It's official: if you're using a voice-assistant -- pretty much any voice-assistant -- someone could be listening in. The Guardian reports that Apple has joined an ever-growing list of tech companies that listens in on commands uttered through its virtual voice-assistant. Snippets of audio, reports The Guardian, are sent to contractors who are responsible for listening and grading them for accuracy, including whether or not the command was accidental or whether its assistant, Siri, was able to complete the task. As is the case with other similar programs from Google and Amazon, however, a whistleblower says the program has inadvertently swept up audio data that most might find confidential. Those include, according to an unnamed source in the report, conversations between patients and doctors, sex, criminal activity, and official business talk.

Google will give away 100,000 of its smart home device to those living with paralysis according to a blog post. The company announced that it will be sending its Google Home mini to people with physical disabilities in an effort to help improve their lives at home. According to Garrison Redd, an ambassador for the Christopher & Dana Reeve Foundation, who has been unable to walk for the past 20 years, the devices can make a huge difference for someone who relies on a wheelchair to move. 'When you're paralyzed, your home goes from being a place of comfort and security to a reminder of what you've lost,' wrote Redd in a blog post for Google. 'Light switches and thermostats are usually too high up on the wall and, if my phone falls on the floor, I may not be able to call a friend or family member if I need help.

Many of today's consumer-focused AI are fashioned after women: Siri, Alexa, Cortana, just to name a few. Female voices are considered friendlier and more helpful, but consumer AI is also in a subservient role. In reality, women are adding so much more to the field of AI than their voices. Today, though only 18 percent of C-level positions in the artificial intelligence/machine learning space are filled by women, which is not an usual figure in the tech sphere, but that does not take into account all the influential women building companies centered around AI. According to some studies, only 13.5 percent of the machine learning field and fewer than 10 percent in AI are women, and while it's important to acknowledge the gender gap, one must also acknowledge the many chief scientists and researchers making great strides in machine learning and AI.

It can't be overstated how fundamentally different this paradigm for music curation is from what you're used to. To compare it to another example from around your time, Spotify's Daily Drive playlist wove audio snippets from news talk shows with personalized music recommendations. I recall the feature was heralded as innovative for combining multiple audio formats into a single interface, but it was still fundamentally limited in how it relied on metadata around past listening activity. In contrast, the music information retrieval (MIR) techniques used in YouNite draw on real-time and forward-looking predictions around both present physiological states and desired future emotional outcomes. Hope this all makes sense?

In academic literature, recommender systems are often evaluated on the task of next-item prediction. The procedure aims to give an answer to the question: "Given the natural sequence of user-item interactions up to time t, can we predict which item the user will interact with at time t+1?". Evaluation results obtained through said methodology are then used as a proxy to predict which system will perform better in an online setting. The online setting, however, poses a subtly different question: "Given the natural sequence of user-item interactions up to time t, can we get the user to interact with a recommended item at time t+1?". From a causal perspective, the system performs an intervention, and we want to measure its effect. Next-item prediction is often used as a fall-back objective when information about interventions and their effects (shown recommendations and whether they received a click) is unavailable. When this type of data is available, however, it can provide great value for reliably estimating online recommender system performance. Through a series of simulated experiments with the RecoGym environment, we show where traditional offline evaluation schemes fall short. Additionally, we show how so-called bandit feedback can be exploited for effective offline evaluation that more accurately reflects online performance.

There is hardly a single hour that passes by when we are untouched by technology. We are constantly engaged with our mobile devices such as smartphones or tablets, except for when we are asleep. Whether it is work or going about our daily chores, technology has found its way into every aspect of our lives. Contributing to this plot, companies are gearing up their operations to adopt enterprise mobility management to enhance the productivity and efficiency of their employees. Enterprise Mobility enables the company to encourage their employees to work anytime, from anywhere with access to data using their mobile devices.

With a new feature, Tinder says it wants to make the swiping experience safer for its LGBTQ users traveling and living in certain countries. On Wednesday, the dating app introduced a new safety update dubbed "Traveler Alert" that will warn users who have identified themselves as lesbian, gay, bisexual, transgender and/or queer when they enter a country that could criminalize them for being out. The app plans to use the locations from users' devices to determine if there is a threat to the user's safety, where users can opt to have their profile hidden during their stay or make their profile public again. The caveat being that if a user decides to have their profile public, their sexual preference or gender identity will no longer be disclosed on the app until they return to a location where the user is deemed safer to disclose their identity. In the statement, Tinder says they developed the feature so that users "can take extra caution and do not unknowingly place themselves in danger for simply being themselves."

It's not every day that an 18-year-old college student catches the eye of the computing world, but when Ewin Tang took aim at recommendation algorithms similar to those commonly used by the likes of Amazon and Netflix, the University of Texas at Austin mathematics and computer science undergraduate blew up an established belief: that classical computers cannot perform these types of calculations at the speed of quantum computers. In a July 2018 paper, which Tang wrote for a senior honors thesis under the supervision of computer science professor Scott Aaronson, a leading researcher in quantum computing algorithms, she discovered an algorithm that showed classical computers can indeed tackle predictive recommendations at a speed previously thought possible only with quantum computers. "I actually set out to demonstrate that quantum machine learning algorithms are faster," she explains. "But, along the way, I realized this was not the case." Ewin Tang set out to show that quantum machine learning algorithms are faster than classical algorithms, "but ... I realized this was not the case."

Rusty Langford has been spending a lot of time lately helping companies introduce intelligent bots within their customer service centers. So he's learned a lot about when bots actually help customers and when they annoy them. Langford, who's spent more than 25 years in customer service and is the vice president of client services at Harte Hanks, says the first step to exploring bots is understanding what your customers typically need help with, and how much human interaction it takes to provide that support. Look for highly repetitive, easily defined problems, he advises--scenarios in which service center teams generally know what the conversation needs to look and feel like, and what the outcome is likely to be. "Those are safe ways for companies to put their toes in the water and begin to understand what bots can actually do," Langford says.

Personalization is a mission-critical feature of effective marketing, as studies show that a personalized journey leads to increased customer engagement and long-term loyalty. Netflix movie recommendations, Spotify music suggestions and special promotions on Amazon demonstrate that personalized content is not only becoming the norm but a consumer expectation. Businesses are accomplishing this task using machine learning, which is becoming the essential go-to tool in content personalization. Evergage, Monetate, Certona, Dynamic Yield and a number of other personalization engine vendors offer this functionality and are increasingly in demand. Gartner's "Magic Quadrant for Personalization Engines" 2019 report shows that personalization engine adoption is up 28% since 2016.