My last note looked at the problem of overfitting machine learning models and a best practice to avoid it called regularization. This note looks at neural networks, what they are, and how they work conceptually. In its most simple form, a neural network can be just an input, a node (or neuron) where a calculation uses that input, and the output of the calculation. This by itself isn't very powerful, but when you combine many of these nodes into a network, they can operate like how neurons in our own brain work. Using linear algebra and calculus to execute a concept called backpropagation of error, these networks of neurons can automatically adjust the weights of the connections between them by assessing how prediction errors map back to different connections in the network.

A specialist technology app store and artificial intelligence for contract drafting are among the technologies likely to be adopted by the legal arm of Big Four firm Deloitte in an attempt to bridge the gap between startup companies and clients. Deloitte Legal said today that it had'hand-selected' 14 startups after evaluating 400 businesses for its Deloitte Legal Ventures programme. In a significant departure from previous lawtech incubator schemes, Deloitte Legal has said it will become a user of products and services offered by the chosen startups. These include execution technology, artificial intelligence, data analytics and predictive analytics, the firm said. Among the 12 companies named today are Genie AI, a UK pioneer in artificial intelligence and Reynen Court, an online platform designed to simplify the process of buying legal technology.

If there's one thing that AI can't do, it's be human. But that's not necessarily a bad thing because, even the least sophisticated bots today can analyze massive amounts of data in seconds, spit out mathematical results and continue those tasks nigh on forever. So, while AI tools can't technically be human (they are still robots, after all), they can do a lot of things only humans could once do. And bots' sophistication is hardly limited to self-driving cars or board-game world championships: Human or not, this technologycan be used to build healthier, longer-lasting relationships with customers. One very human experience is giving and receiving feedback, quality feedback.

Facebook is launching a new dating app on the social media platform, its CEO, Mark Zuckerberg, announced at an annual developer conference on Tuesday, unveiling a feature designed to compete with popular services like Tinder. Speaking in front of a packed crowd in San Jose, Zuckerberg described the new dating feature as a tool to build "real long-term relationships – not just hookups". "We want Facebook to be somewhere where you can start meaningful relationships," he continued. "We've designed this with privacy and safety in mind from the beginning." The announcement sparked gasps from the crowd and seemed to attract the most interest from the audience during Zuckerberg's short speech, which focused on the company's widening privacy scandal, new safeguards meant to protect users' data and misinformation and fake news on the site.

It isn't just the tech entrepreneurs and Hollywood directors who dream about the role that artificial intelligence can play, or will play, in everyday human life--educators have begin to join them. However, those dreams aren't always pleasant and may, in fact, sometimes turn into nightmares. If computer systems are able to perform tasks that humans have performed for thousands of years, will it render teachers and administrators a thing of the past? Or is artificial intelligence the secret to freeing up educators' time for other, non-routine tasks, like mentoring and spending more one-on-one time with students? To find out, I went straight to the source--eight educators, including superintendents, coaches and teachers--to find out whether AI tickles their fancy or scares them straight.

I brace for rejection, but then the robot lets out a balooop and shimmies back and forth. This, I am to presume, means Kuri loves me too. Interacting with Kuri, a robot set to hit the market in December, is at once fascinating, delightful, and puzzling. Kuri's creators call it a "companion robot," but this is no Furby. Kuri belongs to a new class of machines that actually are intelligent, and actually make useful assistants at home.