Stuart Russell is a distinguished artificial intelligence researcher, a Professor of Computer Science at the University of California, Berkeley, an Adjunct Professor of Neurological Surgery at the University of California, San Francisco, and leads the Center for Human-Compatible Artificial Intelligence at UC Berkeley. Along with Peter Norvig, Stuart is the author of Artificial Intelligence: A Modern Approach, the most widely used textbook on artificial intelligence. In his most recent book, Human Compatible: AI and the Problem of Control, Stuart proposes a fundamentally new approach to developing AI. In this interview, Stuart warns that AI is reshaping society in unintended ways. For example, social media content selection algorithms that choose what individuals watch and read do not even know that human beings exist. As AI becomes more capable, he suggests that we are going to see bigger failures unless we change the way we think about AI altogether. Stuart argues that to ensure AI is provably beneficial for human beings, we must design machines to be inherently uncertain about human preferences. This way, we can ensure they are humble, altruistic, and committed to pursuing our objectives even as they set their own goals. We also discuss why AI needs regulation similar to civil engineering and medicine, the impact AI is going to make over the next decade, autonomous vehicles, and a variety of other topics.

More than 50 years ago Marvin Minsky made the first attempt to mimic the human brain, triggering further research into computers' ability to process information to make intelligent decisions. Over the years, the process of automating image analysis led to the programming of algorithms. However, it was only from 2010 onward, when there was acceleration in deep learning techniques. In 2012 Google Brain built a neural network of 16,000 computer processors which could recognise pictures of cats using a deep learning algorithm. This is an edited extract from the Computer Vision – Thematic Research report produced by GlobalData Thematic Research.

The world of artificial intelligence is constantly evolving, and certainly so is the legal and regulatory environment in which it exists. Michael Hayes, Senior Manager of Government Affairs at the Consumer Technology Association (CTA) is focused on these emerging technology challenges that hit up against existing laws and regulations. Michael previously worked on Capitol Hill on patent reform, stopping patent trolls. As part of his current role, Michael makes sure that the emerging policy discussion is framed in a way that makes sure that the technology can thrive and provide competitive advantages for companies implementing them without introducing new risks. There has been a lot of concern about corporation and government's use of data, and the role of privacy.

Innovation can come from the most unexpected of places and perhaps the last place you would expect digital innovation within the insurance sector to develop is a law firm. And yet, with the launch of its separate technology-driven arm, Kennedys IQ, global insurance law firm Kennedys has placed itself among the most transformative digital companies impacting the development of the insurance sector today. Speaking with Insurance Business, partner and head of the innovations group at Kennedys and board director of Kennedys IQ, Richard West (pictured above), outlined how the unveiling of Kennedys IQ, which combines human and machine intelligence to offer clients'Kennedys, without the lawyers', has been a natural next step in producing innovative software for clients. The journey started 10 years ago with Kennedys' virtual lawyer, KLAiM, which, West said, is essentially the great grandfather of this latest innovation offered by the company. It has come to fruition from Kennedys' ambition of ensuring that lawyers are only used in the insurance sector when strictly necessary.

Artificial intelligence AI takes the lead over intelligent automation IA. Intelligent automation is the combination of "'robotic process automation and artificial intelligence to automate processes,'" according to a recent article on the topic in HR Dive, a publication for human resources professionals. Organizations that embrace intelligent automation may experience a return on investment of 200% or more, according to an Everest Group report cited by HR Dive. However, that doesn't mean organizations can expect a reduction in headcount, according to the report. In fact, projections of a reduction in workforce thanks to intelligent automation may be "highly exaggerated," the Everest Group noted.

A key obstacle to the successful deployment of machine learning (ML) methods to important application domains is the (lack of) explainability of predictions. Explainable ML is challenging since explanations must be tailored (personalized) to individual users with varying backgrounds. On one extreme, users can have received graduate level education in machine learning while on the other extreme, users might have no formal education in linear algebra. Linear regression with few features might be perfectly interpretable for the first group but must be considered a black-box for the latter. Using a simple probabilistic model for the predictions and user knowledge, we formalize explainable ML using information theory. Providing an explanation is then considered as the task of reducing the "surprise" incurred by a prediction. Moreover, the effect of an explanation is measured by the conditional mutual information between the explanation and prediction, given the user background.

There is little argument about whether or not machine learning models are useful for applying to social systems. But if we take seriously George Box's dictum, or indeed the even older one that "the map is not the territory' (Korzybski, 1933), then there has been comparatively less systematic attention paid within the field to how machine learning models are wrong (Selbst et al., 2019) and seeing possible harms in that light. By "wrong" I do not mean in terms of making misclassifications, or even fitting over the'wrong' class of functions, but more fundamental mathematical/statistical assumptions, philosophical (in the sense used by Abbott, 1988) commitments about how we represent the world, and sociological processes of how models interact with target phenomena. This paper takes a particular model of machine learning research or application: one that its creators and deployers think provides a reliable way of interacting with the social world (whether that is through understanding, or in making predictions) without any intent to cause harm (McQuillan, 2018) and, in fact, a desire to not cause harm and instead improve the world, 1 for example as most explicitly in the various "{Data [Science], Machine Learning, Artificial Intelligence} for [Social] Good" initiatives, and more widely in framings around "fairness" or "ethics." I focus on the almost entirely statistical modern version of machine learning, rather than eclipsed older visions (see section 3). While many of the limitations I discuss apply to the use of machine learning in any domain, I focus on applications to the social world in order to explore the domain where limitations are strongest and stickiest.

Clearview AI, the controversial facial recognition startup that's gobbled up more than three billion of our photos by scraping social media sites and any other publicly accessible nook and cranny it can find, has lost its entire list of clients to hackers – including details about its many law enforcement clients. In a notification that The Daily Beast reviewed, the company told its customers that an intruder "gained unauthorized access" to its list of customers, to the number of user accounts they've set up, and to the number of searches they've run. The disclosure also claimed that Clearview's servers hadn't been breached and that there was "no compromise of Clearview's systems or network." The company said that it's patched the unspecified hole that let the intruder in, and that whoever it was didn't manage to get their hands on customers' search histories. Security is Clearview's top priority.

Scientists have proposed a new international framework to keep ethics and human wellbeing at the forefront of our relationship with technology. From gene therapy and AI-predicted disease to self-driving cars and 3D printing, advances in technology can improve health, free up time, and boost efficiency. However despite the best intentions of its creators, technology might lead to unintended consequences for individual privacy and autonomy. There's currently no internationally agreed-upon regulation about who, for example, has access to the data recorded by black boxes in cars, smart TVs and voice enabled personal assistants - and recent findings have shown that technology can be used to influence voting behaviour. Now, Imperial College London researchers have suggested a new regulatory framework with which governments can minimise unintended consequences of our relationship with technology.

During a session that lasted two hours, European Film Market (EFM) Horizon attempted to determine whether artificial intelligence (AI) is film's new normal. Moderated by AC Coppens (The Creatives' Catalysts), the conference, which encompassed five presentations, explored the presence of AI, in its various forms, at different stages of a film's life cycle. During her introductory keynote speech, Maja Cappello, head of the European Audiovisual Observatory's legal department, focused on the real or imaginary legal issues that the use of AI creates in fiction. Given that legal systems vary dramatically between regions, in the UK, it would probably be possible for AI to hold on to the rights, whereas that wouldn't be the case in continental Europe. Also, another possible issue arises in cases where a machine creates original content: who will be the owner of this product?