Carl Vondrick is a doctoral candidate and researcher at MIT, where he studies computer vision and machine learning. His research focuses include leveraging large-scale data with minimal annotation and its applications to predictive vision and scene understanding. Recently his work has received a lot of media attention, including features in Forbes, Wired, CNN and PopSci, and other media outlets worldwide. As part of his work with MIT CSAIL, Carl built a deep learning vision system for AI to learn and understand human behaviour and interactions, using popular TV shows like The Office, Desperate Housewives, and YouTube videos. The resulting algorithm analyzes videos, then uses what it learns to predict how humans will behave.

Intelligence was the talk of Dreamforce - Salesforce's annual tech conference in San Francisco this month - with the SaaS giant's latest announcement'Einstein' promising to bring complex data science techniques and predictive algorithms seamlessly into all of their cloud-based CRM products. Here's how Salesforce used a spending spree on artificial intelligence (AI) startups and talent to bring these smart features to customers, all without opening up their precious data. Before he went on an AI acquisition binge, Salesforce CEO Marc Benioff said there was anxiety within the organisation around applying predictive algorithms to customer data they can't see, because customers want to keep their data private and secure. Read next: What is Salesforce's AI powered Einstein product? When can customers try Einstein and how much will it cost?

Shivon Zilis will participate in a panel discussion at Strata Hadoop World New York 2016, "Where's the puck headed?," considering the big trends in big data and explaining what the field will look like down the road. A year ago today, I published my original attempt at mapping the machine intelligence ecosystem. So much has happened since. I spent the last 12 months geeking out on every company and nibble of information I can find, chatting with hundreds of academics, entrepreneurs, and investors about machine intelligence. This year, given the explosion of activity, my focus is on highlighting areas of innovation, rather than on trying to be comprehensive.

You almost certainly should do (a) subject-wise cross-validation rather than (b) record-wise cross-validation. In some sense, an independent observation is a different subject. If you want to predict performance on new subjects, you must test on subjects you didn't train on! In typical settings, repeated observations of the same individual are correlated with each other even after conditioning on features. Hence with record-wise cross-validation, your test set isn't independent of your training set!

Elon Musk famously called it "our greatest existential threat". Physicist Stephen Hawking said that, limited by slow biological evolution, humans wouldn't be able to compete and would be superseded. But the technology that sparked those fears - artificial intelligence - is also being touted as the biggest potential advance in our history. A recent international study found that 50 per cent of experts questioned believe that artificial intelligence - or AI - will be smarter than humans within the next 24 years. And 90 per cent of those surveyed believed that milestone would be reached within 60 years.

Many data scientists are getting excited about advances in large-scale machine learning, particularly recent success stories in computer vision and speech. While correctly identifying meaningful patterns in data sets is the promise of machine learning, AI in the data center currently is the buzz for big tech companies like Google, Amazon, Facebook AI Research, Twitter and many more startups with shared goals of making useful machine learning software. For researchers looking at the scientific and engineering challenges of understanding the brain and building computers, Neural Computation highlights common problems and techniques in modeling the brain, and in the design and construction of neurally-inspired information processing systems. H2O is open source (AI for Business, deep learning with H20), Spark is open source (SparkNet batch processing framework). It's very easy to see the algorithm development and the value-add.

The use of robotics and artificial intelligence in businesses is on the rise, but there are still significant challenges for organisations adopting the technologies. Two executives from global IT consulting and outsourcing group Capgemini spoke to IoT Hub about how best to meet these challenges and why the returns make the effort worthwhile. "The amount of data that's available now in places like social media and enterprises means it is becoming for efficient for machines to make decisions rather than humans, taking the human bias out of it and making decisions objectively," said Saugata Ghosh, senior manager of digital services at Capgemini. This trend, together with the maturity of robotic process automation (RPA) technologies over the last three to five years, has contributed to the growth in adoption of robotics and AI, Ghosh said. "If you look at the spectrum of robotic automation, at one end you have simple rules-based automation where the economics of those are such that they are quite easy to implement and have strong returns on investment," he explained.

This story was delivered to BI Intelligence Apps and Platforms Briefing subscribers. To learn more and subscribe, please click here. The White House released a lengthy report Wednesday on artificial intelligence (AI) and its potential impact on multiple industries. Understanding and pinpointing possibly harmful factors that underline the rapid evolution of AI will help prepare future government bodies, businesses, and users for the broad deployment of AI. The report, titled "Preparing For The Future Of Artificial Intelligence," highlighted four key areas of AI and offered recommendations for each: To receive stories like this one directly to your inbox every morning, sign up for the Apps and Platforms Briefing newsletter.

Modern neural networks are good at making quick, reactive decisions and recognizing patterns, but they're not very skilled at the careful, deliberate thought that you need for complex choices. Google's DeepMind team may have licked that problem, however. Its researchers have developed a memory-boosted neural network (a "differentiable neural computer") that can create and work with sophisticated data structures. If it has a map of the London Underground, for example, it could figure out the quickest path from stop to stop or tell you where you'd end up after following a route sequence. The key is how the AI uses its memory.

Amazon, DeepMind/Google, Facebook, IBM and Microsoft last month announced the creation of the Partnership on AI, a nonprofit organization dedicated to formulating best practices in artificial intelligence and educating the public about the field. The group will invite academics, other nonprofits, and specialists in policy and ethics to join its board, which will represent corporate and noncorporate members equally. The partnership will conduct research, recommend best practices, and publish research under an open license. It will address topics such as ethics, fairness and inclusivity, transparency, privacy, and interoperability collaboration between people and AI systems, as well as AI technology's trustworthiness, reliability and robustness. There are no plans to lobby government or other policymaking bodies.