Humans cannot see, smell or taste airborne asbestos fibers. Identifying them through a microscope requires the eye of a trained analyst -- but perhaps not for long. Australian engineer Jordan Gruber is working on technology that can automatically detect asbestos from the air around a worksite. Exposure to airborne asbestos fibers is the primary cause of mesothelioma, an aggressive form of cancer. The past use of asbestos in building materials has led to great suffering among Americans and Australians alike.

Neural networks (NNs) and deep learning (DL) currently provide the best solutions to many problems in image recognition, speech recognition, natural language processing, control and precision health. NN and DL make the artificial intelligence (AI) much closer to human thinking modes. However, there are many open problems related to DL in NN, e.g.: convergence, learning efficiency, optimality, multi-dimensional learning, on-line adaptation. This requires to create new algorithms and analysis methods. Practical applications both require and stimulate this development.

What does a student coding in a white mask have in common with a New Zealand man struggling with a passport photo? Both individuals found themselves on the wrong side of computational decisions. The man of Asian descent erroneously had his eyes registered as closed by facial recognition software that analyzed his photo during an interaction with an automated passport renewal system. In my case, I wore a mask because my bare face was not consistently detected by facial recognition software. We both experienced exclusion from seemingly neutral machines programmed with algorithms -codified processes.

In January this year, a new app was released that gives users the ability to swap out faces in a video with a different face obtained from another photo or video – similar to Snapchat's "face swap" feature. It's an everyday version of the kind of high-tech computer-generated imagery (CGI) we see in the movies. You might recognise it from the cameo of a young Princess Leia in the 2016 Star Wars film Rogue One, which used the body of another actor and footage from the first Star Wars film created 39 years earlier. Now, anyone with a high-powered computer, a graphics processing unit (GPU) and time on their hands can create realistic fake videos – known as "deepfakes" – using artificial intelligence (AI). The problem is that these same tools are accessible to those who seek to create non-consensual pornography of friends, work colleagues, classmates, ex-partners and complete strangers – and post it online. Read more: The picture of who is affected by'revenge porn' is more complex than we first thought In December 2017, Motherboard broke the story of a Reddit user known as "deep fakes", who used AI to swap the faces of actors in pornographic videos with the faces of well-known celebrities.

An illustration shows what the earliest stars in the universe might have looked like. Stars are our constant companions in the night sky, but seas of twinkling lights weren't always a feature of the cosmos. Now, scientists peering back into deep time suggest that the earliest stars didn't turn on until about 180 million years after the big bang, when the universe as we know it exploded into existence. For decades, teams of scientists have been chasing--in fact, racing--to detect the signatures of these first stars. The new detection, from a project called EDGES, is in the form of a radio signal triggered when light from those stars began interacting with the hydrogen gas that filled primordial empty space.

Generative adversarial networks (GAN) have been effective for learning generative models for real-world data. However, existing GANs (GAN and its variants) tend to suffer from training problems such as instability and mode collapse. In this paper, we propose a novel GAN framework called evolutionary generative adversarial networks (E-GAN) for stable GAN training and improved generative performance. Unlike existing GANs, which employ a pre-defined adversarial objective function alternately training a generator and a discriminator, we utilize different adversarial training objectives as mutation operations and evolve a population of generators to adapt to the environment (i.e., the discriminator). We also utilize an evaluation mechanism to measure the quality and diversity of generated samples, such that only well-performing generator(s) are preserved and used for further training. In this way, E-GAN overcomes the limitations of an individual adversarial training objective and always preserves the best offspring, contributing to progress in and the success of GANs. Experiments on several datasets demonstrate that E-GAN achieves convincing generative performance and reduces the training problems inherent in existing GANs.

NatWest is testing an artificial intelligence-powered "digital human" called Cora that will converse with customers in branches – raising fears that bank tellers could be replaced by avatars. Cora, described by the bank as "highly lifelike", is the result of a link-up with a New Zealand tech company whose co-founder was involved with creating digital characters in the blockbuster films Avatar, King Kong and Spider-Man 2. Cora is currently able to answer basic verbal questions such as "How do I login to online banking?", "How do I apply for a mortgage?" NatWest said it could help cut down on waiting times because it would be able to deal with simple problems, adding that Cora's AI skills would eventually expand to answering hundreds of different questions, even detecting human emotions and reacting verbally and physically with facial expressions. The bank implied that a digital human could improve on the real thing by "providing consistent, accurate answers all the time in a way that humans can't always do". While some will be excited by the news that the "AI revolution" is coming to the UK's high streets, others will fear that the 71% taxpayer-owned bank is investing in this technology in order to replace branches or staff.

It was always going to be a momentous week for legal technology with the world's largest Global Legal Hackathon taking place in over 40 cities. But for the Aussies, it was a double whammy with ALTA's inaugural event and the launch of its Demo Day series, showcasing some of Australia's best legal technology start-ups all on one stage, in very lawyer-like six minute-long demonstrations. Hosted by Macquarie Bank at their Melbourne and Sydney headquarters and also supported by industry heavyweights Janders Dean and Elevate Services, the dual city tour was attended by a diverse variety of industry stakeholders including in-house legal departments, government, law societies, academics, multi-nationals and'techies', whilst Big Law right through to NewLaw and even'tiny' law were all well represented. What emerged from the two days was the incredible depth of our home-grown talent with many commenting on the impressive diversity; the companies showing off everything from data-driven tools for in-house teams to even an AI powered'law firm without lawyers'; but most importantly, a true sense of community was born. The ideas, conversations and relationships that transpired during the coffee breaks, in the conference rooms, and that continued out the doors well after the final demo will have a lasting effect.

New Zealand has a thriving industry working with AI technologies at all levels. Ahead of the 2018 launch of the AI Forum NZ's research report, the forum has released an ecosystem map. Discover who is currently investing in,working with and considering AI in New Zealand. Explore our thriving ecosystem and view the full infographic here. Sign up for the AI Forum NZ update, it's free and will take less than a minute!

This Chapter, "High-dimensional ABC", is to appear in the forthcoming Handbook of Approximate Bayesian Computation (2018). It details the main ideas and concepts behind extending ABC methods to higher dimensions, with supporting examples and illustrations.