... includes all of the major AI methods for (a) representing knowledge about a task or a problem area, and (b) reasoning about a problem.
We can already see glimmers of this zeitgeist in some of the consumer applications now on the market – in any number of offerings from Google, for instance: Google Lens, which gives consumers the ability to learn more about a business simply by snapping a picture; Google Assistant and GoogleHome, which account for and respond to trends in buyers' daily habits. It only makes sense for B2B marketing technologies to follow suit, enabling personalized, one-to-one interactions in real-time across the customer journey. AI will empower marketers to build Adaptive Journeys that adapt to each individual and their preferred channel and send time and delivers the most relevant message in real-time as the customer goes about their research and discovery process. More importantly: this is the kind of continuous learning marketers require in order to deliver the customized experiences their buyers expect, and also the functionality they should prioritize in solutions they implement.
Whether it's Apple's Siri, Microsoft's Cortana, Amazon's Alexa, or virtually any GPS system, chances are the computerized personalities in your life are women. Hanson Robotics recently demoed Sophia, a learning and expressive robot designed to help humans in areas like healthcare and customer service. Market research is likely the main factor that influences tech companies when constructing AI personalities. Whether its stereotypes about women in service roles, the desire for a female companion, or simply that feeling of trust that a woman's voice instills, female AI personalities are easier for most consumers to adopt.
Since leaving Barclays in 2015, Mr Jenkins has spent time speaking to fintech startups and banking chief executives to get a sense of the gulf between the two parties and how to bridge it. Mr Jenkins founded 10x Technologies, a startup that offers a cloud-based core banking platform – a modern operating system for finance. However, it's the last challenge, cultural resistance, that Mr Jenkins says is "the most difficult and the most powerful" obstacle. "Bank management is largely technologically illiterate," says Mr Derhalli.
Technology companies of all sizes and in locations all around the world are developing AI-driven products aimed at reducing operating costs, improving decision-making and enhancing consumer services across a range of client industries. The sum of these drivers -- new programming techniques, more data and faster chips -- has seen AI converge with human-level performance in the key areas of image classification and speech recognition over recent years (see EXHIBIT 2). Chipmakers stand to benefit from increased demand for processing power, particularly makers of graphical processing units for AI program training. And internet companies with AI at the core of their consumer services (such as digital assistants and new software features) stand to benefit directly from improvements in speech recognition and image classification.
Amazon added Fire TV voice controls to the Echo speaker and other Alexa devices a couple weeks ago, while Google has offered similar voice controls for Chromecast through its own Google Home speaker since last December. Amazon Echo voice commands work with any Fire TV device (including first-generation models). In May, Google announced that it was expanding voice controls to HBO Now, Hulu, YouTube TV, Google Play Movies & TV, CBS All Access, Food Network, CW, HGTV, Red Bull TV, Travel Channel, Crackle, DIY Network, Viki, and Cooking Channel, but it's unclear when this support will become available. Although Google Home and Amazon Alexa don't offer direct control over most televisions and sound systems, they both work with Logitech Harmony hubs, which in turn can control various TVs, sound systems, and streaming boxes.
The hardware now supports Alexa, Amazon's voice-enabled assistant, through an Echo or Echo Dot-equipped speaker. So you can ask, "Alexa, ask EE TV what's on tonight" and hear some suggestions from the Freeview-powered guide. Amazon has long touted Alexa as a key part of its Fire TV and Fire TV stick, however those devices are for on-demand apps only. The problem, at least for now, is that the service requires an Echo speaker -- with Amazon's Fire TV, you get a purpose-built remote in the box.
The EPSRC Principles of Robotics, British Standards' BS8611, IEEE global ethics initiative and IEEE P7001 – Transparency in Autonomous Systems are some of the current guidelines being used by the industry. Interestingly, DeepMind and OpenAI have collaborated on research into an algorithm that can reinforce machine learning, based on human preferences. Chohan adds that Natera, which is involved in genetic testing, filed the most patents for AI and machine learning related ethics or morality, mainly because it is concerned with the use of extremely personal patient data. According to Chohan, the number of artificial intelligence patents concerning ethics or morality spiked in 2014 – when the European Union's General Data Protection Regulation was being widely discussed, particularly the "right to be forgotten".
Building successful virtual assistants requires a combination of magic and logic. Automation of manual processes, particularly in legacy businesses with legacy technology, has significant cost base implications. Without a sustainable capability model, businesses are struggling to attract people with the relevant skills, particularly when trying to compete with Google, Amazon and Facebook. While developing internal data analytics capabilities, migrating data from silos into an extensible cloud solution and building key strategic partnerships may not provide visceral evidence of progress in the short term, it is vital to long term sustainable success.
About this course: Probabilistic graphical models (PGMs) are a rich framework for encoding probability distributions over complex domains: joint (multivariate) distributions over large numbers of random variables that interact with each other. They are also a foundational tool in formulating many machine learning problems. It describes the two basic PGM representations: Bayesian Networks, which rely on a directed graph; and Markov networks, which use an undirected graph. The course also presents some important extensions beyond the basic PGM representation, which allow more complex models to be encoded compactly.