Category Archives: Ai

When you think of Artificial Intelligence, you think of all the great technological advancements that are outperforming humans at almost all tasks. This is great, but a lot of AI apps have emerged in the market today that make you question the societal and human impact of this highly-advanced technology. Speaking of which, a group of prominent researchers are alarmed by the harmful societal impact of AI and called for a ban on automated analysis of facial expressions in hiring and other related decisions, reports Reuters.

SEE ALSO: Facebook Has Made An Artificial Intelligence Assistant For Minecraft

As an example of problematic and harmful AI, researchers cited example of the company HireVue, that sells systems for remote video interviews of employers like Hilton and Unilever. HireVue is an AI-based app that analyzes facial movements, tone of voice, speech patterns without disclosing the score to the candidates who have applied for the job. In fact, the nonprofit Electronic Privacy Information Center has also filed a complaint about HireVue to the U.S. FTC.

AI Now, a New-York-based research institute released its fourth annual report on the effects of artificial intelligence tools, where it mentioned that job screening is one of the ways where these kind of software is used without accountability and favors privileged groups. The institute also stated that it wants to take priority action against these software-driven affect recognition, since science doesnt justify its use and there hasnt been widespread adoption of the technology yet. AI Now has also criticized Amazon for its Rekognition software.

HireVue said that it wasnt aware of the AI Now report and didnt respond to any questions surrounding criticism or complaint about the app. Many job candidates have benefited from HireVues technology to help remove the very significant human bias in the existing hiring process, said spokeswoman Kim Paone.

SEE ALSO: Chromes New Feature Uses AI To Describe Images For Blind And Low-Vision Users

A lot of other AI apps have come under scanner over concerns related to harmful use of Artificial Intelligence. For instance, an AI-based babysitter app, Predictim, that uses advanced AI to analyze the risk levels attached to a babysitter. The app gives you a risk score for the babysitter as well as complete details on the babysitter by scanning their social media accounts. Other AI app that received flak recently was the Image-Net Roulette, an online tool that used racist and offensive labels to describe and classify humans.

Follow this link:

AI Software That Analyzes Human Emotions Criticized By Researchers - Mashable India

UW Computer Science Chair Hank Levy speaks at the TechAliance AI Policy Matters Summit. (GeekWire Photo / Monica Nickelsburg)

About eight years ago, leaders of the University of Washingtons computer science department decided to zero in on artificial intelligence. The goal was to recruit machine learning and AI superstars to lead the department into this new frontier of technology, according to UW Computer Science Chair Hank Levy.

The challenge? Competing for experts with bigger names in academia, like Stanford and MIT.

But the University of Washington had an asset those institutions didnt: Jeff Bezos in its backyard.

So we thought, what the hell? Well send Jeff an email, Levy said Thursday, speaking at the Technlogy Alliances AI Policy Matters Summit in Seattle. The event brought together industry experts, scholars, and elected officials to discuss the state of AI and machine learning and recommend policies to govern the new technology.

UW was trying to recruit the married duo Carlos Guestrin and Emily Fox from Carnegie Mellon University and the University of Pennsylvania, respectively, as well as other tech experts.

Within 24 hours, Bezos responded with two $1 million professorship endowments for Fox and Guestrin. Bezos also stepped in to charm the scholars. He spent a half-hour with them in-person, which Levy said helped seal the deal.

Whether you like Jeff or not, hes very funny, he has the biggest laugh in the world, and hes incredibly impressive, Levy said. It had an impact.

That was the start of a new era for UWs computer science department, Levy said. The department made the front page of the New York Times Sunday business section for its AI efforts in 2012. Microsoft co-founder Paul Allen donated $40 million to create a new computer science department in his name, kicking off several multimillion-dollar rounds of donations from tech companies and leaders. And the new Paul G. Allen school allowed UW to triple its number of computer science majors.

This was the beginning of propelling our department to being one of the very best and best-known in the country, Levy said.

The recruiting efforts also paid unexpected dividends to Apple. Despite his Amazon endowment, the startup that Guestrin spun out of UW was eventually acquired by Apple for $200 million in 2016.

This was the biggest exit out of the department, Levy said. It was a really big deal. One of the reasons that its a big deal is Apple did not have much presence in the region at that point.

Apple is now planning to grow to 2,000 employees at its new Seattle campus in Amazons backyard. Guestrin certainly wont be the last superstar the two tech titans compete over.

The success of Guestrins startup, Turi, benefited UW with more than just prestige. On the eve of the Apple acquisition, Turi gave the UW computer science department a $1 million professor endowment, like the ones Amazon provided to lure Guestrin and Fox.

Remember it was Jeffs and Amazons professorships who helped us to recruit Carlos and Emily and now this company, as it was being acquired by Apple, gave us $1 million to create and another professor in AI, to help us hire the next person in this area, Levy said. So that was an incredible thing that they did for us.

View original post here:

How Jeff Bezos personally helped the University of Washington recruit its AI superstars - GeekWire

Latent Logic uses "imitation learning" to create simulations of human behavior which can be used in vehicle testing. Most AI training uses reinforcement learning, in which an AI gives answers to problems that are coded as either correct or incorrect. Over time, reinforcement-based AI can learn the correct answer more quickly.

However, this can be rather inefficient. By contrast, imitation learning has machines mimic human behaviors to learn some of the implicit knowledge that people have about the world, making it faster for the AI to model the optimal solution. Waymo could use this technique to train autonomous vehicles by having AI model complex human behaviors like cars cutting each other off or a pedestrian appearing in an unexpected location.

Latent Logic is based in Oxford, UK, which is something of a hub for self-driving vehicle research. For example, there's Oxbotica, a group which has trialed an autonomous grocery delivery vehicles, self-driving taxis and driverless shuttles. BAE Systems worked with researchers in Oxford to develop a hefty off-world autonomous vehicle based on a Bowler Wildcat. There's also the University of Oxford, which performs research into autonomous vehicles as well.

Acquiring the company gives Alphabet a foothold in a key location in the UK and access to a hub of local talent. "We see an exciting opportunity in Europe, not only in continuing to build our partnerships with major automakers but also in benefitting from the world-class technology and engineering capabilities in Oxford and beyond," Drago Anguelov, Waymo's principal scientist and head of research told The Guardian.

Waymo does not plan to launch self-driving car services in the UK yet, but the company has confirmed it has plans to operate in Europe in the future.

Here is the original post:

Waymo enters the UK with acquisition of self-driving AI startup Latent Logic - Engadget

AIT News Desk 13 Dec 2019 Computing, News Comments Off on Observe.AI Raises $26 Million to Digitally Transform the $300 Billion Voice Customer Service Market with Artificial Intelligence 52 Views

Company Also Announces a Relationship with Microsoft to Bring AI-Powered Coaching and Conversational Insights to Joint Customers

Observe.AI, the leader in AI-powered agent enablement for voice customer service, announced a $26 million Series A financing led by Scale Venture Partners, with participation from Nexus Venture Partners, Steadview Capital, 01 Advisors, and Emergent Ventures. This funding allows Observe.AI to expand its US-India team globally and accelerate product development.

In conjunction with the funding, Andy Vitus, partner at Scale, will be joining Observe.AIs board. This brings the companys total funding to $34 million.

Legacy speech analytics systems are simply not meeting the needs of the worlds top brands, said Swapnil Jain, CEO and co-founder of Observe.AI. Todays customer service agents have a unique ability to emotionally connect with customers and are often a brands only frontline representatives. This investment will fuel our mission to elevate agent performance through AI-based coaching and insights.

Many support teams monitor just 1-2 percent of calls and use three or more systems to access conversational insights and enable agents. Observe.AI uses the latest speech, natural language processing, and deep learning technologies to analyze 100 percent Observe.AI of customer conversations and provide adaptive coaching, including completely automating some parts of the quality assurance and compliance tracking processes. The platform becomes smarter with each call analysis.

Read More: ImmersiveTouch Launches New Personalized VR Imaging Platform into the Radiology Market

Observe.AI is already disrupting the $300 billion voice customer service market by rethinking how agents are coached and the way top brands provide personalized customer experiences, said Andy Vitus, Partner at Scale Venture Partners.

Observe.AI also announced that it has been accepted into the Microsoft for Startups program. With this relationship, Microsoft customers can leverage Observe.AIs platform through its Azure marketplace.

At Microsoft, were thrilled to see one of our Microsoft for Start-Up members excel Observe.AI as one of the fastest-growing startups in the Bay Area. Observe.AI continues to define how AI can transform the customer experience, impacting enterprise support teams to improve quality of service, agent performance, and productivity, said Shaloo Garg, Managing Director, Microsoft for Start-Ups.

Read More: AiThority Interview with Robert Cruz, Senior Director of Information Governance at Smarsh

In the past 12 months, Observe.AI has signed 100 customers and formed partnerships with leading organizations like Microsoft, Talkdesk, ERCBPO, and itelBPO. Some of the worlds largest enterprises and emerging brands use Observe.AI, including TripAdvisor, Concentrix, ClearMe, and Root Insurance. Thousands of global agents are coached with Observe.AI, which provides a detailed look at how top agents successfully structure calls so those tactics can be replicated.

We expect to see a 4X increase in annual recurring revenue in 2020, said Sharath Keshava Narayana, CRO of Observe.AI. With plans to significantly expand our sales, marketing, and customer success teams over the next few months, were both eager and grateful to build on the momentum.

Observe.AI is set to transform voice customer service for the AI era. We are delighted to have partnered with them from the early days of their journey, said Jishnu Bhattacharjee, Nexus Managing Director and Observe.AI board member.

Read More: OptimalPlus Opens German Office and Partners with ZF

Read more:

Observe.AI Raises $26 Million to the $300 Billion Voice Customer Service - AiThority

Intelligence is moving to the edge. As the growth of data acquisition accelerates, migration of compute closer to the edge allows more efficient use of data at reduced latency and infrastructure cost. Computer visionis a field of artificial intelligence that trainsmachinesto interpret and understand the visual world.

One of the goals ofcomputer visionis for machines to see and process images in the same way humans do. Computer vision through machine learning uses a method called supervised learning where a large annotated image set is used to construct a computational model. While model training can be intense and time-consuming, once trained, this model can quickly and effectively perform a variety of tasks.

Some examples of image and video analysis include identifying, classifying, counting and estimating pose of objects; compressing and encoding visual information for transmission or matching; estimating camera perspective; 2-D segmentation of foreground and background; estimating depth and performing 3-D segmentation; and inpainting or inferring visual data for occluded regions of an image.

With the recent technological advancement of edge AI, the ability to process and analyze data locally at the camera as opposed to streaming data to the cloud means that computer vision may be at the forefront of leveraging the wood pallet industry.

Three types of computer vision architecture exist: video or images that are sent to the cloud for computation; partial computing on the edge where only a few modalities are transferred to the cloud to search, sort and compute; and edge AI, which computes all image data on the edge. The latter requires training a model to be installed on the edge and updated at frequent intervals. These architectures have unique advantages and disadvantages. With the growing body of computer vision research, it is possible to find a model that is well suited to each application.

Pallets transport goods throughout the world and are an integral part of the supply chain logistics industry. The European Federation of Wooden Pallet and Packaging Manufacturers (FEFPEB) reported that more than 3 billion wooden pallets are in circulation in the EU, while nearly 2 billion wooden pallets are used each day in the U.S. The majority of these pallets are owned by large pooling solution companies (such as my company), which lease pallets from a shared pool, reducing complexity of pallets procurement, management and recovery for companies managing the supply chain of their products.

Pooling solution companies want to monitor pallet movement throughout the supply chain to understand pallet losses and recovery, pallet damage and pallet cycle time. To achieve comprehensive monitoring, each pallet could be labeled with a unique identifier or a tracking device. If a pallet has a unique ID such as a barcode or a QR code, computer vision can be used to track the pallet as it flows through the supply chain. If a pallet is instrumented with a tracking device, it can be detected through computer vision for device replacement or maintenance as the pallet flows through the sortation process in a service center.

Pallets are made from a wide variety of wood, including beech, ash, poplar, pine and spruce. The type of wood the pallets are made from forms the unique composition of wood tissue contours that not only provide insight into how strong and durable a single pallet can be, but can also provide a unique identifier for a pallet that can be used for tracking. Additionally, the footprint of nails that fasten the wooden boards with blocks form a topology that can provide additional information about the pallet life cycle, such as how long the pallet will last in the supply chain before it hits the repair belt.

Through well-trained computer vision models, the unique grain patterns of each pallet can be identified at birth, and this identity can be managed as the pallet flows through its life cycle. Changes to the wood patterns and structure can be tracked as the pallet is damaged and repaired as it cycles through the supply chain. Computer vision will not only allow tracking of pallets by just images, but it will also give insights into pallet strength and durability. This low-cost solution equips pallet companies to take action to filter unreliable pallets at birth or after repeated use. Furthermore, pallet logistics companies can gather insights into the number of damages and types of damages by customer and industry verticals to enhance the business model and improve the design to make the platforms more rugged.

With 95% of organizations and institutions reporting their continued use, wooden pallets continue to dominate the supply chain market. Wood is the only material that is 100% renewable, recyclable, reusable and rated for hygienic transport several classes of goods.

Computer vision can also play a crucial role in maintaining an accurate inventory count of pallets in any warehouse. From an image that contains one or more stacks of pallets, a well-trained neural network can produce a count of all pallets in less time and with more accuracy than a well-trained eye.

Beyond supply chain KPIs, computer vision can be applied to both worker safety and efficiency within the service centers where pallets are stored, inspected and repaired. Any repetitive task, such as a nailing activity on a repair bench, can be fed into a model to identify worker accuracy, fatigue and many more actions from a live video stream. Additionally, human detection can be used to identify when people might be present in unauthorized or hazardous areas of a facility and if they are wearing appropriate PPE. This can help prevent workplace accidents, which is the highest priority in this business.

Excerpt from:

Confluence Of AI On The Edge And Computer Vision In The Wood Pallets Industry - Forbes

Photo: Artem Peretiatko | Getty Images

Why We Need More Women in STEM and How AI Could Help Us Get There

Recently, Dr. France A. Crdova, director of the National Science Foundation, gave a presentation at the U.S. Council on Competitiveness meeting in Washington, D.C. She holds an extraordinary record of accomplishment and has made a tremendous impact on academia and the U.S.'s scientific community. Crdova is also the youngest person and first woman to serve as Chief Scientist at NASA. Her journey began with her love for STEM.

In some ways, the future of workis largely linked to STEM. Yet in this day and age, despite role models like Crdova, women continue to remain significantly underrepresented across the board in this industry. Whats more, the inevitable reality of an AI-integrated workforce is coming.

According to recent research, women make up just 26 percentof those who hold computer- and math-related jobs. Moreover, data from UNESCO indicates that only 35 percentof women go into STEM, of which a mere 3 percentdecided to pursue fields like IT. And when it comes to computer science degrees in the U.S.,only 18 percentofthem are earned by female college graduates.

Interestingly, in Eastern Europe, more women tend to pursue STEM. About 74 percent of women occupy medical professions in countries like Estonia and Latvia. In Bulgaria, 27 percentof IT workers are women a nine-fold lead over the U.S. AndEastern European countries boast the highest proportion of women who work in high-tech companies.

These results are partly a product of socialist-era policies that encouragedfemales to pursue the maths and sciences in the name of the States advancement. American can learn from the policy initiative of driving the populace to pursue STEM via large-scale campaigns for societal betterment.

Though Eastern Europe trumps America when it comes to women in STEM, many females who do choose to explore the industry report experiencing gender discrimination. Results from a survey of 1,000 college-aged women conducted by Girls Who Code suggested that "half [of the women] had either had a negative experience applying for a job in tech or know a woman who has. Furthermore, of the survey respondents that reported a negative encounter, 25 percentsaid their interviewers focused on their personal attributes rather than their skills and 21 percentof women said they encountered biased questions.

In light of this pressing issue, female-centric STEM initiatives has appeared across the US. Among the best known national programs include the previously mentioned Girls Who Code organization, as well as Kode With Klossy, run by former supermodel Karlie Kloss. And though specialized STEM programs for girls are a step in the right direction, we need to make a leap. Current efforts arent nearly comprehensive enough to adequately prepare women for an AI-augmented reality and work towards solving the problem of discrimination and the gender gap.

We can take advantage of the inescapable marriage of technology and biology to craft a novel multi-part solution for helping solve the discrimination and STEM gender gap:

Women make up just under half (47 percent) of the workforce, but they are 58 percent of workers at the highest risk of automation, states a recent report by the Institute for Womens Policy Research. Therefore, the digital workforce revolution could drive some women to encounter high levels of job insecurity.

But just how much of a threat does automation pose? According to the U.S.Census, between 2006 and 2010, 96 percentof secretarial and administrative positions were occupied by women. Females also reportedly hold77 percentof teaching positions and 78 percentof central-office administrator roles.

Learning from eastern Europe, America can introduce STEM awareness campaigns, large-scale private-public initiatives through whichthe government, academic and private institutions work in tandem to educate the public about STEM. Oneexample of a current STEM awareness initiative is STEMFuture, an international non-profit organization that provides education and workshops for adolescents to encourage careers in technology, mathand science.

STEM awareness campaigns have the potential to significantly lessen the strain of automation on womenand deliver a new set of opportunities and benefits to the female workforce of tomorrow.

Scientific research suggeststhe female brain matures faster than the male brainand possesses unique structural attributes. For example, girls tend to have stronger neural networks in the temporal lobe, leading to better memorization and listening abilities. In addition, the corpus callosum (a weave of fibers that conjoin the left and right hemispheres of the brain), can be up to 25 percentlarger in developing female adolescents than in their male counterparts.

Currently, schools teach boys and girls at an equivalent pace, neglecting their separate biological needs. If educators take advantage of development differences, special STEM curriculums could be crafted for girls at an early age. This could help bring STEM to girls across classrooms in the U.S. and encourage them to explore the field more deeply.

Using AI to improve the HR hiring process isnt news. Many companies use AI recruitment software that aims to make hiring more efficient or cut down on bias. This innovation suggests that AI will remain central to the future workplace environment.

According to a Pew Research report, about four in ten working women (42 percent) in the United States say they have faced discrimination on the job because of their gender. Moreover, another another study by Pew suggests that 50 percentof women in STEM jobs have experienced gender discrimination. Carefully-vetted AI couldhelp decrease gender bias discrimination in STEM by exclusively assessing candidates based on skills.

A digital society is a dynamic one. In the future, new technologies will regularly enter the marketplace, continuing to make lifelong learning necessary. A skills-based economy means that degrees and hierarchies will no longer be as relevant. When abilities are prioritized above factors like gender, more women could feel empowered toenter the STEM industry, knowing they'd be less likely to be assessed on the basis of gender.

Related:

Originally posted here:

Why We Need More Women in STEM and How AI Could Help Us Get There - Thehour.com

The evolution from research to product

Its one thing for a Microsoft researcher to use all the available bells and whistles, plus Azures powerful computing infrastructure, to develop an AI-based machine translation model that can perform as well as a person on a narrow research benchmark with lots of data. Its quite another to make that model work in a commercial product.

To tackle the human parity challenge, three research teams used deep neural networks and applied other cutting-edge training techniques that mimic the way people might approach a problem to provide more fluent and accurate translations. Those included translating sentences back and forth between English and Chinese and comparing results, as well as repeating the same translation over and over until its quality improves.

In the beginning, we were not taking into account whether this technology was shippable as a product. We were just asking ourselves if we took everything in the kitchen sink and threw it at the problem, how good could it get? Menezes said. So we came up with this research system that was very big, very slow and very expensive just to push the limits of achieving human parity.

Since then, our goal has been to figure out how we can bring this level of quality or as close to this level of quality as possible into our production API, Menezes said.

Someone using Microsoft Translator types in a sentence and expects a translation in milliseconds, Menezes said. So the team needed to figure out how to make its big, complicated research model much leaner and faster. But as they were working to shrink the research system algorithmically, they also had to broaden its reach exponentially not just training it on news articles but on anything from handbooks and recipes to encyclopedia entries.

To accomplish this, the team employed a technique called knowledge distillation, which involves creating a lightweight student model that learns from translations generated by the teacher model with all the bells and whistles, rather than the massive amounts of raw parallel data that machine translation systems are generally trained on. The goal is to engineer the student model to be much faster and less complex than its teacher, while still retaining most of the quality.

In one example, the team found that the student model could use a simplified decoding algorithm to select the best translated word at each step, rather than the usual method of searching through a huge space of possible translations.

The researchers also developed a different approach to dual learning, which takes advantage of round trip translation checks. For example, if a person learning Japanese wants to check and see if a letter she wrote to an overseas friend is accurate, she might run the letter back through an English translator to see if it makes sense. Machine learning algorithms can also learn from this approach.

In the research model, the team used dual learning to improve the models output. In the production model, the team used dual learning to clean the data that the student learned from, essentially throwing out sentence pairs that represented inaccurate or confusing translations, Menezes said. That preserved a lot of the techniques benefit without requiring as much computing.

With lots of trial and error and engineering, the team developed a recipe that allowed the machine translation student model which is simple enough to operate in a cloud API to deliver real-time results that are nearly as accurate as the more complex teacher, Menezes said.

In the rapidly evolving AI landscape, where new language understanding models are constantly introduced and improved upon by others in the research community, Bings search experts are always on the hunt for new and promising techniques. Unlike the old days, in which people might type in a keyword and click through a list of links to get to the information theyre looking for, users today increasingly search by asking a question How much would the Mona Lisa cost? or Which spider bites are dangerous? and expect the answer to bubble up to the top.

This is really about giving the customers the right information and saving them time, said Rangan Majumder, partner group program manager of search and AI in Bing. We are expected to do the work on their behalf by picking the most authoritative websites and extracting the parts of the website that actually shows the answer to their question.

To do this, not only does an AI model have to pick the most trustworthy documents, but it also has to develop an understanding of the content within each document, which requires proficiency in any number of language understanding tasks.

Last June, Microsoft researchers were the first to develop a machine learning model that surpassed the estimate for human performance on the General Language Understanding Evaluation (GLUE) benchmark, which measures mastery of nine different language understanding tasks ranging from sentiment analysis to text similarity and question answering. Their Multi-Task Deep Neural Network (MT-DNN) solution employed both knowledge distillation and multi-task learning, which allows the same model to train on and learn from multiple tasks at once and to apply knowledge gained in one area to others.

Bings experts this fall incorporated core principles from that research into their own machine learning model, which they estimate has improved answers in up to 26 percent of all questions sent to Bing in English markets. It also improved caption generation or the links and descriptions lower down on the page in 20 percent of those queries. Multi-task deep learning led to some of the largest improvements in Bing question answering and captions, which have traditionally been done independently, by using a single model to perform both.

For instance, the new model can answer the question How much does the Mona Lisa cost? with a bolded numerical estimate: $830 million. In the answer below, it first has to know that the word cost is looking for a number, but it also has to understand the context within the answer to pick todays estimate over the older value of $100 million in 1962. Through multi-task training, the Bing team built a single model that selects the best answer, whether it should trigger and which exact words to bold.

Earlier this year, Bing engineers open sourced their code to pretrain large language representations on Azure. Building on that same code, Bing engineers working on Project Turing developed their own neural language representation, a general language understanding model that is pretrained to understand key principles of language and is reusable for other downstream tasks. It masters these by learning how to fill in the blanks when words are removed from sentences, similar to the popular childrens game Mad Libs.

You take a Wikipedia document, remove a phrase and the model has to learn to predict what phrase should go in the gap only by the words around it, Majumder said. And by doing that its learning about syntax, semantics and sometimes even knowledge. This approach blows other things out of the water because when you fine tune it for a specific task, its already learned a lot of the basic nuances about language.

To teach the pretrained model how to tackle question answering and caption generation, the Bing team applied the multi-task learning approach developed by Microsoft Research to fine tune the model on multiple tasks at once. When a model learns something useful from one task, it can apply those learnings to the other areas, said Jianfeng Gao, partner research manager in the Deep Learning Group at Microsoft Research.

For example, he said, when a person learns to ride a bike, she has to master balance, which is also a useful skill in skiing. Relying on those lessons from bicycling can make it easier and faster to learn how to ski, as compared with someone who hasnt had that experience, he said.

In some sense, were borrowing from the way human beings work. As you accumulate more and more experience in life, when you face a new task you can draw from all the information youve learned in other situations and apply them, Gao said.

Like the Microsoft Translator team, the Bing team also used knowledge distillation to convert their large and complex model into a leaner model that is fast and cost-effective enough to work in a commercial product.

And now, that same AI model working in Microsoft Search in Bing is being used to improve question answering when people search for information within their own company. If an employee types a question like Can I bring a dog to work? into the companys intranet, the new model can recognize that a dog is a pet and pull up the companys pet policy for that employee even if the word dog never appears in that text. And it can surface a direct answer to the question.

Just like we can get answers for Bing searches from the public web, we can use that same model to understand a question you might have sitting at your desk at work and read through your enterprise documents and give you the answer, Majumder said.

Top image: Microsoft investments in natural language understanding research are improving the way Bing answers search questions like How much does the Mona Lisa cost? Image by Muse du Louvre/Wikimedia Commons.

Related:

Jennifer Langston writes about Microsoft research and innovation. Follow her on Twitter.

Read the original:

From search to translation, AI research is improving Microsoft products - Microsoft

Qualitest, the worlds largest software testing and quality assurance company, has acquired AI and machine learning company AlgoTrace for an undisclosed amount. This acquisition marks the first step of Qualitests growth strategy following an investment from Bridgepoint earlier this year.

The acquisition will allow Qualitest to radically expand the number of AI-powered testing solutions available to clients, as well as develop its capabilities in assisting companies test and launch new AI-powered solutions with greater confidence and speed. As software grows in complexity and the pressure to launch faster and more frequently increases, according to Gartner, companies that do not use AI to enhance their Quality Assurance will be at a significant disadvantage.

AlgoTraces machine learning tools help brands answer business critical questions as they launch new software: what, where, when, and how to test and in what order to ensure consistently high quality. With multiple clients already using Qualitests suite of AI-testing tools, this expansion of capabilities creates opportunities not only for new Qualitest clients, but also allows for the growth of existing relationships with current customers around the world.

Read More: Higher Adoption of Emerging Technologies in Commercial Vehicles Stoke OEM Collaborations with

Qualitest began working with the AlgoTrace team more than a year ago, with AlgoTraces AI platform powering Qualitests market-leading test predictor tool, which applies pioneering autonomous AI capabilities and predictive modeling to unstructured data without the need for code or complex interfaces. Following multiple successful joint projects, the teams saw that, together, they would be able to apply AlgoTraces powerful prediction engine in a variety of ways across the software development lifecycle to improve quality and speed to market.

Qualitests AI-testing solutions have two main features focused on increasing confidence and assurance. First, to assist and enhance quality assurance efforts giving brands, in a more rapid fashion, high levels of confidence that software releases will go smoothly. Second, helping companies who are using AI in their own offerings to have a higher level of confidence that their AI algorithms are generating correct, unbiased results.

Ron Ritter, CEO at AlgoTrace, said: We are thrilled to be joining with Qualitest. Following successful implementations with the company in the past, we have complete faith that we will help Qualitest change the testing paradigm forever enhancing their quality engineering with machine learning. While there is a lot of hype surrounding AI, were deploying real, hard-nosed and practical tools that significantly change the rules.

Read More: AiThority Interview with Jason Braverman, Chief Technology Officer at SkyX Systems Corp.

Norm Merritt, CEO of Qualitest, said:Applying AI to quality engineering is a perfect fit. Just as software becomes increasingly complex, the companies producing it are under competitive pressure to increase the speed and frequency of their rollouts. AI is the only way companies can scale software testing and quality engineering and the AlgoTrace team have shown that they understand this. In our view, companies that do not use AI to improve quality will be at a significant disadvantage.

Aviram Shotten, Chief Knowledge and Innovation Officer at Qualitest, said: Ron and his team are just the kind of innovators we love: smart, customer-obsessed and attacking a big market problem with cutting edge technology. This acquisition will not only help us accelerate AI adoption within quality engineering by providing a holistic solution to our clients, it provides an avenue for our teams to access AlgoTraces unique expertise to build new models, tools and solutions to improve how technology is developed, tested and deployed.

Read More: ImmersiveTouch Launches New Personalized VR Imaging Platform into the Radiology Market

See the original post here:

Qualitest Acquires AI and ML Company AlgoTrace to Expand Its Offering - AiThority

By Julien Legrand, Operation Security Manager, Socit Gnrale

Artificial intelligence is a double-edged sword that can be used as a security solution or as a weapon by hackers. AI entails developing programs and systems capable of exhibiting traits associated with human behaviors. The characteristics include the ability to adapt to a particular environment or to intelligently respond to a situation. AI technologies have extensively been applied in cybersecurity solutions, but hackers are also leveraging them to develop intelligent malware programs and execute stealth attacks.

Security experts have conducted a lot of research to harness the capabilities of AI and incorporate it into security solutions. AI-enabled security tools and products can detect and respond to cybersecurity incidents with minimal or zero input from humans. AI applications in cybersecurity have proved to be highly useful. Twenty-five percent of IT decision-makers attribute security as the primary reason why they adopt AI and machine learning in organizational cybersecurity. AI not only improves security posture, but it also automates detection and response processes. This cuts on the finances and time used in human-driven intervention and detection processes.

Organizations use AI to model and monitor the behavior of system users. The purpose of monitoring the interactions between a system and users is to identify takeover attacks. These are attacks where malicious employees steal login details of other users and use their accounts to commit different types of cybercrimes. AI learns the user activities over time such that it considers unusual behavior as anomalies. Whenever a different user uses the account, AI-powered systems can detect the unusual activity patterns and respond either by locking out the user or immediately alert system admins of the changes.

Antivirus tools with AI capabilities detect network or system anomalies by identifying programs exhibiting unusual behavior. Malware programs are coded to execute functions that differ from standard computer operations. AI antiviruses leverage machine learning tactics to learn how legitimate programs interact with an operating system. As such, whenever malware programs are introduced to a network, AI antivirus solutions can immediately detect them and block them from accessing systems resources. This contrasts from signature-based traditional antiviruses which scans a signature database to determine whether a program is a security threat.

Automated analysis of system or network data ensures continuous monitoring for prompt identification of attempted intrusions. Manual analysis is nearly impossible due to the sheer volume of data generated by user activities. Cybercriminals use command and control (C2) tactics to penetrate network defenses without being detected. Such tactics include embedding data in DNS requests to bypass firewalls and IDS/IPS. AI-enabled cyber defenses utilize anomaly detection, keyword matching, and monitoring statistics. As a result, they can detect all types of network or system intrusion.

Cybercriminals prefer email communication as the primary delivery technique for malicious links and attachments used to conduct phishing attacks. Symantec states that 54.6 percent of received email messages are spam and may contain malicious attachments or links. Anti-phishing emails with AI and machine learning capabilities are highly effective in identifying phishing emails. This is by performing in-depth inspections on links. Additionally, such anti-phishing tools simulate clicks on sent links to detect phishing signs. They also apply anomaly detection techniques to identify suspicious activities in all features of the sender. These include attachments, links, message bodies, among other items.

Hackers are turning to AI and using it to weaponize malware and attacks to counter the advancements made in cybersecurity solutions. For instance, criminals use AI to conceal malicious codes in benign applications.They program the codes to execute at a specific time, say ten months after the applications have been installed, or when a targeted number of users have subscribed to the applications. This is to maximize the impacts such attacks will cause. Concealing such codes and information requires the application of AI models and deriving private keys to control the place and time the malware will execute.

Notwithstanding, hackers can predefine an application feature as an AI trigger for executing cyber-attacks. The features can range from authenticating processes through voice or visual recognition to identity management features. Most applications used today contain such features, and this provides attackers with ample opportunities of feeding weaponized AI models, deriving a key, and attacking at will. The malicious models can be present for years without detection as hackers wait to strike when applications are most vulnerable.

Besides, AI technologies are unique in that they acquire knowledge and intelligence to adapt accordingly. Hackers are aware of these capabilities and leverage them to model adaptable attacks and create intelligent malware programs. Therefore, during attacks, the programs can collect knowledge of what prevented the attacks from being successful and retain what proved to be useful. AI-based attacks may not succeed in a first attempt, but adaptability abilities can enable hackers to succeed in subsequent attacks. Security communities thus need to gain in-depth knowledge of the techniques used to develop AI-powered attacks to create effective mitigations and controls.

Also, cyber adversaries use AI to execute intelligent attacks that self-propagate over a system or network. Smart malware can exploit unmitigated vulnerabilities leading to an increased likelihood of fully compromised targets. If an intelligent attack comes across a patched vulnerability, it immediately adapts to try compromising a system through different types of attacks.

Lastly, hackers use AI technologies to create malware capable of mimicking trusted system components. This is to improve stealth attacks. For example, cyber actors use AI-enabled malware programs to automatically learn the computation environment of an organization, patch update lifecycle, preferred communication protocols, and when the systems are least protected. Subsequently, hackers can execute undetectable attacks as they blend with an organizations security environment. For example, TaskRabbit was hacked compromising 3.75 million users, yet investigations could not trace the attack. Stealth attacks are dangerous since hackers can penetrate and leave a system at will. AI facilitates such attacks, and the technology will only lead to the creation of faster and more intelligent attacks.

Disclaimer: CISO MAG does not endorse any of the claims made by the writer. The facts, opinions, and language in the article do not reflect the views of CISO MAG and CISO MAG does not assume any responsibility or liability for the same. Views expressed in this article are personal.

Here is the original post:

Artificial Intelligence as Security Solution and Weaponization by Hackers - CISO MAG

1. Facial recognition and AI in video interviewsA number of video interview platforms available on the market today focus on scheduling a Q&A with a candidate, recording a clip of the exchange, and forwarding it to the HR team for assessment.

Recruitment tech specialist HireVue, however, takes candidate screening to the next level by combining facial analysis with AI.

The software behind the platform purportedly relies on 25,000 data points taken from the facial expressions, movements and tone of voice of past successful candidates then uses them as a benchmark for screening new applicants. These subtle clues from their interaction with the AI reportedly help determine their suitability to the role.

2. Reducing unconscious bias in candidate interviewsAmid efforts to build a more diverse workforce, organizations still struggle against the influence of unconscious bias in hiring. Stockholm-based tech firm Furhat Robotics, however, is working to reduce the impact of human bias on candidate screening all with the help of the robotic interviewer Tengai.

Unlike human recruiters, the robot skips the small talk and goes through the questions in the same manner, tone and order. The developers hope this approach leads to a fairer recruitment process.

Read more:

Five ways AI changed the workplace in 2019 - Human Resources Director