Bank of America debuted a virtual assistant bot today at Money2020, a fintech conference being held this week in Las Vegas. Named Erica, the bot uses artificial intelligence and predictive analytics to learn your personal spending habits and offer helpful advice. The bot will be available by voice command or plain text in Bank of America smartphone apps next year, according to CNBC. Erica is designed to be not just a virtual assistant but each customer's "personal advocate," said Bank of America head of digital banking Michelle Moore. It can tell you about your spending habits, notice if you spend more than usual on a certain product or category of products, present opportunities to reduce debt or save money, and alert you if your credit score dips.

Figuring out how much a business trip will cost can be a hassle. Employees are often put in the position of figuring out how to get somewhere, and once they get there, they don't have an incentive to keep their spending to a minimum. A new startup called TravelBank aims to help by providing a predictive budget for a trip, based on when an employee is traveling and the destination. After that, it helps the employee to document spending and file an expense report that shows how much money was spent against budget. More than just helping employees capture expenses, TravelBank is focused on helping to change their behavior so they spend less money.

Darknet is one of the most interesting and thought provoking sci fi books that I have read in awhile. As someone who is deeply immersed in the fields of machine learning and artificial intelligence, I have come across or thought about many of the ideas that have been presented in this book, especially the central theme – autonomous agents that aided by our increasingly digital online worlds, become capable enough to run their own corporate entities. Technology is not quite there yet for a creation of such an agent, but it's probably much closer than most people realize. That's why it was really interesting to go through the intellectual exercise of imagining what kind of things would such an entity engage in if it comes to be. For that reason alone Darknet is very worthwhile read for all AI geeks out there.

Applied Materials' (NASDAQ:AMAT) business strategy that enables major technological inflections for customers has earned it the top spot in many of its markets served, resulting in increased sales and EPS since 2013. With that, management has set lofty targets to reach by 2018, including EPS of 2.00, though we think that may be stretching it a bit. Applied is currently working through what it sees as the second phase of WFE drivers in mobile and social media. While it expects such macro trends to provide solid growth, even more robust proliferation of its chips is expected to come with the next wave of visual computing and artificial intelligence. Applications include virtual/augmented reality, the Internet of Things, big data, artificial intelligence, smart vehicles, and additive manufacturing.

When it comes to machine learning frameworks, Google's Tensorflow is clearly the most popular option right now, but with CNTK, Microsoft also released its own internal framework at the beginning of the year. The company is launching the first beta of the next version (2.0) of CNTK today and with it, it hopes to challenge Tensorflow's leadership position. CNTK used to stand for'Computational Network Toolkit' but the software has now been renamed to Microsoft Cognitive Toolkit instead. Xuedong Huang, Microsoft's Chief Speech Scientist, told me that he believes CNTK/Cognitive Toolkit has always had plenty of advantages over Tensorflow and similar frameworks -- especially with regards to performance. According to Microsoft's benchmarks, Cognitive Toolkit continues to outperform its competitors in most tests and unsurprisingly, this new version is faster than the previous releases, especially when working on big data sets.

Parag Mital is director of machine intelligence at Kadenze, as well as an artist and interdisciplinary researcher obsessed with the nature of information, representation and attention. Artificial intelligence allows machines to reason and interact with the world, and it's evolving at a breakneck pace. Many advances in AI can be attributed to machine learning, which works by tapping massive computing power to crunch through enormous amounts of digitized data. Now consider that most of our data, the best minds in the business and more computing power than you could ever imagine sit with just a handful of companies. For these reasons, only a few companies in the world are best situated to understand the true potential -- and the current limits -- of AI.

We hope you like that better than us dying. For you impatient types, we'll return to normal Video Friday in two weeks, so if you have video suggestions, keep them coming as usual.

This is a guest post. The views expressed here are solely those of the author and do not represent positions of IEEE Spectrum or the IEEE. When it comes to new technologies and products, we tend to think of "digital" as synonymous with advanced, modern, and high-def, while "analog" is considered retrograde, outmoded, and low-resolution. But if you think analog is dead, you'd be wrong. Analog processing not only remains at the heart of many vital systems we depend on today, it is now going to make its way into a new breed of compute and intelligent systems that will power some of the most exciting technologies of the future: artificial intelligence and robotics.

Artificial intelligence programs have beaten chess masters and TV quiz show champions. The'machine learning' software it is developing will be able to look beyond those set patterns and understand which situations truly warrant red flags The technology would not necessarily prevent events such as the 2010 'flash crash.' However, it could be quicker to catch manipulative behavior thought to contribute to them. Executives are hoping computers with humanoid wit can help mere mortals catch misbehavior more quickly. No comments have so far been submitted. Why not be the first to send us your thoughts, or debate this issue live on our message boards.

Recurrent neural networks are one of the staples of deep learning, allowing neural networks to work with sequences of data like text, audio and video. They can be used to boil a sequence down into a high-level understanding, to annotate sequences, and even to generate new sequences from scratch! The basic RNN design struggles with longer sequences, but a special variant – "long short-term memory" networks – can even work with these. Such models have been found to be very powerful, achieving remarkable results in many tasks including translation, voice recognition, and image captioning. As a result, recurrent neural networks have become very widespread in the last few years. As this has happened, we've seen a growing number of attempts to augment RNNs with new properties. Individually, these techniques are all potent extensions of RNNs, but the really striking thing is that they can be combined together, and seem to just be points in a broader space.