We are at the cusp of the next industrial revolution--or maybe it's in full swing already. Artificial intelligence, Internet of Things, cloud computing, smartphones, and a slew of other technologies that were unknown or sci-fi before the turn of the century are redefining and disrupting different aspects of life as we know it today. As with every industrial revolution, most of the changes overcoming our lives are pleasant. These are just some of the advantages brought by these technologies. But the same trends drag in tow some less appreciated disruptions, namely the upheaval of the socio-economic landscape.

The chief scientist helping drive Baidu Inc.'s push into artificial intelligence is quitting the Chinese search giant, putting at risk its efforts to put AI at the center of a business revival. Andrew Ng, a Stanford University academic who worked on deep learning at Alphabet Inc. before joining Baidu in 2014, said he's leaving the business next month. Ng doesn't plan to join another technology company and will seek to bring AI into sectors such as health care and education around the world. The departure comes at a crucial point for the Beijing-based company as it attempts to revive its fortunes by embracing machine intelligence across all of its business units. His decision to leave comes after Qi Lu was hired in January as Baidu's group president and chief operating officer with a mandate to reshape the business, whose online-ad business is under threat from rivals including Alibaba Group Holding Ltd. and Tencent Holdings Ltd. Prior to joining, Lu was an executive at Microsoft Corp. leading efforts to develop artificial intelligence.

The use of artificial intelligence applications in business is growing, but AI and machine-learning aren't yet an efficient use for every business task, according to an infographic. Published by LatentView, a marketing automation and digital analytics platform, the infographic details what criteria make a task a good candidate for AI, and goes on to explain how to use AI in your business. For example, in marketing, decisions about personalized offers to customers may work well with AI, but decisions about marketing strategy are best left to the humans (at this point,anyway), the infographic says. In customer service, AI can learn to improve responses based on historical chat data, but decisions that are based on empathy still require human intervention.

Sometimes the relationships between the data you've gathered are more important than the data itself. That's when a graph processing system comes in handy. It's an important but often poorly understood method for exploring how items in a data set are interrelated. Microsoft's been exploring this area since at least 2013, when it published a paper describing the Trinity project, a cloud-based, in-memory graph engine. The fruits of the effort, known as the Microsoft Graph Engine, are now available as an MIT-licensed open source project as an alternative to the likes of Neo4j or the Linux Foundation's recently announced JanusGraph.

One of the harder problems that chatbot developers face is, how to maintain the context of conversation. While all the popular frameworks provide an opinionated take on how to maintain this context, none of them seem to be either simple or complete. Here let me introduce a reactive approach to maintain the context. The elegance of this approach is that the pushed callback captures the state of variables as a part of its closure context. When the callback is called, this state is retained and we can time travel back to the state when the context was actually set.

Using virtual technology to take on the work of humans when interacting with applications will transform the procure-to-pay function. Automating process through robotics will allow you to perform repetitive tasks continuously and flawlessly all the time with minimal set up. Technology ensures error-prone payable processes and eliminates much of the mundane and time-consuming tasks such as data-entry. Also, robotics in AP will accelerate cycle times, easily manage tasks in volume, ensure consistency in operating procedures and streamlines audit & compliance activities. Human work can be combined with robotic process automation to streamline the supplier management and to tightly align accounts payable with other departments.

I will be resigning from Baidu, where I have been leading the company's AI Group. Baidu's AI is incredibly strong, and the team is stacked up and down with talent; I am confident AI at Baidu will continue to flourish. After Baidu, I am excited to continue working toward the AI transformation of our society and the use of AI to make life better for everyone. I joined Baidu in 2014 to work on AI. Since then, Baidu's AI group has grown to roughly 1,300 people, which includes the 300-person Baidu Research.

In recent years, we have become increasingly good at training deep neural networks to learn a very accurate mapping from inputs to outputs, whether they are images, sentences, label predictions, etc. from large amounts of labeled data. What our models still frightfully lack is the ability to generalize to conditions that are different from the ones encountered during training. Every time you apply your model not to a carefully constructed dataset but to the real world. The real world is messy and contains an infinite number of novel scenarios, many of which your model has not encountered during training and for which it is in turn ill-prepared to make predictions. The ability to transfer knowledge to new conditions is generally known as transfer learning and is what we will discuss in the rest of this post. Over the course of this blog post, I will first contrast transfer learning with machine learning's most pervasive and successful paradigm, supervised learning. I will then outline reasons why transfer learning warrants our attention. Subsequently, I will give a more technical definition and detail different transfer learning scenarios.

As a general rule of thumb, the law and government are slow-moving and deliberate. That's really handy for important things that you have to get right, but the trouble is that disruptive technology tends to move much faster. That's bad enough if the disruptive technology you're talking about is (say) the sharing economy, but it's more serious when it's something that Elon Musk once described as "potentially more dangerous than nukes". That thing is artificial intelligence. And while it's hard to feel too threatened when it's your Amazon Echo failing to understand you saying "Play REM" for the tenth time in a row, the threat – potentially – is a real one.