Modern life runs on intelligent algorithms. The data-devouring, self-improving computer programmes that underlie the artificial intelligence revolution already determine Google search results, Facebook news feeds and online shopping recommendations. Increasingly, they also decide how easily we get a mortgage or a job interview, the chances we will get stopped and searched by the police on our way home, and what penalties we face if we commit a crime, too. So they must be unimpeachable in their decision-making, right? Skewed input data, false logic or just the prejudices of their programmers mean AIs all too easily reproduce and even amplify human biases – as the following five examples show.

It's about to roll out in #India: https://bit.ly/2Jz4SIl

A spokeswoman for SenseTime, a facial-recognition software company dual-based in Beijing and Hong Kong, said it was "exploring the situation" and declined further comment. The company includes Singapore's state investor Temasek as one of its backers following a $600 million funding round which closed on Monday.

The future of finance will be derived from Artificial Intelligence (AI), so much so, that AI in finance is already taking the industry by storm. Be it faster speeds, personalized response, reduced errors and identifying opportunities, AI technology is already at the epicentre of the biggest financial revolution to happen since the last fifty or more decades. Be it consumer electronics, transportation, retail, healthcare or marketing, Artificial Intelligence companies are ready to ride the machine learning wave of the future. Thanks to the strong technological push towards an Internet of Things (IoT) based world, even the banks are putting their money (and resources) where their mouth is. One of the biggest advantages of machine learning is that it can be designed and trained to deliver a smarter way of assisting customers.

Prospects of universal AI ethics seem slim. However the five design pillars will serve organizations well beyond the social fads and fears. The goal – build controls that will identify biases, show attribution, and enable course correction as needed. Ready to roll out your plans for AI? Do you understand the business model implications? Who will you partner with for AI? Add your comments to the blog or reach me via email: R (at) ConstellationR (dot) com or R (at) SoftwareInsider (dot) org. Please let us know if you need help with your Digital Business transformation efforts. Here's how we can assist: Reprints can be purchased through Constellation Research, Inc. To request official reprints in PDF format, please contact Sales .

"To ensure the safety of road tests, we will not only not only require that road tests take place on prescribed streets, but also that the test driver sits in the driver position throughout, monitoring the car and the surrounding environment and ready to take control of the car at any time," he added.

The US and China have imposed tariffs on each other's goods. But will a skirmish between the world's two biggest economies turn into a full-on trade war? Perhaps it has already started. Both sides have struck initial blows. The US has imposed tariffs on imports of steel and aluminium.

This week we were treated to a veritable carnival attraction as Mark Zuckerberg, CEO of one of the largest tech companies in the world, testified before Senate committees about privacy issues related to Facebook's handling of user data. Besides highlighting the fact that most United States senators -- and most people, for that matter -- do not understand Facebook's business model or the user agreement they've already consented to while using Facebook, the spectacle made one fact abundantly clear: Zuckerberg intends to use artificial intelligence to manage the censorship of hate speech on his platform. Over the two days of testimony, the plan for using algorithmic AI for potential censorship practices was discussed multiple times under the auspices of containing hate speech, fake news, election interference, discriminatory ads, and terrorist messaging. In fact, AI was mentioned at least 30 times. Zuckerberg claimed Facebook is five to ten years away from a robust AI platform. All four of the other Big 5 tech conglomerates -- Google, Amazon, Apple, and Microsoft -- are also developing AI, many for the shared purposes of content control.

A Chinese fugitive was arrested after an AI-powered facial recognition system alerted authorities to his presence in a crowd of 60,000 people attending a pop concert. Welcome to the age of robot snitches. Wanted for "economic crimes," the 31 year-old man was reportedly surprised when police apprehended him. He'd traveled nearly 100 km (about 60 miles) with his wife and friends to attend the event, a concert headlined by Cantopop star Jacky Cheung, before authorities nabbed him on a tip from a venue camera. Chinese authorities have entirely embraced facial recognition systems and AI-powered surveillance monitoring.

NVIDIA cuDNN is a low-level library that provides GPU kernels frequently used in deep learning. Specifically, cuDNN implements several equivalent convolution algorithms, whose performance and memory footprint may vary considerably, depending on the layer dimensions. When an algorithm is automatically selected by cuDNN, the decision is performed on a per-layer basis, and thus it often resorts to slower algorithms that fit the workspace size constraints. We present {\mu}-cuDNN, a transparent wrapper library for cuDNN, which divides layers' mini-batch computation into several micro-batches. Based on Dynamic Programming and Integer Linear Programming, {\mu}-cuDNN enables faster algorithms by decreasing the workspace requirements. At the same time, {\mu}-cuDNN keeps the computational semantics unchanged, so that it decouples statistical efficiency from the hardware efficiency safely. We demonstrate the effectiveness of {\mu}-cuDNN over two frameworks, Caffe and TensorFlow, achieving speedups of 1.63x for AlexNet and 1.21x for ResNet-18 on P100-SXM2 GPU. These results indicate that using micro-batches can seamlessly increase the performance of deep learning, while maintaining the same memory footprint.