Google has kept its education initiatives, ranging from Chromebooks for schools to teaching software, within the confines of the classroom. "Bolo" (Hindi for "speak") is essentially a reading assistant for elementary school children that harnesses Google's speech recognition and text-to-speech smarts. Kids can read through a mix of Hindi and English stories with the help of an in-app learning assistant, who appears as a young girl named "Diya." She offers tips and corrections and can even translate Hindi words into English. The app supports multiple users and works offline in order to avoid connectivity issues in rural areas -- while Google works on getting those regions online through WiFi hotspots in remote train stations.

The utility of learning a dynamics/world model of the environment in reinforcement learning has been shown in a many ways. When using neural networks, however, these models suffer catastrophic forgetting when learned in a lifelong or continual fashion. Current solutions to the continual learning problem require experience to be segmented and labeled as discrete tasks, however, in continuous experience it is generally unclear what a sufficient segmentation of tasks would be. Here we propose a method to continually learn these internal world models through the interleaving of internally generated rollouts from past experiences (i.e., pseudo-rehearsal). We show this method can sequentially learn unsupervised temporal prediction, without task labels, in a disparate set of Atari games. Empirically, this interleaving of the internally generated rollouts with the external environment's observations leads to an average 4.5x reduction in temporal prediction loss compared to non-interleaved learning. Similarly, we show that the representations of this internal model remain stable across learned environments. Here, an agent trained using an initial version of the internal model can perform equally well when using a subsequent version that has successfully incorporated experience from multiple new environments.

To me, the answer is simple. Companies like NVIDIA represent the future state of Artificial Intelligence, not the current state of AI. The reason for that divide is simple – the most important word in AI technology is: "democratization." GPUs are expensive, especially compared to their mainstream counterpart, the Central Processing Unit. GPUs are not mainstream, whereas the CPU is found in nearly every computer on the market today.

Let's now look at a concrete example. The problem is to predict math, reading and writing grades for high-school students in the U.S. We are given historical data that include features like -- gender, race/ethnicity (which is anonymized), parent level of education, whether the student ate a standard/free/subsidized lunch and the level of preparation for tests. Given this data, I trained a multi-class random forest model [source code]. In order to explain what the model has learned, one of the simplest techniques is to look at the relative feature importance. Feature importance measures how big an impact a given feature has on predicting the outcome.

Along with the proliferation of personalized learning platforms also comes the rise of the presence of artificial intelligence in educational technology. Before going further, we must first acknowledge the truth about the term artificial intelligence -- it is a buzzword that connotes many different applications. It holds different meaning in different contexts and, much of the time, is used indiscriminately without clear boundaries or specifications. To try and avoid just adding to the noise, we will focus on three main topics: (1) deconstructing the buzzword into its component parts, (2) characterizing current use cases and predicting future ones, and (3) exploring risks and trade-offs. First, let us define, as clearly as possible, what all is under consideration when we mention the term.

Earlier this year, the UK Education Secretary called for the IT industry to work with educators to make "smarter use" of technologies, such as artificial intelligence (AI), to cut teachers' burgeoning workloads. Speaking at the 2019 BETT educational IT show in London, Damian Hinds said: "Education is one of the few sectors where technology has been associated with an increase in workload rather than the reverse." But he added that, if used wisely, it could also reduce the amount of time educators had to spend on non-teaching tasks, such as lesson planning, marking and general admin. Hinds cited the instance of Bolton College, which has deployed IBM's Watson AI programme as a virtual clerk called Ada. Ada, which can answer natural language questions, provides about 14,000 students with personalised learning assessments and handles queries about the curriculum and attendance issues, both of which teachers would previously have had to do in their own time.

Training employees how to perform specific tasks isn't difficult, but building their soft skills -- their interactions with management, fellow employees, and customers -- can be more challenging, particularly if there aren't people around to practice with. Virtual reality training company Talespin announced today that it is leveraging AI to tackle that challenge, using a new "virtual human platform" to create realistic simulations for employee training purposes. Unlike traditional employee training, which might consist of passively watching a video or lightly interacting with collections of canned multiple choice questions, Talespin's system has a trainee interact with a virtual human powered by AI, speech recognition, and natural language processing. Because the interactions use VR headsets and controllers, the hardware can track a trainee's gaze, body movement, and facial expressions during the session. Talespin's virtual character is able to converse realistically, guiding trainees through branching narratives using natural mannerisms and believable speech.

Andreas Schleicher is a German data scientist--tall and precise with a grey mustache and a steely gaze. The head of the education division at the Organisation for Economic Cooperation and Development (OECD), he gives off an impression of determined focus. That's useful, considering that he's on a mission to change the way countries around the world teach their children. Society, according to Schleicher, is preparing for the future of work all wrong. We're scared that human jobs will be replaced by robots. But we're still teaching kids to think like machines. "What we know is that the kinds of things that are easy to teach, and maybe easy to test, are precisely the kinds of things that are easy to digitize and to automate," Schleicher said at the LearnIt conference in London in January. It's fairly easy to teach and test math, for example--but robots happen to be pretty good at math, too.

All our knowledge is about the past, but all our strategic decisions are about the future Conway 2003 What we don't know we don't know about the future What we know What we know we don't know 22. While 63% of CEOs believe AI will have a larger impact on the world than the internet,…. Boards overestimate their digital savviness 62% of boards report they are digitally savvy Source: MIT CISR 2014 Board Survey, 81 companies. Digital Savvy Boards - Higher performance • Companies with digitally savvy boards • 38% higher revenue growth • 34% higher ROA • 34% higher market growth • 3 digitally savvy directors required to impact performance Source: MIT Research 2018 26. A single tech savvy director in the boardroom risks being lonely and misunderstood.

Inspired by Google Interview University. This is my multi-month study plan for going from mobile developer (self-taught, no CS degree) to machine learning engineer. My main goal was to find an approach to studying Machine Learning that is mainly hands-on and abstracts most of the Math for the beginner. This approach is unconventional because it's the top-down and results-first approach designed for software engineers. Please, feel free to make any contributions you feel will make it better.