Over the past few years, bots have started taking over parts of the tech world, with good bots like web crawlers indexing your site to boost your traffic and chat bots helping with more efficient communication in the office. Unfortunately, malicious bots are also on the rise, exposing vulnerabilities and stealing information. In fact, 2014 was the first year that bots were purported to have outnumbered actual people online. Distil Networks, a company that provides bot detection and mitigation services, recently raised 21 million in a Series C financing round to boost its efforts against bad bots. For those unfamiliar, a bot is simply a piece of software that runs automated scripts online.

Traditionally, this meant finding tens of thousands of labeled examples that your model can learn from. We created our Custom Collections API to allow you to build effective models with 10x-1000x less training data. And now, we introduce CrowdLabel – a new utility that is designed to help you get labeled data to accelerate model development even further.

This is a short follow-up to our last blog post about Hashpipe, our experiment in machine learning. In our early research for that project, we experimented with a convolutional neural network for image classification, using ConvNetJS. We understand that's pretty technical language. In a nutshell, the idea here was to see if we could teach a machine to recognize specific objects. Through those experiments we used deep learning to classify images into one of two buckets: pictures of Porsches and pictures of burgers.

Computers are only as good, or as bad, as the people who program them. And it turns out that many individuals who create machine learning algorithms are presumably and unintentionally building in race and gender bias. In part one of a two-part interview, Richard Sharp, CTO of predictive marketing company Yieldify, explains how it happens. The Enterprisers Project (TEP): Machines are genderless, have no race, and are in and of themselves free of bias. How does bias creep in?

Artificial Lawyer caught up with Cian O'Sullivan, founder of Beagle, the automated contract analysis system that is just celebrating a year and a half of operations and landing VW as a client. We discussed how Beagle came about, why maybe sometimes it's better not to talk to lawyers about AI and how come the company has one of the world's largest auto companies as a client, and then some. Cian O'Sullivan's web camera is not working when Artificial Lawyer calls for a video conference and so is treated to a picture of a soccer pitch in Colombia that the legal tech company founder took on his travels. The international reference makes sense once you start to talk to O'Sullivan. The Canadian travels a lot.

Our once-subservient machines are encroaching on tasks that have been firmly in the human domain. Even seasoned carnival barkers might struggle to exaggerate the current feats of cognitive technologies. We now marvel at artificial intelligence-fueled creations that allow a teenager to win 160,000 parking ticket cases or an app to quickly diagnose health symptoms at a fraction of the cost of a doctor alone. Unless, of course, you manage a law firm or medical practice now in the position of competing against these machines. Cognitive technologies are artificial intelligence-based systems that increasingly do tasks that once required humans.

When Google introduced RankBrain as its third most important ranking signal, it caught many people off guard. It also frightened many SEO providers who were afraid they might have to completely overhaul their approach to optimization. And in many cases, they were right. Anyone who approached SEO from a perspective of trying to play off algorithms or guess how to best game the system is in trouble, because the system is no longer one with which we can keep up. RankBrain processes too much data too quickly for tricks to work. As artificial intelligence gains and increases its foothold in Search Engine Rankings, SEO webmasters will need to replace gamesmanship with quality.

LONDON: Technological developments will dramatically change the role of agency strategists as they move from a free-associating, subjective approach to a more empirical, objective and advisory role, a leading industry figure has said. Writing in the current issue of Admap, Mark Holden, Worldwide Strategy and Planning Director at PHD, outlines the future direction of strategy that will start to emerge once attribution modelling and demand-side platforms come together. Currently, users log in to the former to pull out insights and then log in to the latter to execute their strategies. "When they are finally joined up, this will create the first closed system our industry has ever experienced," Holden says, "with this the basis into which we can drop a reinforcement learning algorithm." The point about reinforcement learning – an emerging area of artificial intelligence – is that it requires a closed system, where action and outcome are inextricably linked, in order to further develop.

Scientists may have found an off switch for alcoholism. Targeting only specific neural pathways that are specialized just for alcohol consumption, researchers at the Scripps Research Institute were able completely eliminate compulsive alcohol consumption in mice populations, according to research published Wednesday in the Journal of Neuroscience. "It's like they forgot they were dependent," Olivier George, an assistant professor at Scripps and lead researcher on the study, said of the findings. "We can completely reverse alcohol dependence by targeting a network of neurons." When a person or mouse drinks alcohol, they develop neural reward pathways specifically for alcohol.

We propose Turing Learning, a novel system identification method for inferring the behavior of natural or artificial systems. By observing the behavior of the system as well as the behaviors produced by the models, two sets of data samples are obtained. The classifiers are rewarded for discriminating between these two sets, that is, for correctly categorizing data samples as either genuine or counterfeit. Conversely, the models are rewarded for'tricking' the classifiers into categorizing their data samples as genuine. Unlike other methods for system identification, Turing Learning does not require predefined metrics to quantify the difference between the system and its models.