On 9 September 2019, the UN Human Rights Council begins its 42nd Session in Geneva (HRC42). Over 3 weeks, major human rights issues will be debated and acted on, with significant implications for the protection of freedom of expression and right to information globally. The UN Human Rights Council, with its 47 Member States, is an essential forum for the protection of freedom of expression, in particular for the rights of journalists, human rights defenders, and minorities and groups facing discrimination. As stakeholders prepare for HRC42, the UN's Human Rights Chief, Michelle Bachelet, has set out a series of thematic priorities for States to act upon, including to reverse shrinking civic space for protesters and dissenters, to push back against heavy censorship of the Internet and attacks on digital rights, and to end killings of human rights defenders, journalists, and trade unionists. As attacks on the multilateral system intensify, with autocrats even resorting to thuggish and personal jibes at Bachelet herself, it is crucial that rights-respecting States demonstrate that the Council can still deliver strong outcomes for freedom of expression.

If you have a story for Horizon, please email us at: editorial@horizon-magazine.eu Are you sure you want to view these Tweets? Is #AI a double-edged sword when it comes to fighting #climatechange? We need to assess the environmental impact of #AI – both good and bad – say experts. Water, fertilisers, soil: just a few of the resources we need to manage better to take care of Europe's land say experts http://bit.ly/Land-use

During its 1353rd meeting on 11 September 2019, the Committee of Ministers of the Council of Europe set up an Ad Hoc Committee on Artificial Intelligence – CAHAI. The Committee will examine the feasibility and potential elements on the basis of broad multi-stakeholder consultations, of a legal framework for the development, design and application of artificial intelligence, based on Council of Europe's standards on human rights, democracy and the rule of law.

Fox News Flash top headlines for Sept. 12 are here. Check out what's clicking on Foxnews.com California could soon become the largest state to ban the use of facial recognition technology in law enforcement body cameras, a significant milestone in the regulation of the burgeoning technology. The State Assembly on Thursday passed AB 1215, a bill that would impose a three-year moratorium on the technology, garnering praise from privacy and civil liberties advocates. The legislation now heads to Gov. Gavin Newsom's desk.

In fact, Gartner predicts that the market size of robotics process automation (RPA) alone is projected to reach $1.3B by the end of 2019. The concept of robotics is no longer limited to assembly lines in manufacturing, it is an overall technological approach that can have game changing impact for all workers in all industries. However, while there is palpable momentum for robotics, its adoption is still relatively nascent in legal operations, and for good reason. Today, most software robotics technology is focused on recording user actions and building task-based bots that perform repetitive tasks like data entry. While these bots can automate individual tasks, they don't have significant impact on processes as a whole, which require more advanced logic.

Information retrieval (IR) systems often leverage query data to suggest relevant items to users. This introduces the possibility of unfairness if the query (i.e., input) and the resulting recommendations unintentionally correlate with latent factors that are protected variables (e.g., race, gender, and age). For instance, a visual search system for fashion recommendations may pick up on features of the human models rather than fashion garments when generating recommendations. In this work, we introduce a statistical test for "distribution parity" in the top-K IR results, which assesses whether a given set of recommendations is fair with respect to a specific protected variable. We evaluate our test using both simulated and empirical results. First, using artificially biased recommendations, we demonstrate the trade-off between statistically detectable bias and the size of the search catalog. Second, we apply our test to a visual search system for fashion garments, specifically testing for recommendation bias based on the skin tone of fashion models. Our distribution parity test can help ensure that IR systems' results are fair and produce a good experience for all users.

Text-based Question Generation (QG) aims at generating natural and relevant questions that can be answered by a given answer in some context. Existing QG models suffer from a "semantic drift" problem, i.e., the semantics of the model-generated question drifts away from the given context and answer. In this paper, we first propose two semantics-enhanced rewards obtained from downstream question paraphrasing and question answering tasks to regularize the QG model to generate semantically valid questions. Second, since the traditional evaluation metrics (e.g., BLEU) often fall short in evaluating the quality of generated questions, we propose a QA-based evaluation method which measures the QG model's ability to mimic human annotators in generating QA training data. Experiments show that our method achieves the new state-of-the-art performance w.r.t. traditional metrics, and also performs best on our QA-based evaluation metrics. Further, we investigate how to use our QG model to augment QA datasets and enable semi-supervised QA. We propose two ways to generate synthetic QA pairs: generate new questions from existing articles or collect QA pairs from new articles. We also propose two empirically effective strategies, a data filter and mixing mini-batch training, to properly use the QG-generated data for QA. Experiments show that our method improves over both BiDAF and BERT QA baselines, even without introducing new articles.

Explainable machine learning seeks to provide various stak ehold-ers withinsights into modelbehavior via feature importancescores, counterfactual explanations, and influential samples, among other techniques. Recent advances in this line of work, however, h ave gone without surveys of how organizations are using these te ch-niques in practice. This study explores how organizations v iew and use explainability for stakeholder consumption. We find that the majority of deployments are not for end users affected by t he model but for machine learning engineers, who use explainability to debug the model itself. There is a gap between explainability in practice and the goal of publictransparency, since explanations primarily serve internal stakeholders rather than external on es. Our study synthesizes the limitations with current explainabi lity techniques that hamper their use for end users. To facilitate end user interaction, we develop a framework for establishing clear goals for explainability, including a focus on normative desiderata.

Investment in artificial intelligence (AI) is growing, with 60% of adopters raising their budgets 50% year over year, according to Constellation Research. But working with AI under emerging privacy standards is complex, requiring a dynamic balance that allows for continued innovation without misstepping on regulatory requirements. Under privacy regulations, businesses are responsible for gaining consent to use personal data and being able to explain what they are doing with that data. There is a real concern that black box automation systems that offer no explanations and require the long-term storage of large customer data sets will simply not be permitted under these regulations. Data regulations often have a negative connotation for companies, but with AI, the regulations could have the opposite effect.

Prominent members of Europe's so-called "Skype Mafia," all co-founders or early employees of the voice-over-Internet conferencing service, are backing Pactum, a startup that uses artificial intelligence to automate business contract negotiations. Founded late last year but only emerged from stealth mode on Wednesday, Pactum uses a chatbot-like interface to conduct contract talks. The bot can offer changes to standard terms, including price, delivery conditions and days to pay, in order to reach a better deal. The company is based in Mountain View, Calif., with engineering offices in Tallinn, Estonia, where Skype's first engineering offices were also located. Among those investing in the small startup are Jaan Tallinn, a Skype co-founder who has become a prominent backer of A.I.-related startups and research groups, Taavet Hinrikus, an early Skype employee who went on to found international payments firm TransferWise, Ott Kaukver, another early Skype employee who is now the chief technology officer at Twilio, and Sten Tamkivi, a general manager at Skype who is now chief product officer at Topia.