," explains Nadella. "So now Xiaoice can be conversing with you in WeChat and stop and call you. Then you can just talk to it using voice." (The term "full duplex" here refers to a conversation where both participants can speak at the same time; it's not a reference to Google's product, which was named after the same jargon.)

Youtube

Google Launches YouTube Music Service With Creepy AI To Predict Listening Habits (audioholics.com) 87

Audiofan writes: Will the new YouTube Music streaming service provide the soundtrack to your life? Google believes that its ability to harness the power of artificial intelligence will help the new service catch up to its rivals in the music streaming business. Google's latest attempt to compete with Spotify and Apple Music may finally have what it takes if it doesn't creep users out in the process. While the service officially rolls out on Tuesday, May 22nd, only some users will be able to use it at launch. What separates YouTube's music streaming service from the competition is its catalog of remixes, live versions, and covers of official versions of songs. It also uses the Google Assistant to make music recommendations based on everything it knows (and can learn) about you and your listening habits. "When you arrive at the gym, for example, YouTube Music will offer up a playlist of hard-hitting pump-up jams (if that's your thing)," reports Audioholics. "Late at night, softer tunes will set a more relaxing mood."

YouTube Music is free with ads, but will cost $9.99 for ad-free listening. There is also YouTube Premium, which will cost $11.99 per month, and will include both the ad-free music service and the exclusive video content from the now-defunct YouTube Red.
Microsoft

The Whole World is Now a Computer, Says Microsoft CEO Satya Nadella (zdnet.com) 182

Thanks to cloud computing, the Internet of Things and artificial intelligence, we should start to think of the planet as one giant computer, according to Microsoft chief executive Satya Nadella. From a report: "Digital technology, pervasively, is getting embedded in every place: every thing, every person, every walk of life is being fundamentally shaped by digital technology -- it is happening in our homes, our work, our places of entertainment," said Nadella speaking in London. "It's amazing to think of a world as a computer. I think that's the right metaphor for us as we go forward."

[...] AI is core to Microsoft's strategy, Nadella said: "AI is the run time which is going to shape all of what we do going forward in terms of applications as well as the platform." Microsoft is rethinking its core products by using AI to connect them together, he said, giving an example of a meeting using translation, transcription, Microsoft's HoloLens and other devices to improve decision-making. "The idea that you can now use all of the computing power that is around you -- this notion of the world as a computer -- completely changes how you conduct a meeting and fundamentally what presence means for a meeting," he said.

AI

New Toronto Declaration Calls On Algorithms To Respect Human Rights 168

A coalition of human rights and technology groups released a new declaration on machine learning standards, calling on both governments and tech companies to ensure that algorithms respect basic principles of equality and non-discrimination. The Verge reports: Called The Toronto Declaration, the document focuses on the obligation to prevent machine learning systems from discriminating, and in some cases violating, existing human rights law. The declaration was announced as part of the RightsCon conference, an annual gathering of digital and human rights groups. "We must keep our focus on how these technologies will affect individual human beings and human rights," the preamble reads. "In a world of machine learning systems, who will bear accountability for harming human rights?" The declaration has already been signed by Amnesty International, Access Now, Human Rights Watch, and the Wikimedia Foundation. More signatories are expected in the weeks to come.

Beyond general non-discrimination practices, the declaration focuses on the individual right to remedy when algorithmic discrimination does occur. "This may include, for example, creating clear, independent, and visible processes for redress following adverse individual or societal effects," the declaration suggests, "[and making decisions] subject to accessible and effective appeal and judicial review."

Slashdot Top Deals