<h2>The Godfather</h2>
<p>Yoshua Bengio, a computer scientist who&#8217;s widely considered one of the three &#8220;godfathers&#8221; of deep learning, opened up in a <a href="https://www.bloomberg.com/news/articles/2019-02-02/deep-learning-godfather-bengio-worries-about-china-s-use-of-ai">new interview</a> with <em>Bloomberg</em> about his anxiety about how China is using the technology he helped develop to surveil and manipulate large populations.</p>
<p>&#8220;This is the 1984 Big Brother scenario,&#8221; he told the business magazine, referring to George Orwell&#8217;s classic dystopic novel about universal government surveillance. &#8220;I think it’s becoming more and more scary.&#8221;</p>
<h2>Mug Shot</h2>
<p>Bengio, who&#8217;s a professor at the University of Montreal and the co-founder of deep learning company Element AI, is particularly troubled by China&#8217;s use of facial recognition technology.</p>
<p>Though American tech companies including Amazon also offer facial recognition software, Benio argued that the tech should be &#8220;highly regulated&#8221; because of its capacity for abuse, according to <em>Bloomberg.</em></p>
<h2>Bleak Outlook</h2>
<p>For a person whose work underlies much of the current boom in artificial intelligence technologies, Bengio is strikingly pessimistic about the role of the tech in the future.</p>
<p>&#8220;Technology, as it gets more powerful, outside of other influences, just leads to more concentration of power and wealth,&#8221; he told <em>Bloomberg</em>. &#8220;That is bad for democracy, that is bad for social justice, and the general well-being of most people.&#8221;</p>
<div>
<p><strong>READ MORE: </strong><a href="https://www.bloomberg.com/news/articles/2019-02-02/deep-learning-godfather-bengio-worries-about-china-s-use-of-ai">Deep Learning ‘Godfather’ Bengio Worries About China&#8217;s Use of AI</a> [<i>Bloomberg</i>]</p>
</div>
<div>
<p><em><strong>More on deep learning: </strong><a href="https://futurism.com/artificial-intelligence-automating-hollywood-art">Artificial Intelligence Is Automating Hollywood. Now, Art Can Thrive.</a></em></p>
</div>


"That is bad for democracy, that is bad for social justice, and the general well-being of most people."

Godfather of deep learning: China's AI tech is "1984 big brother"

Artificial Intelligence

Yoshua Bengio opened up in a new interview about his anxiety about how China is using deep learning to surveil and manipulate large populations.

Godfather of Deep Learning: China’s AI Tech Is “1984 Big Brother”

"This is the 1984 Big Brother scenario. I think it’s becoming more and more scary."

<h2>Out With the Old</h2>
<p>Artificial intelligence developers may soon find themselves on the brink of a paradigm shift. Deep learning has dominated the field for several years — but may be on its way out.</p>
<p>The field of AI has shifted focus roughly every decade since it began in the 1950s. Now, a new analysis suggests that the 2020s will be no different.</p>
<h2>Number Crunching</h2>
<p>A <a href="https://www.technologyreview.com/s/612768/we-analyzed-16625-papers-to-figure-out-where-ai-is-headed-next/">team from <em>MIT Technology Review</em></a> scanned all 16,625 research papers in the artificial intelligence section of arXiv, an open-source repository for sharing research, that were published between 1993 and November 18.</p>
<p><em>MIT Tech</em> found that on machine learning started to pick up over the last 20 years — rapidly increasing in prevalence since about 2008 — but now that research fervor seems to be dying down.</p>
<h2>Unpredictable Future</h2>
<p>Many of the new developments in artificial intelligence that we hear about nowadays are actually <a href="https://futurism.com/artificial-intelligence-hype">just applications of machine learning</a> techniques that have been hammered out for years.</p>
<p>And as the research community&#8217;s attention shifts from deep learning, it remains unclear what will take its place, according to <em>MIT Tech</em>. In the past, older types of artificial intelligence that didn&#8217;t really take off when they were first developed later resurfaced and taken off with new life. For instance, scientists first developed machine learning decades ago, but it only became commonplace about a decade ago.</p>
<p><em>MIT Tech</em> didn&#8217;t predict what will come next. It may be that some form of existing technology will finally hits its stride, but it&#8217;s also possible that an AI engineer will develop some brand-new type of AI that&#8217;ll shape the future.</p>
<p><strong>READ MORE: <a href="https://www.technologyreview.com/s/612768/we-analyzed-16625-papers-to-figure-out-where-ai-is-headed-next/">We analyzed 16,625 papers to figure out where AI is headed next</a> </strong>[<em>MIT Technology Review</em>]</p>
<p><strong><em>More on AI:</em></strong> <strong><em><a href="https://futurism.com/deep-learning-help-first-responders-offer-critical-aid-wake-disasters">Deep Learning Could Help First Responders Offer Critical Aid in the Wake of Disasters</a></em></strong></p>


But we don't yet know what will take its place.

Analysis: "The era of deep learning is coming to an end"

Time to Upgrade

According to a massive analysis by MIT Technology Review, the AI research community may soon move away from deep learning toward something new.

Analysis: “The Era of Deep Learning Is Coming to an End”

Nothing lasts forever.

<h2>Prime-ary Care</h2>
<p>Bezos and company are betting big on the future of healthcare. Amazon began <a href="https://futurism.com/amazons-healthcare-plans">hinting at</a> healthcare ambitions in late 2016 and has taken incremental steps since, including <a href="https://futurism.com/amazon-just-opened-a-new-lab-dedicated-to-healthcare-tech/">forming a healthcare lab</a> and purchasing an <a href="https://futurism.com/the-byte/pillpack-amazon-prescription-drugs">online pharmacy company</a>. Now Amazon has <a href="https://www.wsj.com/articles/amazon-starts-selling-software-to-mine-patient-health-records-1543352136">begun selling software</a> that analyzes patient medical records for hospitals and their physicians.</p>
<p>Although hospitals across the United States have been creating electronic patient records for years, Amazon&#8217;s software makes these records easier to search for key information, saving time and money by freeing up essential staff to take care of patients. Most notably the software transcribes handwritten notes, including medical jargon and abbreviations, saving doctors the time it takes to manually re-record their notes.</p>
<h2>Dr. AI</h2>
<p>If you&#8217;ve ever struggled to decipher a doctor&#8217;s handwriting, rest assured it&#8217;s hard for AI too. Amazon&#8217;s text-analysis software performed on par or better with other software and is capable of successfully gleaning the most important information from patient records, it claims.</p>
<p>&#8220;We’re able to completely, automatically look inside medical language and identify patient details,&#8221; including diagnoses, treatments, dosage and strengths, &#8220;with incredibly high accuracy,&#8221; Matt Wood, general manager of artificial intelligence at Amazon Web Services, told <em>The Wall Street Journal</em>.</p>
<p>Healthcare professionals can then export this data into easy-to-read records of specific information they can use to improve their services. Patient data is subject to the notoriously strict Health Insurance Portability and Accountability Act, so medical records processed by Amazon can only be able to be accessed using a customer-specific authorization key.</p>
<h2>Out-of-Network Provider</h2>
<p>This isn&#8217;t the first time algorithms have been used to aggregate and analyze data. A 2015 initiative at New York&#8217;s Mount Sinai Hospital produced <a href="https://www.technologyreview.com/s/604087/the-dark-secret-at-the-heart-of-ai/">Deep Patient</a>, which, similar to Amazon&#8217;s software, processed patient records and even went as far as to predict potential illnesses.</p>
<p>Deep learning algorithms are excellent at processing large datasets, but the sensitive nature of the data in question means these processes must be handled carefully. Maybe Amazon&#8217;s early steps could foretell a shift from two-day shipping to two-day diagnosis — or even quicker.</p>
<p><strong>READ MORE:</strong> <a href="https://www.wsj.com/articles/amazon-starts-selling-software-to-mine-patient-health-records-1543352136">Big Tech Expands Footprint in Health</a></p>
<p><strong>More on healthcare:</strong> <a href="https://futurism.com/the-byte/biggest-tech-companies-hack-heart-health">Big Tech Companies Want to Hack Your Heart Health</a></p>


Big data has a role in healthcare, if it's handled properly.

Amazon's next AI bet: digitizing medical records

Alexa, I Don't Feel So Good

Amazon continues to invest in healthcare and has begun selling software which analyzes patient medical records for hospitals and their physicians.

Amazon's Next AI Bet: Digitizing Medical Records

Amazon wants in on medtech.

Amazon’s Next AI Bet: Digitizing Medical Records

<p>It’s date night! Time to throw on your oversized graduation cap and ruffly shirt, spritz yourself with some of your new “Hivinga” perfume, and meet your better half for some sweet potato, beans, and brie pizza covered with &#8220;snipped caramel cheese.&#8221;</p>
<p>Wait, what? Things that don&#8217;t seem quite right here. Oh, right, that&#8217;s because all these date night tips came from a series of artificial intelligence algorithms custom-built to generate new clothing designs, pizza recipes, and perfumes.</p>
<p>Whatever bizarre (and, frankly, unappetizing) ideas the AI systems conjured were built by a group of MIT researchers and students called <a href="https://howtogeneratealmostanything.com/">How to Generate (Almost) Anything</a>. The team builds and trains algorithms that can tackle seemingly creative tasks to come up with brand new designs for each week&#8217;s theme. The researchers then build, bake, or stitch together whatever the AI system came up with. Their goal: these strange, unusable things will help the average person understand and appreciate what <a href="https://futurism.com/artificial-intelligence-hype">AI can do</a>, and what it can&#8217;t yet.</p>
<p>“The larger goal of the project is to democratize AI for the public. Currently, the creative usage of AI is within the reach of AI practitioners,” computer scientist <a href="http://www.mit.edu/~pinary/">Pinar Yanardag</a>, the MIT postdoctoral researcher who founded the group, told Futurism.</p>
<p>Most of the projects used machine learning algorithms that, upon being trained with thousands of examples of pizzas and perfume recipes, spat out unique creations that were sort of like the thing they were supposed to create, but just a little bit off. Often, the results were outlandish and alien. For instance, several pizzas called for fictional ingredients or skipped crucial steps in the baking process.</p>
<p>But two of the projects, the algorithms that designed dresses and graffiti, used a generative adversarial network (GAN), a type of algorithm used to create things that seem like the real deal. As a result, the fashionbot and graffitibot were a touch less goofy than all the others.</p>
<figure id="attachment_139906" style="width: 600px" class="wp-caption alignnone"><img class="wp-image-139906 size-full" src="https://wordpress.futurism.com/wp-content/uploads/2018/10/bizarre-ai-creations-generate1.jpg" alt="" width="600" height="399" /><figcaption class="wp-caption-text">Image Credit: How to Generate (Almost) Anything</figcaption></figure>
<p>Granted, the algorithms could produce slightly less ridiculous results — like, for instance, pizza recipes that actually result in something edible. But that wasn&#8217;t what Yanardag and her team wanted to accomplish here. Their primary goal was to make something that would be fun and accessible for the average person to use.</p>
<figure id="attachment_139907" style="width: 600px" class="wp-caption alignnone"><img class="wp-image-139907 size-full" src="https://wordpress.futurism.com/wp-content/uploads/2018/10/bizarre-ai-creations-generate2.jpg" alt="" width="600" height="399" /><figcaption class="wp-caption-text">Image Credit: How to Generate (Almost) Anything</figcaption></figure>
<p>Every week, the team releases a new video highlighting the algorithm they trained and what it came up with. They&#8217;re working on a user-friendly tool that would let people play around with the algorithm — no technical skills required — and generate their own versions of <a href="https://howtogeneratealmostanything.com/food/2018/08/30/episode2.html">that week’s theme</a>, whether it’s graffiti, pizza, perfume, or music. That tool is still a work in progress, but the group shares extra designs and recipes from their algorithms for people to try making at home.</p>
<p>Normally, bizarre AI creations are merely the <a href="https://futurism.com/scripts-written-ai-hilarious-fake-how-to-spot-difference">stuff of twitter threads</a>. By making the algorithmic output tangible, Yanardag hopes that people will develop a better grasp of machine learning&#8217;s capabilities. Advanced AI, Yanardag&#8217;s project suggests, can be used to <em>help </em>us in our creative endeavors, not simply replace us.</p>
<p><strong><em>More on “creative” AI: <a href="https://futurism.com/artificial-intelligence-video-games">Artificial Intelligence Is Making Video Game Levels So Good That Even Other AI Thinks They’re Man-Made</a></em></strong></p>
<p><em>(10/17/2018, 3:15 PM) Editor&#8217;s note: a previous version of this article said that the group&#8217;s tool for letting the public generate their own AI creations was live. The article has been updated to reflect that the tool is a work in progress. For now, the group is sharing extra recipes and designs that their algorithm generated but they didn&#8217;t make in real life.</em></p>


A graduation cap with a super large rectangular brim really makes a statement.

Meet the researchers bringing bizarre AI creations to life

WE HAVE THE TECHNOLOGY

How to Generate (Almost) Anything is collaborating with their custom AI systems to bring bizarre foods, clothes, and art into the real world.

Meet the Researchers Bringing Bizarre AI Creations to Life

The MIT group calls itself "How to Generate (Almost) Anything," and it wants to make AI available to the masses.

<h2>Easier, Faster, Holograms</h2>
<p>In two new studies, researchers at UCLA used artificial neural networks to <a href="https://www.eurekalert.org/pub_releases/2018-02/cioo-dlr010218.php">reconstruct a hologram</a> — not just any hologram, though. Not only is the technique an advancement of holographic technology, but the resulting microscopic images could have fascinating medical applications.</p>
<p>In <a href="https://www.nature.com/articles/lsa2017141">the first study,</a> published in the journal, <i><a href="https://www.nature.com/articles/lsa2017141" target="_blank" rel="noopener">Light: Science &amp; Applications,</a> </i>researchers used deep learning to create images of biological samples like blood, Pap smears, and other thin tissue samples.</p>
<p>The neural network technique proved to be easier and faster than the usual methods used to make holograms, which often require an abundance of physical measurements and computational input.</p>
<p>In <a href="https://www.osapublishing.org/optica/abstract.cfm?uri=optica-4-11-1437">the second study</a>, the team applied their deep learning framework to improve the resolution and quality of the microscopic images, which could help doctors detect very small abnormalities in a large blood or tissue sample.</p>
<p>As holograms are rendered they can lose information in the process, which can sometimes make &#8220;artifacts&#8221; appear in the image; something that looks significant (like a dark spot that&#8217;s really just a shadow but could be interpreted as a cancerous growth). This also happens at times with radiological scans, especially if the patient moves when the scan is being performed.</p>
<p>However, the UCLA team&#8217;s deep learning AI was successfully able to address the problem: once it was properly trained, the neural network could separate the spatial features of the true image from any interference (often caused by light).</p>
<p><iframe width="500" height="281" src="https://www.youtube.com/embed/1cQbMP3I5Sk?feature=oembed" frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe></p>
<h2>Practical Holograms</h2>
<p>Layers of artificial neural networks allow deep learning algorithms to automatically analyze data. The technology has already demonstrated its use for improving <a href="https://futurism.com/microsofts-speech-recognition-is-now-as-good-as-a-human-transcriber/">real-time speech translation,</a> video captioning, and many other tasks previous left up to humans — who, perhaps, don&#8217;t accomplish them as speedily or accurately as the algorithms do.</p>
<p>Since machine learning can sort through enormous amounts of information much faster than humans can, one area that the technology has begun to establish a presence is medicine. The algorithms are already finding applications in diagnostic radiology, where they&#8217;ve been used to <a href="https://futurism.com/watch-out-doc-an-ai-is-going-to-start-interpreting-x-rays/">read x-rays</a> and have even managed to catch cancers <a href="https://futurism.com/human-radiologists-missed-7-of-cancers-in-a-study-a-deep-learning-algorithm-missed-0/">that were missed by human physicians</a> on scans.</p>
<p>As we strive to develop tools to help bring these views &#8220;to life&#8221; <a href="https://arxiv.org/ftp/arxiv/papers/1705/1705.04286.pdf">h</a><a href="https://arxiv.org/ftp/arxiv/papers/1705/1705.04286.pdf">olographic image reconstruction</a> will be integral — especially if they&#8217;re able to repair and recover the images.</p>
<p>Beyond the remarkable potential deep learning has to advance hologram technologies, Aydogan Ozca, who led the research, believes it could open up new possibilities in imaging. In <a href="http://newsroom.ucla.edu/releases/ucla-engineers-use-deep-learning-to-reconstruct-holograms-and-improve-optical-microscopy">a UCLA press release, Ozca said</a> the technology could even lead to the creation of entirely new, coherent imaging systems. Future imaging systems could even use UCLA&#8217;s findings to incorporate other parts of the electromagnetic spectrum into the tech, like X-rays and visible wavelengths.</p>
<p>If the future is anything like we&#8217;ve seen in science fiction for the last 40 to 50 years, <a href="https://futurism.com/holograms-life-worlds-first-interactive-lightfield-display/" target="_blank" rel="noopener">holograms will play a major role</a>, but UCLA&#8217;s research doesn&#8217;t just support the fantastic future of holograms; it offers realistic <a href="https://futurism.com/a-new-artificially-intelligent-hologram-was-just-born/" target="_blank" rel="noopener">applications for the incredible technology</a>.</p>


The research doesn't just support the fantastic future of holograms — it offers realistic applications, too.

How deep learning AI will help hologram technology find practical applications

Not only is the technique an advancement of holographic technology, but also, the holograms could have fascinating (and practical) medical applications.

How Deep Learning AI Will Help Hologram Technology Find Practical Applications

Especially in medicine. 

<h2>Applied Deep Learning</h2>
<p>From <a href="https://futurism.com/this-hurricane-season-has-been-exceptionally-bad-and-its-not-going-to-get-any-better/">hurricanes</a> to <a href="https://futurism.com/this-could-be-the-wests-worst-fire-season-ever/">wildfires</a>, 2017 brought the world a number of natural disasters — as well as some <a href="https://futurism.com/the-red-cross-officially-launched-the-first-drone-program-for-disasters/">tech to deal with them</a>. We have more information than ever following a disaster thanks to unmanned aerial vehicles (UAVs) and sophisticated satellites that can capture images of disasters from the air, but we are still working on ways to process the data so it is valuable for relief efforts. That&#8217;s where deep learning comes in, says the <a href="http://www.worldbank.org/">World Bank</a> in collaboration with <a href="https://werobotics.org/">WeRobotics</a> and <a href="https://openaerialmap.org/">OpenAerialMap</a>.</p>
<figure style="width: 360px" class="wp-caption alignright"><a href="https://futurism.com/?post_type=image&p=58365"><img src="https://wp-assets.futurism.com/2016/11/HomeThumb_V1.jpg" alt="The 5 Weirdest AI Applications [INFOGRAPHICS]" ></a><figcaption class="wp-caption-text center"><strong><a href="https://futurism.com/?post_type=image&p=58365" target="_blank">Click to View Full Infographic</a></figcaption></strong></figure>
<p>On Jan. 10, 2018, World Bank issued an artificial intelligence (AI) challenge to explore how deep learning could <a href="https://www.forbes.com/sites/kalevleetaru/2018/01/20/using-ai-for-good-a-new-data-challenge-to-use-ai-to-triage-natural-disaster-aerial-imagery/#5f5f6261d26e" target="_blank" rel="noopener">be used in the wake of natural disasters</a>. Deep learning  is what<a href="https://futurism.com/deep-learning-is-going-to-teach-us-all-the-lesson-of-our-lives-jobs-are-for-machines/" target="_blank" rel="noopener"> enables AI to recognize patterns</a> in images, sounds, and other data using a neural network that mirrors our own grey matter. This deep learning software is what helps Alexa recognize speech patterns, Google Translate to interpret entire sentences, and Facebook&#8217;s AI labs to <a href="https://www.wired.com/2013/12/facebook-yann-lecun-qa/" target="_blank" rel="noopener">automatically identify and tag users in uploaded photographs</a>.</p>
<p>AI could be used to catalog aerial images in the critical periods following disasters and help first responders and humanitarian aid agencies aggregate information. Sorting images quickly en masse would make it easier to assess which areas need immediate assistance, what the clearest paths in and out of a disaster site are, and where the most infrastructure damage is.</p>
<p>The <a href="https://werobotics.org/blog/2018/01/10/open-ai-challenge/" target="_blank" rel="noopener">AI challenge announcement</a> by WeRobotics founder Patrick Meier focuses on Pacific Island countries, which are vulnerable to earthquakes, tsunamis, storm surges, volcanic eruptions, landslides, and droughts. In the last decade alone, major cyclones have caused millions of dollars of damage in hundreds of islands, including Fiji and Samoa, Meier wrote.</p>
<h2>Identifying Trees and Roads</h2>
<p>The World Bank’s UAVs for Disaster Resilience Program captured about 80 square km (31 square miles) of high-resolution aerial imagery in the island of Tonga. Now, the World Bank is challenging participants to develop machine learning algorithms that will analyze this imagery without human assistance. In future, that learning will be &#8220;applied to new imagery to speed up baseline analysis and damage assessments,&#8221; <a href="https://docs.google.com/document/d/16kKik2clGutKejU8uqZevNY6JALf4aVk2ELxLeR-msQ/edit" target="_blank" rel="noopener">according to the announcement</a>.</p>
<figure id="attachment_119610" style="width: 800px" class="wp-caption aligncenter"><img class="wp-image-119610 size-full" src="https://wp-assets.futurism.com/2018/01/Damage_at_Nela_Village_Temotu_Province_Solomon_Islands_10690986314.jpg" alt="" width="800" height="533" /><figcaption class="wp-caption-text">In 2013, a magnitude 8 earthquake and its aftershocks destroyed infrastructure in the Solomon Islands. Credit: <a href="https://commons.wikimedia.org/wiki/File%3ADamage_at_Nela_Village%2C_Temotu_Province%2C_Solomon_Islands_(10690986314).jpg">Wikimedia Commons</a>.</figcaption></figure>
<p>In particular, developers should focus on trees and roads. The algorithms need to identify all coconut, banana, papaya, and mango trees and their locations with at least 80 percent accuracy, since the loss of those critical food production trees would impact both food security for island residents and their economies following a disaster.</p>
<p>The automated imagery analysis should also assess road conditions, like whether they are paved and how many lanes they have. Road assessments for disaster area could allow first responders to plan which roads to use to transport aid effectively.</p>
<p>In an era of increased social media, tailored advertising, and big data, it&#8217;s easy to forget that AI can be used for more than just improving home technology and the user experience. This challenge from the World Bank and its collaborators is a welcomed reminder that deep learning could prove useful in humanitarian aid efforts as well.</p>


This could improve our responses to earthquakes, tsunamis, and droughts.

Deep learning could help first responders offer critical aid in the wake of disasters

Developers are being challenged to craft deep learning algorithms to assess the conditions of roads and food production trees following natural disasters.

Deep Learning Could Help First Responders Offer Critical Aid in the Wake of Disasters

Artificial intelligence could help us deal with the aftermath of hurricanes and wildfires.

<h2>&#8220;Naïve&#8221; Neural Networks</h2>
<p>Engineers at <a href="http://www.purdue.edu/" target="_blank" rel="noopener">Purdue University</a> in Lafayette, Indiana are developing a new system for keeping nuclear reactors safe with artificial intelligence (AI). In the paper published in the <i><a href="http://ieeexplore.ieee.org/document/8074762/" target="_blank" rel="noopener">IEEE Transactions on Industrial Electronics</a></i> journal, the researchers introduced a deep learning framework called a naïve Bayes-convolutional neural network that can effectively identify cracks in reactors by analyzing individual video frames. The method could potentially make safety inspections safer.</p>
<p>“Regular inspection of nuclear power plant components is important to guarantee safe operations,” Mohammad Jahanshahi, an assistant professor at Purdue’s Lyles School of Civil Engineering, said in<a href="http://www.purdue.edu/newsroom/releases/2017/Q4/system-uses-deep-learning-to-detect-cracks-in-nuclear-reactors.html" target="_blank" rel="noopener"> a press release</a>. “However, current practice is time-consuming, tedious, and subjective and involves human technicians reviewing inspection videos to identify cracks in reactors.”</p>
<p>Trained using a dataset of some 300,000 crack and non-crack patches, Purdue&#8217;s AI works by viewing video images of the reactors, which are often submerged under water to keep them cool, making them even more difficult to inspect manually. The AI scans each and detects cracks in overlapping &#8220;patches&#8221; of the video frames. Each crack is tracked from one frame to another using a data fusion algorithm.</p>
<p>&#8220;Data fusion comes up with more robust decision making than would otherwise be possible,” said Jahanshahi, who noted that their AI scored a 98.3 success rate in identifying the cracks, which is significantly higher than even today&#8217;s most advanced approaches.</p>
<p><iframe width="500" height="281" src="https://www.youtube.com/embed/8O8FFey4GJo?feature=oembed" frameborder="0" gesture="media" allowfullscreen></iframe></p>
<h2>The &#8220;Holy Grail&#8221; of Renewable Energy</h2>
<p>As the world continues to shift towards more renewable sources of energy, nuclear has presented itself as an option. One reason there&#8217;s increasing interest in nuclear as an alternative energy source is that it&#8217;s devoid of the usual limitations of solar and wind, which depend on the right conditions to generate power. The ultimate goal as many see it, however, is to harness the so-called &#8220;holy grail&#8221; of renewable energy: nuclear fusion.</p>
<p>While researchers have achieved considerable success in stabilizing and sustaining a fusion reaction, it will be some time yet before we can safely rely on it for our daily energy needs.</p>
<p>The viable type of nuclear power currently available to us come in the form of nuclear <em>fission, </em>which researchers around the world are working to improve. Experts are already making headway using <a href="http://www.world-nuclear.org/information-library/current-and-future-generation/molten-salt-reactors.aspx">molten-salt nuclear reactors</a>; advanced nuclear reactors that use fluid rather than solids, meaning the salt can function as both the fuel <em>and </em>the coolant.</p>
<p>The safety system developed at Purdue could help bolster public support, which would ideally keep the nuclear option for alternative energy open. Since the Chernobyl disaster in 1986,<a href="http://content.csbs.utah.edu/~mli/Economies%205430-6430/Sovacool-Nuclear%20Power%20and%20Renewable%20Electricity%20in%20Asia.pdf"> 56 of the 99 major nuclear power accidents</a> have occurred on U.S. soil (a nation where nuclear power accounts for 20 percent of the electricity generated).</p>
<p>“One important factor behind these incidents has been cracking that can lead to leaking,” Jahanshahi explained. “Nineteen of the above incidents were related to cracking or leaking, costing $2 billion. Aging degradation is the main cause that leads to function losses and safety impairments caused by cracking, fatigue, embrittlement, wear, erosion, corrosion, and oxidation.”</p>
<p>Keeping nuclear energy safe is, therefore, an important step in guaranteeing a wider adoption of what is, essentially, the most renewable energy source we have at our disposal. As such, nuclear energy would contribute a significant blow to the threat of climate change and global warming.</p>


The deep learning framework can effectively identify cracks in nuclear reactors by analyzing individual video frames.

How artificial intelligence is making nuclear reactors safer

Earth & Energy

Using artificial intelligence, engineers at Purdue University have found a new method to track potentially disasterous cracks in nuclear reactors.

Researchers Are Training AI to Help Make Nuclear Reactors Safer

The AI has a 98.3 success rate in identifying the cracks, which is significantly higher than even the most advanced current methods.

How Artificial Intelligence Is Making Nuclear Reactors Safer

<h2>Cracking CAPTCHAs</h2>
<p>A group of researchers has developed a computer model that&#8217;s capable of cracking text-based CAPTCHA keys. Given that the purpose of CAPTCHA is to test whether the entity attempting to access a service is human, this is a considerable step forward for developing computers that think like humans.</p>
<p>Text-based CAPTCHAs–the name coming from the phrase Completely Automated Turing Test to Tell Computers and Humans Apart–are often used by web services that might suffer an attack based around an influx of automated users. Account sign-up services and ticketing sites often use them to confirm that visitors are human.</p>
<p>CAPTCHA work by displaying letters that are relatively easy for a human read, but hard for a computer that&#8217;s trained on the standard shapes of the alphabet to decipher. Previously, there have been attempts to crack the CAPTCHA system, but they&#8217;ve revolved around weaknesses in a particular implementation of the concept, and as such have been easy to counteract. This project is capable of parsing the text much more effectively without much training.</p>
<p>This study used an approach known as a Recursive Cortical Network, which builds a model of how different letters should look based on standard examples. When presented with an image to analyze, the network tries to explain it by looking at all the pixels and comparing them to its prior knowledge. This makes it possible for the computer to make assumptions about things, like one character being partially hidden behind another.</p>
<h2>Machine Mind</h2>
<p>This particular study may have focused on CAPTCHAs, but they&#8217;re really just a means to an end for a broader goal: to figure out how to make computers that think like humans.</p>
<p>Many attempts to make computers that think like humans rely on deep learning, where the system is trained to respond to a situation in a certain way by being exposed to a large quantity of examples. This has its limitations, though; in this case, showing a computer lots of examples of the letter &#8216;A&#8217; and lots of examples of the letter &#8216;B&#8217; wouldn&#8217;t prepare it for a scenario where both letters overlap one another.</p>
<p>&#8220;Robots need to understand the world around them and be able to reason with objects and manipulate objects,&#8221; said research leader Dileep George, co-founder of AI company Vicarious, in an <a href="http://www.npr.org/sections/thetwo-way/2017/10/26/560082659/ai-model-fundamentally-cracks-captchas-scientists-say?" target="_blank" rel="noopener">interview with NPR</a>. &#8220;So those are cases where requiring less training examples and being able to deal with the world in a very flexible way and being able to reason on the fly is very important, and those are the areas that we&#8217;re applying it to.&#8221;</p>
<p>While this study does demonstrate how computers might be taught to defeat text-based CAPTCHA tests, the technology already seems to be on its way out; Google has <a href="https://futurism.com/googles-new-ai-can-tell-if-you-are-human-or-not/" target="_blank" rel="noopener">already discontinued its usage</a>. Artificial intelligence is getting more and more capable, and as such our tests to determine <a href="https://futurism.com/rights-synthetic-lifeforms-civil-rights-controversy/" target="_blank" rel="noopener">who is human and who isn&#8217;t</a> will have to become more sophisticated, too.</p>


The key was training the computer to reason on the fly when encountered with new variations, like overlapping letters.

Computer system beats CAPTCHA checks by thinking more like a human

computer desktop captcha

An AI company trained a computer to beat CAPTCHA tests, which test if a user is human. This brings us one step closer to computers that think like humans.

Computer System Beats CAPTCHA Checks By Thinking More Like a Human

The study trained a system to cross-reference what it knows about individual characters.

<h2>An AI Horror</h2>
<p>Developers of artificial intelligence (AI) systems have been dabbling in the arts for a while now. There&#8217;s already an AI capable of <a href="https://futurism.com/the-worlds-first-album-composed-and-produced-by-an-ai-has-been-unveiled/" target="_blank" rel="noopener">composing original music</a> — it even has its own album — <a href="https://futurism.com/ai-wrote-this-short-sci-fi-film-starring-david-hasselhoff/" target="_blank" rel="noopener">writing film screenplays</a>, and <a href="https://futurism.com/ai-now-produce-better-art-humans-heres-how/" target="_blank" rel="noopener">even painting</a>. Now, as an early Halloween treat, the latest AI artist is aspiring to be a horror novelist.</p>
<p>Meet <a href="http://shelley.ai/" target="_blank" rel="noopener">Shelley AI</a>, a deep learning algorithm that was developed by researchers at the Massachusetts Institute of Technology (MIT) and named after Victorian-era novelist Mary Shelley who penned <i>Frankenstein</i>. The AI was trained using stories collected from<a href="https://www.reddit.com/r/nosleep/" target="_blank" rel="noopener"> a subreddit </a>dedicated to sharing original eerie tales. <a href="http://shelley.ai/#about" target="_blank" rel="noopener">According to its developers</a>, Shelley &#8220;takes a bit of inspiration in the form of a random seed, or a short snippet of text, and starts creating stories emanating from her creepy creative mind.&#8221;</p>
<figure style="width: 820px" class="wp-caption aligncenter"><img class="wp-image-108194 size-large aligncenter" src="https://wp-assets.futurism.com/2017/10/Shelley-AI.png" alt="Image credit: Shelley AI, screenshot" width="820" height="515" /><figcaption class="wp-caption-text">Shelley is fully equipped to terrify you. Image Credit: Shelley AI, screenshot</figcaption></figure>
<p>Here&#8217;s an example of how creative and scary Shelley can be:</p>
<blockquote><p>I could feel something watching me and I could see this dark shadow standing there with a torrent of hatred in its face. I was beyond scared so I didn&#8217;t take my eyes off this thing and turned back towards the mirror to see what was behind me!</p></blockquote>
<p>If that&#8217;s not blood-curdling enough, here&#8217;s another snippet:</p>
<blockquote><p>She fell to the floor from her cries and muttered a soft &#8216;Come to  meeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee&#8217;. He stood rooted with fear, she slowly crawled spider-like towards him, and started shrieking &#8216;no escape nowhere to hide&#8217;.</p></blockquote>
<h2>Even AI Needs Inspiration</h2>
<p>Shelley, it seems, can write like a pro, although most of this AI&#8217;s work is currently in snippets or short forms. The developers designed Shelley to work in tandem with humans, &#8220;Shelley truly enjoys is to work collaboratively with humans, learning from their nightmarish ideas, creating the best scary tales ever.&#8221; If you&#8217;re interested in contributing, you can add your own horror ideas <a href="https://twitter.com/shelley_ai" target="_blank" rel="noopener">via Twitter</a>.</p>
<p>Here&#8217;s one original by Shelley entitled &#8220;Mirror&#8221;:</p>
<blockquote><p>I slowly moved my head away from the shower curtain, and saw the reflection of the face of a tall man who looked like he was looking in the mirror in my room. I still couldn&#8217;t see his face, but I could just see his reflection in the mirror. He moved toward me in the mirror, and he was taller than I had ever seen. His skin was pale, and he had a long beard.</p></blockquote>
<p>After Shelley incorporated contributions from Twitter, the story continues like this.</p>
<blockquote><p>&#8220;Please, don&#8217;t be scared, I&#8217;m just looking for her,&#8221; he said pointing to the mirror. I nodded, frozen, because something touched me. I remember he looked down at me, and screamed in agony. He didn&#8217;t say anything, but he looked over at me with that unnatural expression.</p></blockquote>
<p>If you hadn&#8217;t been told those lines were written by AI, you wouldn&#8217;t have suspected a thing. <a href="http://stories.shelley.ai/hospital/" target="_blank" rel="noopener">From some of her original sentences</a>, Shelley&#8217;s even able to produce two versions.</p>
<p>What might be most fascinating about Shelley is how it&#8217;s learned to identify which words best project horror, and it combines these to produce sentences or paragraphs as if they were taken from a Stephen King novel. Perhaps that shouldn&#8217;t be surprising since AIs have become fairly adept at understanding language. If anything, Shelley is another demonstration of how well AI algorithms can learn — which, for some, is an ability that could end up being a horror story for humanity.</p>


There's a new horror novelist that wants to scare the living breath out of you. Developed by researchers from MIT, Shelley AI takes inspiration from original works and human Twitter contributions to spin the scariest words in a short horror prose.

A New AI is Writing Perverse Horror Fiction

The world just got a new author.

<h2>Birth of the Nervana Neural Network Processor</h2>
<p><a href="https://futurism.com/artificial-intelligence-is-our-future-but-will-it-save-or-destroy-humanity/" target="_blank" rel="noopener">The future of technology is artificial intelligence</a>, but before it can radically change the world, we need to develop proper hardware that can give the machines the ability to learn.</p>
<p>To that end, <i><a href="https://www.wsj.com/articles/facebook-pitches-in-on-intels-coming-artificial-intelligence-chip-1508263200" target="_blank" rel="noopener">The Wall Street Journal</a></i> reports that Intel is working with Facebook and a number of other companies on a new processor <a href="https://futurism.com/intel-is-releasing-a-processor-thats-built-for-neural-networks/" target="_blank" rel="noopener">called Nervana</a> —also known as the Nervana Neural Network Processor. The chip is the result of work done by <a href="https://www.intelnervana.com/" target="_blank" rel="noopener">Intel Nervana</a>, and was designed to accelerate the AI learning process known as deep learning, in which a computer is taught to recognize objects, patterns, speech, and complex concepts so that it can mimic the same learning capabilities exhibited by humans.</p>
<p><iframe width="500" height="281" src="https://www.youtube.com/embed/6twMD2DGXDM?feature=oembed" frameborder="0" allowfullscreen></iframe></p>
<p>Intel&#8217;s partners are helping &#8220;fine tune&#8221; Nervana, while Facebook has started taking a deeper look at the processor and realizing its capabilities.</p>
<p>&#8220;[They said], ‘Hey, this really could change the way we think about artificial intelligence. And help us really steer how we build software and hardware,” said Intel Chief Executive Brian Krzanich at The WSJ D.Live conference on Tuesday.</p>
<h2>Efficient Deep Learning</h2>
<p>When it comes to deep learning, existing computers must use a lot of processing power to handle the amount of data used during the teaching phase. Nervana, however, is meant to take shortcuts that can reduce the amount of processing power required, enabling it to perform calculations.</p>
<p>Intel has yet to detail how it&#8217;s tracking the chip&#8217;s performance, but the company seems confident in Nervana&#8217;s ability to impact technologies like weather forecasts and self-driving cars — the latter of which uses deep learning to recognize other cars on the road, as well as driving lines in order to safely navigate an environment. Intel also <a href="https://futurism.com/intel-and-waymo-have-teamed-up-to-roll-out-fully-autonomous-vehicles/" target="_blank" rel="noopener">recently partnered with Google&#8217;s Waymo</a> to developed better autonomous vehicles.</p>
<p><iframe width="500" height="281" src="https://www.youtube.com/embed/Dz_FfcimdUg?feature=oembed" frameborder="0" allowfullscreen></iframe></p>
<p>In <a href="https://newsroom.intel.com/editorials/intel-pioneers-new-technologies-advance-artificial-intelligence/?_ga=2.141275634.1353331571.1508529180-1083698026.1508529180" target="_blank" rel="noopener">a press statement</a>, Krzanich stated Intel has multiple Nervana projects in the works that will &#8220;deliver higher performance and enable new levels of scalability for AI models.&#8221; The company is currently on track to achieve 100 times greater AI performance by 2020.</p>
<p>&#8220;We designed the Intel Nervana NNP to free us from the limitations imposed by existing hardware, which wasn’t explicitly designed for AI,&#8221; said Naveen Rao, vice president of the AI Products Group at Intel Nervana, <a href="https://www.intelnervana.com/intel-nervana-neural-network-processors-nnp-redefine-ai-silicon/?_ga=2.149508598.1353331571.1508529180-1083698026.1508529180" target="_blank" rel="noopener">in a blog post</a>. &#8220;Our hope is to open up the possibilities for a new class of AI applications that are limited only by our imaginations.&#8221;</p>


"This really could change the way we think about artificial intelligence."

Facebook and Intel are joining forces for artificial intelligence

The chip is an integral part of multiple projects that are expected to help Intel achieve 100 times better AI performance by 2020.

/Deep Learning

Artificial Intelligence

Godfather of Deep Learning: China’s AI Tech Is “1984 Big Brother”

Time to Upgrade

Analysis: “The Era of Deep Learning Is Coming to an End”

Alexa, I Don't Feel So Good

Amazon’s Next AI Bet: Digitizing Medical Records

WE HAVE THE TECHNOLOGY

Meet the Researchers Bringing Bizarre AI Creations to Life

Artificial Intelligence

How Deep Learning AI Will Help Hologram Technology Find Practical Applications

Artificial Intelligence

Deep Learning Could Help First Responders Offer Critical Aid in the Wake of Disasters

Earth & Energy

How Artificial Intelligence Is Making Nuclear Reactors Safer

Artificial Intelligence

Computer System Beats CAPTCHA Checks By Thinking More Like a Human

Artificial Intelligence

A New AI is Writing Perverse Horror Fiction

Artificial Intelligence

Facebook and Intel Are Joining Forces for Artificial Intelligence

Load More